United States v. Johnson

Full title: UNITED STATES OF AMERICA, Plaintiff, v. TERRY S. JOHNSON, in his official…

Court: UNITED STATES DISTRICT COURT FOR THE MIDDLE DISTRICT OF NORTH CAROLINA

Date published: Aug 7, 2015

122 F. Supp. 3d 272 (M.D.N.C. 2015)

Opinion

No. 1:12cv1349.

2015-08-07

UNITED STATES of America, Plaintiff, v. Terry S. JOHNSON, in his official capacity as Alamance County Sheriff, Defendant.

Michael J. Songer, Anika Gzifa, Jude Volek, Silvia J. Dominguez–Reese, Timothy Mygatt, United States Department of Justice, Washington, DC, for Plaintiff. S.C. Kitchen, Karlene S. Turrentine, Turrentine Law Firm, PLLC, Raleigh, NC, for Defendant.

THOMAS D. SCHROEDER, District Judge.

Michael J. Songer, Anika Gzifa, Jude Volek, Silvia J. Dominguez–Reese, Timothy Mygatt, United States Department of Justice, Washington, DC, for Plaintiff.

S.C. Kitchen, Karlene S. Turrentine, Turrentine Law Firm, PLLC, Raleigh, NC, for Defendant.

MEMORANDUM OPINION AND ORDER

THOMAS D. SCHROEDER, District Judge.

TABLE OF CONTENTS

I.	FINDINGS OF FACT	282

	A.	Alamance County	282

		1.	Population and Demographics of Alamance County	282
		2.	Law Enforcement Challenges in Alamance County	283

	B.	ACSO and Sheriff Johnson	285

		1.	The Introduction and Implementation of ICE's 287(g) Program in Alamance County	285

			a.	287(g) Program's TFO	286
			b.	Termination of TFO Position	289
			c.	Gun Permit Investigations	290
			d.	ACSO's Processing and Booking Procedures	291
			e.	Fairness Alamance	294

		2.	ACSO Arrest Policy and Practice	295
		3.	Orders to Target Hispanics	297
		4.	ACSO Checkpoint Policy and Implementation	298
		5.	ACSO's Stops and Searches	302

			a.	Particular Stops	302
			b.	Statistical Evidence	304

				i.	Dr. John Lamberth	304
				ii.	Dr. John MacDonald	307

					(a)	Post–Stop Outcome Study	309
					(b)	"Hit–Rate" Study	310

				iii.	Officer Mark Dockery and ACSO Data on Searches Not Incident to Arrest	311
				iv.	Dr. David Banks	312

					(a)	ACSO's Checkpoint Siting	313
					(b)	ACSO's Checkpoint Stops	314
					(c)	ACSO's Checkpoint Arrests	315
					(d)	ACSO's Traffic Stops	316
					(e)	ACSO's Citations	316

	C.	Evidence of Racial and Ethnic Bias Within ACSO	317

		1.	Racial and Ethnic Epithets and Jokes	318
		2.	Derogatory Emails	319

	D.	ACSO's Supervision and Discipline of Its Officers	322

		1.	Supervision	322

			a.	ACSO's Complaint Policy	322
			b.	ACSO's Review of Stops, Arrest, and Searches	323
			c.	ACSO's Traffic Stop Data Reporting	324
			d.	ACSO's Training of Its Officers	325
			e.	ACSO's Harassment Prevention Policy	325
			f.	Implementation of Email and Video Filtering Software	326

		2.	Discipline	326
		3.	Margo Frasier	327

II.	CONCLUSIONS OF LAW	329

	A.	Motion to Exclude Dr. Lamberth's Testimony	329

		1.	Admissibility	330

			a.	Testability	330
			b.	Peer Review	332
			c.	Error Rates	332
			d.	General Acceptance	334

		2.	Credibility	337

	B.	Motion to Exclude Dr. Banks' Testimony	338

		1.	Dr. Banks' Qualifications	338
		2.	The Admissibility of Dr. Banks' Testimony	339

			a.	Dr. Banks' Study of ACSO Traffic Stops	339
			b.	Dr. Banks' Analysis of ACSO Checkpoint Stops	340
			c.	Dr. Banks' Study of ACSO Checkpoint Arrests	341
			d.	Dr. Banks' Permutation Test	341

	C.	Motion for Adverse Inference	342
	D.	Motion to Exclude Kenneth Evans' Testimony	345
	E.	42 U.S.C. § 14141 Claims	347

		1.	Fourteenth Amendment Claim	349

			a.	Express Classification	350
			b.	Facially Neutral Classification	354

				i.	Traffic Stops	357
				ii.	Checkpoint Placement and Stops	358
				iii.	Post–Stop Outcomes	360
				iv.	Searches After Stops	365
				v.	287(g) Practices	369
				vi.	ACSO's Culture, Supervision, and Discipline	371

		2.	Fourth Amendment Claim	372

	F.	Statute of Limitations	379

III.	CONCLUSION	380

The United States of America (the "Government") alleges that, from at least January 2007 to the present, Defendant Terry S. Johnson, in his official capacity as Sheriff of Alamance County, North Carolina, engaged in a pattern or practice of discriminatory law enforcement activities directed against Hispanics,¹ in violation of the Fourth and Fourteenth Amendments to the United States Constitution. The Government brings this action via Section 210401 of the Violent Crime Control and Law Enforcement Act of 1994, 42 U.S.C. § 14141. (Doc. 1.) Principally, the Government charges that Alamance County Sheriff's Office ("ACSO") disproportionately subjects Hispanics to unreasonable searches, arrests them for minor infractions (in lieu of issuing warnings or citations), targets them at vehicle checkpoints located in predominantly Hispanic neighborhoods, uses ethnically-offensive epithets to refer to Hispanics and otherwise tolerates activities of deputies that evidence anti-Hispanic bias, automatically and selectively refers Hispanic arrestees to U.S. Immigration and Customs Enforcement ("ICE") investigators for deportation, and otherwise engages in deficient policies, training, and oversight that facilitates discriminatory enforcement. (Id. ) Sheriff Johnson denies any wrongdoing and maintains that ACSO's law enforcement is legitimate and lawful.

1 The Government refers to Latinos and Hispanics interchangeably.

The parties previously filed cross-motions for summary judgment, which the court considered and, following oral argument, granted in part and denied in part. (Doc. 118.)

A bench trial on the merits was conducted from August 12 through 22, 2014. The Government presented twenty-nine fact witnesses and three expert witnesses, and Sheriff Johnson presented sixteen fact witnesses and one expert witness. At the close of the Government's evidence, Sheriff Johnson moved for judgment pursuant to Federal Rule of Civil Procedure 52(c), which the court took under advisement. Following trial, the parties sought to file proposed findings of fact and conclusions of law and requested additional time to do so. They have now submitted them. (Docs. 157, 158.) The case is therefore ready for decision.

Pursuant to Federal Rule of Civil Procedure 52(a), the court enters the following findings of fact—based upon an evaluation of the evidence, including the credibility of witnesses, and the inferences that the court has found reasonable to draw therefrom—and conclusions of law. To the extent any factual statement is contained in the conclusions of law, it is deemed a finding of fact as well.

As explained by the following analysis, and after careful consideration, the court concludes that the Government has failed to demonstrate that ACSO has engaged in a pattern or practice of unconstitutional law enforcement against Hispanics in violation of § 14141.

I. FINDINGS OF FACT

A. Alamance County

1. Population and Demographics of Alamance County

Alamance County (the "County") lies within central North Carolina and is currently home to over 150,000 residents. See State & County QuickFacts: Alamance County, North Carolina, U.S. Census Bureau, http://quickfacts.census.gov/qfd/states/37/37001.html (last visited June 29, 2015).² The Government has consistently contended, and Sheriff Johnson does not dispute, that Alamance County's population, and particularly its Hispanic population, has risen sharply since 1990. (See Doc. 1 ¶ 12; Doc. 158 at 9.) As of the 1990 U.S. Census, Alamance County was home to about 108,000 residents, with the County's Hispanic population totaling fewer than 800—or less than 1% of the total population. See 1990 Census of Population and Housing Public Law 94–171 Data (Official) Age by Race and Hispanic Origin, U.S. Census Bureau, http://censtats.census.gov/pl94/pl94.shtml (displaying population totals for North Carolina by county) (last visited June 29, 2015). Today, just over twenty years later, the County has over 150,000 residents, with the Hispanic population expanding to approximately 17,700—comprising 11.8% of the total population. See State & County QuickFacts: Alamance County, North Carolina, U.S. Census Bureau, http://quickfacts.census.gov/qfd/states/37/37001.html (presenting figures on the demographic layout of Alamance County) (last visited June 29, 2015).

2 Under Federal Rule of Evidence 201, the court may take judicial notice of U.S. Census data. See Hollinger v. Home State Mut. Ins. Co., 654 F.3d 564, 571–72 (5th Cir.2011) ("United States census data is an appropriate and frequent subject of judicial notice."); United States v. Gregory, 871 F.2d 1239, 1245 (4th Cir.1989) (taking judicial notice of population figures); see also United States v. Cecil, 836 F.2d 1431, 1452 (4th Cir.1988) (agreeing that "courts may take judicial notice of official governmental reports and statistics" under Rule 902(5) of the Federal Rules of Evidence ).

2. Law Enforcement Challenges in Alamance County³

3
Any challenge to law enforcement conduct must be examined in the context of legitimate law enforcement concerns, keeping in mind that, while highly relevant, such concerns cannot excuse or justify unconstitutional policing.

The Sheriff contends that the County's recent growth has brought with it increasing crime, including a serious drug trafficking problem with associated drug-related violence. In particular, the Sheriff presented uncontested evidence that major Mexican drug trafficking organizations ("DTOs"), including the Sinaloa drug cartel, have relocated operations to Alamance County, making it a hub for drugs and crime. (Doc. 152 at 109, 112, 114.) According to a 2010 Department of Justice ("DOJ") and Drug Enforcement Administration ("DEA") report, Mexican DTOs "mov[ed] their operations beyond metropolitan Atlanta into rural areas of Georgia, North Carolina, and South Carolina ... to avoid law enforcement pressure in the Atlanta ... region." (Id. at 114; see also Def. Trial Ex. 30 at 7.) The DOJ's criminal enforcement division, a sister division of the one bringing the present action, has specifically identified Alamance County as a national concern. (Def. Trial Ex. 30 at 8.) Throughout the time period at issue in this case, as detailed below, the DEA has actively enlisted ACSO's assistance in enforcing the nation's drug laws against drug traffickers in Alamance County.

Mexican DTOs send illegal drugs, such as cocaine, "directly from Mexico" to distribution hubs like Alamance County. (Doc. 152 at 100.) According to one ICE agent, drugs in North Carolina come "predominately from Mexico" and are received "mainly" by Mexican nationals in North Carolina. (Doc. 153 at 24.) If not seized in Alamance County, the drugs then move farther along the DTO distribution chain to cities like Chicago, Illinois, and states like Virginia and South Carolina. (Doc. 152 at 101.) Proceeds from the DTOs' drug transactions are returned to drug trafficking operators in Mexico. (Id. )

Inside Alamance County, DTO "cell heads" run logistics for the distribution chains. (Id. at 116.) DTOs mostly employ family and friends in the United States as drug recipients, and family and friends in Mexico receive drug proceeds sent from the United States. (Id. at 112.) As a result, many American-based DTO cell heads are Mexican citizens with direct ties to Mexico. (Id. at 112, 116.) When DTO cell heads in places like Alamance County are arrested, they are replaced with DTO members from Mexico or the U.S. border. (Id. at 115–16.) Thus, according to DEA Agent Walter Serniak, Jr., while drug users are "not of any particular ethnicity," (id. at 129), greater than 90% of those arrested in Alamance County in connection with drug trafficking are Hispanic. (Id. at 115–16.)

The increased presence of Mexican DTO cell heads in Alamance County has brought an increase in drugs, drug money, guns, and violence. DTO members often keep stash houses to store drugs, money, and guns. (Id. at 101.) According to Agent Serniak, the DTOs keep a "great number" of stash houses in Alamance County and often operate within the County's mobile home parks. (Id. at 101, 104; see also Doc. 154 at 106–07.) For example, a recent DEA raid of a mobile home park in Alamance County yielded a seizure of hundreds of thousands of dollars in drug money. (Doc. 152 at 104; see also Def. Trial Ex. 4 (showing money seized from a stash house).) DTOs have become so sophisticated and entrenched that they launder money through local businesses within the County. (Doc. 152 at 105.)

The DEA has conducted a number of raids in Alamance County, which have resulted in seizures of drugs and weapons connected to Mexican DTOs.⁴ For example, the DEA discovered approximately 12 kilograms of cocaine destined for the D.C./Baltimore area in 2005, 39 kilograms of cocaine in 2009, and one kilogram of heroin from the Green Level community.⁵ (Id. at 101–03, 109–10.) Drug-related weapons seizures are common, and their prevalence has increased in recent years. (Id. at 104–05.) The DEA has seized high-caliber handguns, rifles, and assault rifles. (Id.; Def. Trial Ex. 4.) And, with the confluence of money laundering, drugs, and weapons, the DEA has investigated drug-related homicides in the County. (Doc. 152 at 111; Def. Trial Ex. 4.) Overall, the DEA's operations have netted seizures of over 100 firearms, millions of dollars, and hundreds of kilograms of cocaine. (Def. Trial Ex. 4.)

4 Despite these operations, Agent Serniak testified that he never observed racial or ethnic profiling by ACSO. (Doc. 152 at 127.)

5 The street value of a kilogram of cocaine is about $30,000 dollars, and a kilogram of heroin is about $60,000. (Def. Trial Ex. 4.)

United States Interstates 40 and 85 join as a common highway in Alamance County, dissecting the County horizontally at its midsection. The town of Green Level lies just north of the highways and offers an example of the collateral damage resulting from the growing drug trafficking in the County. The Green Level community faced, and continues to face, "a drug problem" from DTOs that has left many residents living in fear. (Doc. 152 at 62–63, 65.) Sandra McCollum, Green Level's town clerk, testified that when "you would go down the streets, you could not get through to get to your home because drug activities was [sic] in the street." (Id. at 63.)

Green Level's mobile home communities, especially the Seamsters and Otter Creek mobile home parks, have been frequent sites of significant illegal drug and criminal activity. (Id. at 65, 67, 71, 75.) The Otter Creek mobile home park has suffered murders, drug activity, and kidnappings. (Id. at 75.) The owner of Seamsters has sought law enforcement help in combatting drug and gang activity in his park. (Id. at 72.)

Starting in 2002, Green Level's council members—all African American—contracted with ACSO for placement of a "substation," which included the assignment of an ACSO sergeant, at Green Level's town hall. (Id. at 63.) Council members also requested increased ACSO policing, which required more frequent patrols and checkpoints (id. at 69), and more frequent inspections of the town's mobile homes by both ACSO and the Alamance County Inspection Department to enforce town ordinances against abandoned trailers being used for criminal activity. (Id. at 68–70.)

By 2010, Alamance County's rising drug trade led the DOJ and the DEA to designate the County a "High Intensity Drug Trafficking Area," a classification reserved for counties representing "a high threat of drug trafficking, of violence." (Id. at 106, 113, 115.) The County was one of only seven in the State to be so designated (id. at 115), and it retained the designation at the time of trial. (Id. ) According to Agent Serniak, who participated in many of the DEA's activities in the County, Mexican DTOs remain a "danger" to Alamance County residents. (Id. at 113.)

B. ACSO and Sheriff Johnson

This is the environment in which Terry Johnson was elected ACSO Sheriff in December 2002 (following a 30–year career with the North Carolina State Bureau of Investigation ("SBI")), a position to which he has since been continuously re-elected. (Doc. 154 at 55–56.) In North Carolina, the position of sheriff is constitutionally-provided for, and a sheriff's office is a separate legal entity from each county. See N.C. Const. Art. VII, § 2 ; N.C. Gen.Stat. § 162–1 ; Cranford v. Frick, No. 1:05CV00062, 2007 WL 676687, at *3 (M.D.N.C. Feb. 28, 2007) ("[T]he Office of the Sheriff is separate and distinct from the Board of County Commissioners because a sheriff is elected by the people, not employed by the county." (quoting Little v. Smith, 114 F.Supp.2d 437, 446 (W.D.N.C.2000) ) (internal quotation marks omitted)); Goodwin v. Furr, 25 F.Supp.2d 713, 715–16 (M.D.N.C.1998) (noting that a sheriff is an "independently elected official" and "a county is not liable for the acts of the sheriff and deputies"); Clark v. Burke Cnty., 117 N.C.App. 85, 450 S.E.2d 747, 749 (1994) ("A deputy is an employee of the sheriff, not the county."). ACSO is the largest law enforcement agency in the County, employing approximately 123 full-time deputies and 147 civil employees.

As the head of ACSO, Sheriff Johnson is vested with the authority to set its policies and procedures, which he implements both individually and through his officers. (Doc. 154 at 89; N.C. Gen.Stat. § 162–24.) The Government contends that, since he has taken control, Sheriff Johnson and ACSO have engaged in a pattern and practice of discriminatorily enforcing the law against Hispanics. The Government's evidence as to the various ways in which this was allegedly done is addressed below.

1. The Introduction and Implementation of ICE's 287(g) Program in Alamance County

A principal feature of the Government's case rests on its claim that ACSO sought out and used federally-granted immigration authority in a discriminatory manner against Hispanics. (See Doc. 1 ¶¶ 47–51; Doc. 158 at 61–63.) The thrust of the Government's claim is that Sheriff Johnson abused authority granted under Section 287(g) of the Immigration and Nationality Act, 8 U.S.C. § 1357(g), to direct ACSO officers to target and arrest Hispanics so that they could be detained at the Alamance County Detention Center (the "ACDC") and ultimately processed by federal authorities for deportation.

Section 287(g) authorizes the Attorney General to delegate federal immigration enforcement authority to certain State employees. Known as "287(g) programs," these arrangements permit state and local law enforcement officers to investigate, apprehend, and detain aliens in the U.S. See 8 U.S.C. 1357(g)(1). To investigate the possibility of employing a 287(g) program, Sheriff Johnson contacted the Sheriff of Mecklenburg County (Charlotte), North Carolina, who had the first such program in the State, to learn about his experience with it. (Doc. 154 at 157.) That sheriff found it to be the "best crime fighting tool that he had ever been involved with." (Id. ) Sheriff Johnson eventually flew to Alabama to observe how the program was implemented there, and, in about 2006, he spoke with ICE officers. (Id. ) Satisfied with what he had learned, and at the urging of an Alamance County Commissioner, the Sheriff applied for participation in the program through ICE in Washington, D.C.; by January 2007, ACSO entered into a 287(g) agreement—known as the Memorandum of Agreement ("MOA")—with ICE and Alamance County. (Id. )

According to Sheriff Johnson, the 287(g) program offered a tool to increase safety within the ACDC. (Id. at 81.) Prior to the 287(g) program, ACSO found itself often unknowingly placing rival Hispanic gang members within the same prison cell, causing fights, increasing medical costs, and giving rise to lawsuits. (Id. ) Via the MOA, ACSO officers could better identify detainees and coordinate with federal agents who enforced immigration violations. (Doc. 149 at 127–28.) Also, the 287(g) program offered a financial incentive, as ACSO would be paid by ICE for the cost of housing all ICE detainees at the ACDC, irrespective of whether the detainees were originally arrested by ACSO, ICE, or another of the many law enforcement agencies using the ACDC as a jail. (Doc. 147 at 151–52.)

One aspect of the MOA was a "jail enforcement program," by which ACSO officers trained through ICE would receive immigration enforcement certification and, once certified, would return to ACSO to enforce federal immigration laws within the ACDC. (Doc. 149 at 127–28.) ICE also stationed its own agents within the ACDC to supervise those 287(g)-certified ACSO officers. (Id. at 138–39; Doc. 152 at 157–58; Doc. 154 at 33–34.) While anyone arrested by ACSO and jailed could be turned over to ICE within the ACDC, who could determine whether they were in the country illegally, the ACDC also served as a detention center for persons ICE and eleven other law enforcement agencies arrested (independently of ACSO) in the surrounding counties. (Doc. 152 at 164; Doc. 154 at 23–24.) In practice, about only one detainee a week was processed through the ACDC's 287(g) program, and the "bulk" of those were persons brought in from ICE's other offices—not from ACSO. (Doc. 152 at 164 ("[T]he actual 287(g) processing[s] were not large numbers.... [T]he bulk of the alien population housed at Alamance County jail actually came in from ICE arrests made by ICE officers or other 287(g) officers in other counties.").) The vast majority of ICE's detainees were from Mexico. (Id. at 166.) Thus, while the Government at trial emphasized the financial incentive to ACSO for arresting and detaining persons who were not in the country legally, the reality was that the vast majority of the ICE detainees held at the ACDC were arrested by someone other than ACSO and were simply housed at the ACDC. As a result, ACSO had no involvement in their identification or apprehension.

Throughout its operation in Alamance County, ACSO's 287(g) program received annual reviews from ICE. (Id. at 148–50.) ICE found it to be "an exemplary unit." (Id. at 149.) In fact, after one review, ICE used ACSO's information as an exemplar to provide to ICE's other 287(g) units. (Id. at 149–50.) A 2011 review from ICE concluded, "The 287(g) program in Alamance County is adhering to the priorities and obligations set forth in the MOA." (Def. Trial Ex. 56 at 6.) The review also acknowledged that, in 2011, "no complaints of any kind have been received by the ACSO or ICE concerning the 287(g) program." (Id. at 5.)

ICE eventually withdrew ACSO's 287(g) MOA in 2012, however, following the complaints that led to the filing of the current lawsuit. (Doc. 149 at 126, 128.)

a. 287(g) Program's TFO

Early in Alamance County's 287(g) program in 2007, Gloria Fichou, Special Agent-in-charge of Homeland Security (now retired) in ICE's regional office over Alamance County, contacted Sheriff Johnson because the federal government hoped to designate an ACSO officer as a Task Force Officer ("TFO") for ICE's Homeland Security Investigations. (Doc. 153 at 17–18; Doc. 154 at 80.) A TFO is cloaked with 287(g) authority to perform investigations in the field, as opposed to simply in the jail. (Doc. 149 at 129.) Sheriff Johnson selected Deputy Jeff Randleman to serve as the TFO, based on Deputy Randleman's experience with investigations. (Doc. 154 at 80.) At the time of Deputy Randleman's selection in 2007, Fichou told Sheriff Johnson that Deputy Randleman could serve as a 287(g) detention officer in the ACDC and as a TFO under her supervision. (Id. at 80–81; Doc. 153 at 18–22.)

Stationed in both Alamance County and ICE's Winston–Salem office, Deputy Randleman worked as an ICE TFO for approximately one year. (Doc. 149 at 171–73; Doc. 153 at 20.) In addition to his certification as a 287(g) detention officer, he received training as a TFO. (Doc. 149 at 171; Doc. 153 at 20 (8 weeks of ICE training).) Fichou supervised Deputy Randleman, and an ICE special agent would sometimes work with Deputy Randleman as well. (Doc. 153 at 20.) Deputy Randleman's investigations as a TFO required Fichou's authorization. (Id. at 27.) Fichou, however, would occasionally turn over ICE investigations to ACSO as well. (Id. at 25.)

During this time, Deputy Randleman also remained employed by ACSO. (Doc. 149 at 171.) Thus, while a TFO, Deputy Randleman would receive work from both Fichou and Sheriff Johnson, and he would self-initiate investigations, as he had before. (Id. at 172–73, 177.) Any cases assigned to him by Sheriff Johnson were reported to his training officer, with Fichou receiving monthly reports about his ACSO work, but Deputy Randleman spent "more time" helping ICE during his time as a TFO.⁶ (Id. at 177–78.) Fichou acknowledged that she was aware of all 287(g) matters on which he worked. (Doc. 153 at 20–21.)⁷

6 The Government cites Deputy Randleman's statement that the only persons Sheriff Johnson or then-Deputy Britt requested that he investigate while acting as a TFO were Hispanic. (Doc. 149 at 208.) This statement has little capacity for persuasiveness. There is no indication how many times (if any) such requests were made apart from the limited specific ones detailed here, and Deputy Randleman clearly spent more of his time working for ICE. Moreover, for the reasons explained below, none of the specific investigations ordered by ACSO provided at trial (which may be all of the Hispanic investigations performed by Deputy Randleman) evidences an intent to target Hispanics.

7 Deputy Randleman believed that Fichou was aware of his ACSO matters, too. (Doc. 149 at 178.) But any significant knowledge appears doubtful, as oversight for those matters was the responsibility of Sheriff Johnson.

While serving as a TFO, Deputy Randleman performed a number of tasks. For one, he would assist in gang roundups. (Doc. 149 at 174.) Organized and coordinated by ICE and the DEA, these are operations in which law enforcement agents execute a number of arrest warrants on certified gang members. (Doc. 147 at 75; Doc. 149 at 174; see also Doc. 152 at 102–03 (stating that the DEA would also run gang roundups in Alamance County).) At ICE's request and prior to executing warrants, ACSO's gang unit would provide ICE's 287(g) officers with a list of certified gang members. (Doc. 147 at 75; Doc. 151 at 124, 139–41.) Both before and after his time as a TFO, Deputy Randleman—who had access to several databases that included North Carolina Department of Motor Vehicle records—would conduct background checks on gang members prior to the roundup operations. (Doc. 149 at 173–75.) The majority of the gang members involved in these ICE operations were Hispanic, but Asian, South African, and South American gang members were also involved.⁸ (Id. at 136, 174.)

8 In addition to gang roundups, ICE implemented (and continues to implement) within Alamance County a yearly program called "Operation Community Shield." (Doc. 153 at 22–24.) This program requires ICE to "reach out" to local law enforcement agencies to retrieve a list of identified gang members in the community. (Id. at 23.) The program's ultimate goal is "to identify, arrest, and remove undocumented nationals deemed to be gang members in the United States." (Id. at 22–23.)

Deputy Randleman also worked on numerous identity fraud and theft cases while serving as a TFO. The Government cites one of the cases as evidence of ACSO targeting—an identity fraud case he investigated involving an individual named Marxavi Angel–Martinez. (Id. at 179.) The case arose when an Alamance County employee reported to ACSO that a Hispanic employee at the Alamance County library was receiving maternity benefits and food stamps illegally through another person's Social Security number. (Id. at 179–80, 201–02; Doc. 154 at 86.) ACSO checked the names of the library employees, and only one name stood out as being possibly Hispanic. (Doc. 154 at 87, 102.) Sheriff Johnson asked Deputy Randleman to investigate, and, working with an ICE agent, Angel–Martinez was identified as a suspect. (Doc. 149 at 179–80; Doc. 154 at 87.) The case was referred to federal prosecutors, who later charged Angel–Martinez with Social Security fraud. (Doc. 149 at 201.)

The Government also points to an instance when Deputy Randleman assisted in an identity fraud investigation requested by Alamance County Manager David Smith. (Id. at 181–83; Doc. 154 at 26–28.) Smith asked ACSO Chief Deputy Timothy Britt to verify the Social Security numbers of several newly-hired employees. (Doc. 154 at 27.) Of the approximately fifteen names provided to Chief Deputy Britt, only one was not Hispanic. (Doc. 149 at 182.) Working with Deputy Randleman, an ICE Agent contacted the Social Security Administration with the list of names and learned that one of the names was fraudulent. (Id. at 182–83.)

The Government also cites Deputy Randleman's TFO investigation of a complaint of identity theft raised by Alamance County resident Kay Oliver. (Id. at 183.) Oliver reported to Sheriff Johnson that he was the victim of identity theft, and the Sheriff instructed Deputy Randleman to "look into it." (Id. at 184.) Deputy Randleman ascertained that Juan Ariano Vazquez in Weaverville, North Carolina, was illegally using Oliver's Social Security number. (Id. at 183–85.) Deputy Randleman secured a warrant, arrested Vazquez in Buncombe County (approximately 200 miles away), and brought him to Alamance County to be charged. (Id. at 185–86.) The Government characterizes Deputy Randleman's efforts as an extraordinary measure that reflects an intent to enforce the law more stringently against Hispanics, noting that Oliver had a reputation in the community of being anti-immigrant. (Id. at 189–90.) Deputy Randleman testified, however, that Sheriff Johnson never directed him to make the trip, it was the only occasion as a TFO where he went outside the County to make an arrest, and executing a warrant outside of Alamance County is nevertheless a "part of regular law enforcement." (Id. at 185–86, 203–04.) Ultimately, the U.S. Attorney's Office declined to prosecute the case (for reasons never explained at trial), although the Government never offered any indication that Vazquez's arrest lacked probable cause. (Id. at 190–91; 204.) The case was prosecuted by the Alamance County district attorney. (Doc. 149 at 205.)

Finally, the Government cites a traffic accident investigation Deputy Randleman conducted while a TFO in 2007 or 2008. (Id. at 191; Doc. 154 at 167–68.) The matter began when Alamance County Commissioner Ann Vaughan came to Sheriff Johnson's office, "raising Cain" following an accident. (Doc. 154 at 76.) She complained that the other driver was a Hispanic man who cut her off in traffic, caused her to hit his car, then left the scene of the accident to refuel. (Id. at 168.) Apparently before he left to put gas in his car, the driver told Vaughn not to call the police when she reached for her phone. (Id. at 168, 172.) This struck a nerve with Vaughan, who told the man, "[I]n this country, we report all accidents," and then called the Burlington Police Department. (Id. at 168.) The officers responded but ultimately declined to cite the other driver for the accident or for apparently having three adults and three children unbelted in the front seat. (Id. )

Vaughan was upset. She showed Sheriff Johnson her accident report and demanded that he do something. (Id. at 168–69.) Sheriff Johnson asked if the driver was in the country illegally, and she responded that a red flag went up because of his urging her not to call the police. (Id. ) Sheriff Johnson said, "We are not supposed to do this, but let me see if my man is busy." (Id. at 169.) He summoned Deputy Randleman, showed him a photo of the driver (apparently from a driver's license), and said that the person, who appeared to the deputy to be Hispanic, lived in a mobile home park off North Church Street, an area known to be predominantly Hispanic. (Doc. 149 at 191–92.) Sheriff Johnson gave Deputy Randleman no reason to suspect the driver was an illegal immigrant and simply directed him to "follow up on it." (Id. at 192.)

The Sheriff explains that he was reluctant to get involved because traffic citations within the City of Burlington lay within the jurisdiction of the Burlington Police Department. Nevertheless, Deputy Randleman ran the individual's name through the immigration database, and the search yielded no results. (Id. at 193.) The Sheriff reported to Commissioner Vaughn that the driver had a valid North Carolina driver's license and was not an illegal immigrant. (Doc. 154 at 169–70.) With this, the investigation ended.

The Government characterizes Deputy Randleman's inquiry as an abuse of his TFO authority and evidence of an intent to target Hispanics. While the date of the inquiry was not made clear at trial, if the matter occurred during Deputy Randleman's role as TFO, he had at least apparent authority to investigate the driver's immigration status. It bears noting that Vaughn presented at trial as an outspoken and assertive witness; she is also a political adversary of the Sheriff. The court is persuaded that, but for Vaughn's persistence and will, the Sheriff never would have gotten involved in the inquiry.

b. Termination of TFO Position

In about 2008, an ICE audit revealed an issue with the MOA as to Deputy Randleman's TFO authority. (Doc. 149 at 172; Doc. 152 at 160; Gov't Trial Ex. 58; Def. Trial Ex. 54.) Jill Arndt, ICE supervisor (now retired), had reported a concern about it but had been "getting conflicting information from ICE['s] Office of Investigations" about ICE's ability to use Deputy Randleman as a TFO. (Doc. 152 at 160.) According to Fichou, "there was no clearcut policy in regards to the 287(g) program" at the time she approached Sheriff Johnson to request a TFO. (Doc. 153 at 22.) And Fichou had never questioned Deputy Randleman's authority to operate as a TFO. (Id. at 21–22.) ICE eventually interpreted the MOA not to authorize Deputy Randleman to be a TFO within ICE's Homeland Security Investigations (id. at 21–22), so ICE withdrew Deputy Randleman's TFO authority, and he promptly discontinued any further work as a TFO, returning to his position as a 287(g) detention officer in the ACDC. (Doc. 149 at 172, 175.)

Up to and during trial, the Government portrayed Sheriff Johnson's designation of Deputy Randleman as TFO, the deputy's activities as a TFO, and ICE's withdrawal of TFO authority as evidence that ACSO abused its authority under the 287(g) program by engaging in investigations outside the permission granted by the MOA. However, federal ICE agents—called by the Sheriff—made clear that this was not the case. Fichou, Special Agent-in-charge of ICE's regional office over Alamance County at the time, testified credibly at trial that it was she who had approached the Sheriff about the possibility of such an arrangement (Doc. 153 at 18) and that any mistake with regard to the TFO's authority was ICE's, not ACSO's (id. at 21–22). Arndt confirmed this assessment: "I believe it was purely an ICE issue as far as ICE, and don't get me wrong. Nobody did this on purpose. ICE was giving conflicting information; and once it was discovered in the management audit, as it was supposed to, it was fixed." (Doc. 152 at 166.)

c. Gun Permit Investigations

Under North Carolina law, county sheriffs' offices bear responsibility for issuing gun permits. See N.C. Gen.Stat. § 14–404. The law specifically requires that county sheriffs verify that "it is not a violation of State or federal law for the applicant to purchase, transfer, receive, or possess a handgun." Id.⁹ In accordance with the law, a clerk at ACSO handles background checks for gun permits with subsequent review by ACSO captains and a major. (Doc. 154 at 38–39.) Former Chief Deputy Morris McPherson oversaw this process at ACSO until he retired in 2009.¹⁰ (Id. at 41.)

9 N.C. Gen.Stat. § 14–404 was amended several times during the period of inquiry for this case and has also been amended since the Government initiated its lawsuit. Neither party, however, contends that any amendment to the statute affects the court's analysis.

10 Chief Deputy McPherson died in 2012. (Doc. 154 at 41.)

Through the duration of the 287(g) program, ACSO used its access to immigration databases to investigate individuals applying for gun permits. (Doc. 149 at 132–33.) While working as a certified ICE agent in the ACDC, ACSO Lieutenant Randy Denham ran searches through these databases following requests from ACSO Majors Shelton Brown and Monte Holland. (Id. at 133, 142–43.) These searches checked criminal history and, because it is a violation of federal law for aliens illegally or unlawfully in the United States to possess a firearm, see 18 U.S.C. § 922(g)(5), immigration violations. (Doc. 149 at 152.) At times, Lieutenant Denham conducted such checks weekly. (Id. at 134.)

The Government correctly contends that the use of any ICE database by ACSO officers (other than by a TFO before such authority was rescinded) for gun application checks exceeded ACSO's authority under the MOA because the applicants were not in ICE custody. Lieutenant Denham did not appear to be aware of that, however, and he maintained that his ICE supervisor was aware of his investigatory background checks and approved them. (Id. at 143.) Fichou and Arndt—both female—denied awareness of the background checks (Doc. 152 at 169; Doc. 153 at 29– 30), yet Lieutenant Denham's statement is still credible because he testified that he stopped the checks in 2011, once his ICE supervisor—a male—raised a concern. (Doc. 149 at 133 ("[T]he supervisor over deportation and detention for ICE, come to me, and he was concerned that this may be a problem, and he decided that we need to stop." (emphasis added)), 147, 151.) Sheriff Johnson was never aware that ACSO used the 287(g) program to carry out ACSO's statutory obligation to investigate gun permits. (Doc. 154 at 88.)

The Government does not suggest that the checking of ICE databases in and of itself violated any law, and presumably ACSO officers could lawfully ask ICE to do so in order to discharge their obligation to avoid issuing gun permits to ineligible aliens. The Government's main contention seems to be that all names submitted for background checks sounded Hispanic (Doc. 149 at 134), yet not all applicants were Hispanic, thus evidencing a potentially discriminatory investigative practice. However, the Government elicited no evidence as to the makeup of gun permit applicants or ACSO's procedures for checking such applications. Thus, it is unknown whether ACSO checked all applicants' names for legal status and through some different means determined that the others were U.S. citizens. Although the majors were the officers referring the names to Lieutenant Denham, the Government did not call Major Holland and, when it called Major Brown as a witness, failed to ask a single question about the gun permit background checks.

d. ACSO's Processing and Booking Procedures

The Government contends that ACSO changed its arrestee booking procedures to target Hispanics "for heightened immigration enforcement" after entering into the 287(g) MOA. (Doc. 1 ¶¶ 47–51; Doc. 158 at 65–66, 132.) According to the Government, after the start of the 287(g) program, ACSO required that all persons arrested be "fully booked" into the ACDC, subjecting more individuals—namely Hispanics—to questioning about their immigration status. (Doc. 158 at 65, 132.)

The Government relies mainly on the testimony of North Carolina Magistrate Susan Wortinger. (Id. at 61–62, 65–66, 132.) Magistrate Wortinger has served 12 years in her position, having previously been employed by ACSO from 1996 to 2002. (Doc. 149 at 155–56.) Her office is in the ACDC. (Id. at 157.)

According to Magistrate Wortinger, before ACSO participated in the 287(g) program, it only booked felony arrestees into its jail. (Id. at 159–60.) She stated that ACSO's policy changed after implementation of the 287(g) program to require persons arrested for misdemeanors to be booked as well, even if they could pay their fine or the bond. (Id.; see also Doc. 147 at 43–44 (Evans stating that the policy "probably" changed around 2007).) Magistrate Wortinger observed more Hispanics come through the ACDC after the 287(g) program and the change in booking procedure. (Doc. 149 at 162.) She, however, inaccurately believed that the 287(g) program lasted "[m]aybe two years," when it in fact lasted closer to five years. (Doc. 149 at 158; id. at 126, 128; Doc. 154 at 157.) Citing this testimony, the Government contends that ACSO changed its booking procedures following the 287(g) program to check the citizenship of misdemeanor arrestees to facilitate their deportation. (Doc. 158 at 132.) The persuasive evidence at trial, however, conflicts with this account of ACSO's use of the 287(g) program.

Sheriff Johnson changed ACSO's procedures well before implementation of the 287(g) program. North Carolina law permits law enforcement to photograph and fingerprint arrestees, whether or not they are later committed to a jail, unless they are only charged with certain lower level motor vehicle misdemeanors. N.C. Gen.Stat. § 15A–502(a).¹¹ In accordance with the law and several years before implementation of the 287(g) program, Sheriff Johnson instituted a policy requiring officers to fingerprint and photograph felony as well as misdemeanor arrestees, including those who have no driver's license and cannot be identified, irrespective of ethnicity or race. (Doc. 154 at 58–59; see also Doc. 147 at 33 (Perry—called by the Government—testifying that Sheriff Johnson changed the policy a few years before the 287(g) program was implemented).)¹² The Sheriff explained that he did this to remedy a developing problem:

11 Even when charging those lower level motor vehicle misdemeanors, North Carolina law enforcement may still photograph a person at the scene where cited or, if arrested, at the jail if that person fails to produce a valid driver's license and the officer has reasonable suspicion concerning his or her true identity. N.C. Gen.Stat. § 15A–502(b).

12 ACSO Deputy Steven Perry testified that in the mid–2000s "we were told to start—anybody we brought in on an arrest, we had to fingerprint for any charge, misdemeanor or felony" and not just for felonies as in the past. (Doc. 147 at 26.) The court does not read this summary description of the new policy to necessarily contravene the limitations of N.C. Gen.Stat. § 15A–502.

[T]he reason for that was that we were having people arrested under different names; and when I come in the office, we had an overcrowding in the old jail because our new jail had not been finished. And when I went down to look at the records, I was seeing people's same picture but under different names, and also people were not showing up for court when they were arrested or given a citation.

(Doc. 154 at 58.) The new procedure, permitted by North Carolina law, enabled ACSO to properly identify all persons arrested.

The evidence at trial was unclear as to how Sheriff Johnson's photographing and fingerprinting policy affected, if at all, the number of individuals "booked" into the ACDC and subject to 287(g) questioning. After an arrested individual was brought to the ACDC, the arresting officer completed arrest and property sheets. (Doc. 152 at 135–38; see also Doc. 154 21–23.) In doing so, the arresting officer inquired whether the arrestee was a U.S. citizen. (Doc. 152 at 138.) This process was completed "in the prebooking area" where the arrestee went before the magistrate. (Id. )

Then, after a judicial officer (usually the magistrate, but sometimes a judge if not at the ACDC) found probable cause for the arrest, he or she set the arrestee's bond and determined whether the individual could pay the bond so as to be released. (Doc. 152 at 138–39; Doc. 154 at 23–24.) If an arrestee could not meet bond, he or she would proceed to the booking area and there be subject to possible 287(g) questioning depending on the response to the citizenship question provided on the arrest and property sheets. (Doc. 152 at 138–40; see also Def. Trial Ex. 52.) If an arrestee could meet bond, however, there was persuasive evidence that he or she might be able to leave without being subject to 287(g) questioning.¹³The Government's complaint further alleged that arrestees are referred for 287(g) questioning based on ACSO officers' assumptions about nationality or ethnicity and thus that ACSO targets Hispanics for immigration questioning. (Doc. 1 ¶¶ 48, 50–51.) To support this allegation, the Government relies on its leading question to Major Alan Miles, who agreed that, during the 287(g) program, "it was important to [him] to make sure that people who appeared to be of foreign descent were booked into the jail." (Doc. 149 at 214.) This was the sole statement regarding the misuse of the 287(g) program within the ACDC, and there is no other evidence on record to support it. Even taking Major Miles' response at face value, there is no evidence that he or other detention officers, who were stationed in the ACDC, could control whether an arrestee was booked into the jail.

13 Sergeant Dan Cubino, a Hispanic officer who worked in the ACDC during the entire 287(g) program and whom the court finds highly credible, testified that, if an arrestee made bond, the individual did not enter the booking area and thus presumably would not be subject to 287(g) questioning. (Doc. 152 at 138–40.) Others testified, though, that arrestees subject to fingerprinting and photographing at least had to enter the booking area, so that detention officers could fingerprint and photograph them. (Doc. 147 at 43 (Evans explaining that detention officers would fingerprint and photograph arrestees but making no mention as to whether they would also then be subject to 287(g) questioning); Doc. 149 at 160 (Magistrate Wortinger testifying that "[i]f [an arrestee] had the money for the fine or the bond, they still had to go into the jail"); Doc. 154 at 22–23 (Chief Deputy Britt stating that detention took arrestees' photographs and fingerprints so that ACSO could turn officers "back out on the street").) Neither party satisfactorily addressed whether, after fingerprinting and photographing, an arrestee would likely be subject to 287(g) questioning despite meeting bond. (See Def. Ex. 53.)

Moreover, there is no evidence as to how many (or few) Hispanics arrested by ACSO were booked into the ACDC and subjected to 287(g) questioning. Certainly, the Government never presented any booking reports or other similar evidence from ICE or ACSO to support such a claim. Rather, ICE witnesses called by ACSO, explained that ACSO's 287(g) officers processed only about one detainee per week at the ACDC through the 287(g) program, and "the bulk" of those detainees actually came from other ICE offices. (Doc. 152 at 163–64.) Thus, in actuality, ACSO officers contributed only infrequently to the number of ICE detainees housed in the ACDC.

To the extent that Magistrate Wortinger observed more Hispanics "come through the jail" after the implementation of 287(g), this observation aligns with the testimony of former ICE Special Agent Jill Arndt. (Doc. 149 at 162.) At trial, she explained,

[T]he bulk of the alien population housed at Alamance County jail actually came in from ICE arrests made by ICE officers or other 287(g) officers in other counties, and Alamance County jail was actually housing ICE detainees, and the bulk of the numbers at Alamance County jail were not 287(g) arrests from Alamance County. They were, in fact, people who had been arrested from ICE officers at other 287(g) locations, and we housed those individuals at the Alamance County jail.

(Doc. 152 at 164.) Chief Deputy Britt added that eleven "contributing" State law enforcement agencies also used ACSO's facilities for booking. (Doc. 154 at 24; see also Doc. 152 at 138–40 (Sergeant Dan Cubino testifying that, no matter the jurisdiction, an arrestee was taken before the magistrate).) Thus, Magistrate Wortinger's observations of increased Hispanics at the ACDC are consistent with Arndt's account of the 287(g) program's operations but reflect an apparent misunderstanding of the reasons for the increased traffic. The increase in traffic was predominantly caused by some combination of ICE's arrests and those of the eleven law enforcement agencies, other than ACSO, using the ACDC.

Lastly, Magistrate Wortinger testified that, during the operation of the 287(g) program, she saw Officer Troy Anthony on one occasion with a "stack" of alien registration cards and U.S. passports. (Doc. 149 at 163–64.) Officer Anthony referred to them as his "collection." (Id. at 164.) Magistrate Wortinger saw one of the documents, which appeared to belong to a Hispanic. (Id. ) She did not know whether the documents were forged and reported her observation to Officer Anthony's supervisor, Brandon Wilkerson, yet she never followed up to determine what, if anything, became of it. (Id. at 164–65, 166–67.) She concedes that she is unaware of ACSO's policy for taking custody of forged instruments or what the deputies actually did with forgeries they came across during their investigations. (Id. at 166.) Although Officer Anthony "regularly" came before her for some period of time, she never found that any of his arrests lacked probable cause. (Id. at 167.)

Officer Anthony denies that he ever referred to a personal collection to Magistrate Wortinger or that he kept any seized items permanently in his possession. (Doc. 152 at 85–86.) The court credits Magistrate Wortinger's testimony. Officer Anthony's conduct on this occasion was a violation of ACSO's evidence collection policy, which required that all documents seized from detainees be preserved in the evidence locker.¹⁴ (Id.; Doc. 154 at 77–78.)

14 The Government does not tie this event to any claim in the case. (See Doc. 158 at 60–61 (referring to testimony in Findings of Fact).)

e. Fairness Alamance

In 2008, as a result of publicity about the 287(g) program, a grass-roots collection of individuals in Alamance County formed Fairness Alamance to air concerns about what the group believed was ACSO's unfair treatment of Hispanics. (Doc. 151 at 143–44.) Members attended Alamance County Commissioners' meetings, held their own meetings, and organized a candlelight vigil in support of their views. (Id. at 148–50.)

In response to the group's expressed concerns, Sheriff Johnson invited its members to the ACDC in 2008, provided a tour of the facility, outlined the 287(g) program, and addressed members' questions. (Id. at 151.) In February 2009, the Sheriff also gave a presentation on ACSO's traffic stops at an Alamance County Commissioners' public meeting. (Id. at 151–52.) The presentation provided information based on the traffic stop data ACSO sent to North Carolina's SBI. (Id. ) In his presentation, Sheriff Johnson invited citizens to personally examine the traffic stop data on the SBI's website. (Id. at 152–53.)

A leader of Fairness Alamance, Laura Roselle, who served as Professor of Political Science and Policy Studies at Elon University, took the Sheriff up on his invitation. (Id. at 142, 153.) On February 16, 2009, following her review, Professor Roselle emailed Sheriff Johnson, hoping to discuss the SBI's data. (Id. at 154–55.) Sheriff Johnson did not respond to her email, and so she asked him about the data and ACSO's traffic stop forms at the next County Commissioners' meeting. (Id. at 155.) The Sheriff incorrectly responded that he did not have the forms and that they were in Raleigh. (Id. at 154–55.) Ultimately, Professor Roselle learned that the SBI's data contained four times as many citations as traffic stops; although, she did not realize that ACSO's citations also include non-traffic related offenses. (Id. at 158, 182.) Upon finding this discrepancy, and rather than following up with the Sheriff to notify him of a potential problem, Professor Roselle instead met with a County Commissioner and went to a local newspaper to report what she considered to be inconsistencies in the numbers. (Id. at 159.)

After the news story went to print, ACSO acknowledged that it had independently found a "glitch" in its computer software, causing the underreporting of traffic stops. (Id. ) ACSO had been in the process of changing its software for the third time in a year. (Doc. 154 at 20.) ACSO informed the SBI of its underreporting, found a new software provider to correct the problem, and instituted administrative review over its traffic-stop data. (Id. ) After these changes, Professor Roselle's requests for corrected data went unanswered. (Doc. 151 at 161.)

In addition to traffic stop data, Professor Roselle made a request to ACSO for information on the implementation of 287(g)—specifically for ICE's detainee records.¹⁵ (Id. at 161–62.) ACSO referred her to ICE for access to those records. (Id. at 162.) Undeterred, during a March 2009 County Commissioners' meeting, Professor Roselle again requested the records—this time publicly—from the Sheriff, who was in attendance. (Id. at 163–64.) And after the meeting, she approached him to follow up. (Id. at 164.) He explained that, under the MOA with ICE, he lacked the authority to release the records, so she needed to request them from the federal government—specifically ICE.¹⁶ (Doc. 154 at 64.) At some time during his discussion with her, Sheriff Johnson placed his hand on Professor Roselle's shoulder; this caused her to bristle, and she became quite upset. (Id. ) Sheriff Johnson reacted by saying he was "done" with her, raising his hands in the air. (Doc. 151 at 165; Doc. 154 at 64–65.)

15 There is no record of a federal FOIA request or North Carolina public records request under N.C. Gen.Stat. § 132–1.

16 Professor Roselle conceded at trial that she never pursued her requests with ICE. (Doc. 151 at 176–78.)

The Government offers this encounter as evidence of anti-immigrant animus by the Sheriff. This is an over-read of the situation. Professor Roselle presented at trial as a deeply-motivated advocate who was personally opposed to what she perceived as anti-immigrant efforts by ACSO, distrusted Sheriff Johnson, and was zealous in her persistence.¹⁷

17 Professor Roselle testified to several situations where she believed ACSO surveilled her and tapped her phone, although no proof of such was ever presented. She ultimately moved to a neighboring county solely because of her dislike of ACSO. (Doc. 151 at 167–73.)

2. ACSO Arrest Policy and Practice

The Government contends that Sheriff Johnson orders officers to arrest Hispanics, rather than to cite or warn them. (See Doc. 1 ¶¶ 5, 29; Doc. 158 at 2–3, 9–14.) The Government's proof on this point relies on several witnesses and on statistical evidence (largely through Dr. MacDonald, its expert criminologist and statistician, discussed below).

The Government presented three witnesses who testified that, during one ACSO staff meeting, and perhaps two, Sheriff Johnson directed officers to arrest Hispanics stopped for traffic violations. Kenneth Evans, former ACSO chief deputy, testified that, at a monthly staff meeting (normally held the second Wednesday of every month) around 2007, Sheriff Johnson told officers that if they "went to an Hispanic's house or [they] stopped one for a traffic violation or if it was a violation, that [they] were to bring them to jail and not write a citation." (Doc. 147 at 45.) Evans said he did not pass the instruction on to officers under his command, however, "[b]ecause I knew it was wrong," and there is absolutely no evidence he ever carried it out. (Id. at 45–46.) Similarly, Officer Roger Lloyd testified that at a monthly staff meeting Sheriff Johnson told them that Hispanics without a driver's license should be arrested and brought to jail, where they would be run through the 287(g) program. (Id. at 151–54.) Officer Lloyd said that the Sheriff had a similar conversation with him separately and explained that doing so would allow Hispanic detainees to be run through the 287(g) program to be deported, allowing the County to receive money for the federal detention. (Id. at 151–52.) Officer Lloyd was unable to say when such discussion occurred but insisted that he never discriminated against any Hispanic and knew of no ACSO officer who ever had. (Id. at 152–54, 168.) He also conceded on cross-examination that he understood (and his personal practice was) that if he could not identify any suspect, he would arrest him or her so they could be processed and their identity determined. (Id. at 170.) Finally, Brenda Cole—a former ACSO officer—testified that she heard Sheriff Johnson give the same instruction (to arrest Hispanics) at a monthly staff meeting in 2007 or 2008, but could offer no other context for the statement. (Doc. 151 at 5, 7.) Like the others, Cole insisted that she never carried out the instruction. (Id. at 6.)¹⁸

18 All three witnesses started at ACSO before Sheriff Johnson became sheriff; Officer Lloyd left in 2009, Cole in 2011, and Evans in 2011. (Doc. 147 at 36–38, 57–58, 144, 179; Doc. 151 at 3; Doc. 154 at 41–42.)

No party examined any of these three witnesses as to the context of the statements, but Sheriff Johnson did provide some context and presented evidence challenging the credibility of all three. The Sheriff had fired Officer Lloyd in 2009 for lying to an Alamance County Assistant District Attorney. (Doc. 147 at 162, 171–75; Doc. 154 at 43–44; Def. Trial Ex. 67.) He had also fired Evans' wife and nephew from ACSO and demoted Evans upon taking office. (Doc. 154 at 73–74.) And, Cole had supported Sheriff Johnson's opponent, Ron Parrish, in the most recent election. (Doc. 151 at 6.)

Nine ACSO employees—some of whom were offered by the Government—testified that they had never heard Sheriff Johnson give an instruction to single out Hispanics for arrest or to do so irrespective of whether they could show identification. (See Doc. 150 at 30, 134, 157, 184; Doc. 151 at 136; Doc. 152 at 175–76, 215; Doc. 154 at 4, 37.) Sheriff Johnson similarly denied ever doing so. (Doc. 154 at 56–57.) Rather, the Sheriff stated that his command left all arrests to the officer's proper discretion, so that "if" an officer arrested an individual stopped without any identification, the individual should be brought in for processing. (Id. at 57–58.) Lieutenant Allen, who attended many of the monthly meetings between 2007 and 2013 and recorded his notes in a binder, never heard any statement singling out Hispanics for arrest, and his contemporaneous notes, albeit not exhaustive, reflect no instruction to support the claimed statement. (Doc. 152 at 215–18; Doc. 153 at 7–9.) ACSO Detective Carlos Rossi, who is Hispanic and has been with ACSO for five and a half years, never heard the Sheriff or anyone else discriminate against or target Hispanics. (Doc. 153 at 212; Doc. 154 at 4.)

There was no evidence of a written ACSO policy. ACSO's own evidence, while not entirely consistent, did offer some context for such claims, however. ACSO Chief Deputy Britt testified that he understood ACSO's policy to be that an officer was to arrest any stopped individual whom the officer cannot identify and who does not have a driver's license or other method of identification.¹⁹ (Doc. 154 at 37–38.) Lieutenant Brian Allen similarly described ACSO's arrest policy as follows: "[I]f it was anybody that we couldn't positively identify, ... they needed to be arrested, and that's anybody." (Doc. 152 at 215–16.) As noted, this is what Officer Lloyd explained as well.

19 As noted below, this is in fact what transpired with the stop of Jose Luis Arzola, Jr., the only witness offered by the Government as an example of alleged targeting. Once he provided identification (but not proof of citizenship) after being stopped, he was not arrested, cited, or warned, but permitted to go about his way.

However, the more accurate articulation of what was actually implemented by ACSO officers is the following: if an officer stopped an individual without valid personal identification who cannot be reasonably identified, the officer may arrest the individual at the officer's discretion, assuming the arrest is for an otherwise arrestable offense. Captain Kimberly Wilson, head of ACSO's patrol division, stated that "[a]n officer is to use his discretion" when facing an individual stopped without any identification. (Doc. 150 at 136, 157.) When asked about Sheriff Johnson's arrest policy, Captain Wilson reaffirmed, "I expect [officers] to use their discretion, their sound discretion." (Id. at 157–58.) Major Shelton Brown—head of operations, which includes the patrol division—echoed Captain Wilson's reiteration of the arrest policy. (Id. at 165–66.) Major Brown testified that officers making a traffic stop had sole discretion to warn verbally, warn in writing, cite, or arrest. (Id. at 173–74.) He further maintained that complete officer discretion remained following a traffic stop, "even if the driver doesn't have a valid driver's license." (Id. at 173–74.) In sum, the officer "make[s] the choice" to arrest. (Id. at 173.) All of this is consistent with Sheriff Johnson's articulation of ACSO's policy, which appears to comport with N.C. Gen.Stat. § 15A–502.

As to the statements attributed to Sheriff Johnson, it is difficult to accept that the Sheriff made them as portrayed, given the lack of context offered for them and the number of ACSO officers who stated credibly that they never heard them. Had the Sheriff made these statements as characterized, it is implausible that an officer or employee (some of whom are Hispanic) would not have questioned his statements and complained. This is especially true in light of the Government's evidence that all orders of the Sheriff were expected to be obeyed and that all three individuals who claimed to hear such a directive universally testified that they never followed it.

3. Orders to Target Hispanics

The Government's complaint alleges that during a staff meeting after January 2007, Sheriff Johnson yelled "bring me some Mexicans" while banging on the table, and on another occasion instructed his staff to "go out there and catch me some Mexicans." (Doc. 1 ¶ 27(a)-(b).)

At trial, the first claim was supported by Officer Lloyd, who testified that he never heard the Sheriff use the word "Mexicans" but heard him say at one unidentified staff meeting: "bring me Hispanics." (Doc. 147 at 153–54.) In eliciting this statement, the Government made no effort to provide any context, and none was given. There was no testimony about banging on the table, and no other witness supported Officer Lloyd's claim. Sheriff Johnson denied ever making this statement. (Doc. 154 at 57–58.) The court is doubtful that the claimed statement was made, especially in the unsupported context the Government contends.

As to the second claim, the Government presented testimony from two former ACSO officers. In one instance, according to Corporal Adam Nicholson, Sheriff Johnson once said, "Go get the Mexicans." (Doc. 147 at 81.) While he could not recall the context, it was most likely when the owner of the Rocky Top mobile home park, which has a significant Hispanic population, came to Sheriff Johnson's office to complain about crime. Sheriff Johnson summoned Corporal Nicholson, who reported that the "Bandidos," a Mexican gang operating in that area, had been breaking and entering into trailers in the mobile home park. (Id. at 78, 88–89.) Corporal Nicholson testified that he interpreted the Sheriff's comment as a command to arrest every Hispanic in the park. (Id. at 86–87.) But as Corporal Nicholson conceded, because the park was predominantly Hispanic, such an interpretation would have resulted in the arrest of many, if not most, of the owner's paying customers. (Id. ) Insofar as the owner, who was present, never objected to whatever the Sheriff said or directed, Corporal Nicholson's interpretation appears overstated and conflicts with his own testimony. In fact, Corporal Nicholson admitted he never arrested anyone, but instead set up a license check point near the park on his own accord "to appease" the Sheriff. (Id. at 78, 87–89.) When pressed further, he conceded that he never intended to arrest non-violent or non-criminal Hispanics, but was after the "Bandidos," who were the "people who broke the law that day"; and he inferred that is what the Sheriff meant, too. (Id. at 88–89.) Corporal Nicholson also denied that he ever targeted any Hispanics or knew anyone at ACSO who did. (Id. at 92.)²⁰

20 Corporal Nicholson also testified that, in front of the mobile home park owner, Sheriff Johnson told him to "arrest every chili shitter in the park." (Doc. 147 at 77.) Corporal Nicholson has some reason for bias, as he was reprimanded for sexual harassment by Sheriff Johnson. (Doc. 147 at 92–93.). Sheriff Johnson denies he made this statement (Doc. 154 at 70), and Sergeant Christopher Crain—also present at the time—denies ever hearing any such statement. (Doc. 151 at 136.) Even assuming this epithet was used, it is some evidence of ethnic bias but, for the reasons noted, even Corporal Nicholson did not interpret it as any command to arrest non-violent or law-abiding Hispanics, but rather as a reference to the Bandidos gang members thought to be committing the crimes in the park.

This understanding is consistent with the testimony of Sergeant Christopher Crain, the Government's other witness, who worked with Corporal Nicholson and recalled an incident when Johnson said, "go get those Mexicans." (Doc. 151 at 132.)²¹ Sergeant Crain recalled the Sheriff making the statement upon learning that a Mexican gang had spray-painted an ACSO patrol car. (Id. at 132–33, 136–37; Doc. 154 at 69–70.) Notably, Crain's testimony differs from the Government's claim that Sergeant Crain heard an order to "[g]o get some Mexicans." (Doc. 158 at 11, 122 (emphasis added).)

21 This is also consistent with Sheriff Johnson's testimony that he told Corporal Nicholson and Sergeant Crain during a meeting to "go down there and get those Mexicans," referring to the Mexican gang committing the crimes. (Doc. 154 at 70.)

The persuasive evidence is that Sheriff Johnson stated "[g]o get those Mexicans" or "the Mexicans" in reference to a specific Mexican gang then under ACSO investigation for criminal activity. It does not indicate that the Sheriff ever directed his deputies to arrest individuals simply because they were from Mexico or were Hispanic.

4. ACSO Checkpoint Policy and Implementation

The Government contends that ACSO targets Hispanics through its enforcement of vehicle checkpoints. (Doc. 1 ¶¶ 41–46; Doc. 158 at 139–46.) ACSO has a written checkpoint policy in its internal policy manual which outlines the purposes and procedures for its checkpoint operations. (Gov't Trial Ex. 113.) The general purpose of the policy "is to establish guidelines and procedures for members of [ACSO] to conduct suspicion less [sic] seizures of vehicles at vehicle checking stations (checkpoints)." (Id. at 1.)

The policy provides for three permissible purposes or types of checkpoints: "standard"; "informational"; and "special operations." (Id. at 2–4; Doc. 154 at 104–05.) Standard checkpoints were the most commonly used and aim to "determine compliance with motor vehicle laws," such as "verification of drivers' licenses, vehicle registration checks, insurance checks, seat belt compliance checks[,] and driving while impaired checks." (Gov't Trial Ex. 113 at 2; Doc. 150 at 82.) To set up a standard checkpoint, the policy states that the checkpoint "shall be approved, in writing, by a Sheriff's Office supervisor or their designee." (Gov't Trial Ex. 113 at 2.) For standard checkpoints, supervisors must complete ACSO's "Checking Station Authorization Form." (Id. ) Informational checkpoints generally seek "motorists' assistance in solving a crime" and require supervisor approval—written or unwritten—prior to setup. (Id. at 3 (stating that approval must be made "in advance").) Once in place, standard and informational checkpoints must "provide for the stopping of every vehicle." (Id. at 1.) Special operations checkpoints focus on "the apprehension of a suspect who poses a danger to life and property or the rescue of a hostage or abducted person" and, under the policy, do not require the stopping of every passing vehicle. (Id. at 1, 3–4.)

ACSO's checkpoint policy does not require officers to complete a stop form—a North Carolina SBI form filled out by ACSO officers following a traffic stop—for every vehicle passing through a checkpoint, presumably for practical logistical reasons. (See Gov't Trial Ex. 59.) Instead, the policy explicitly ensures that only those vehicles physically detained at a checkpoint will require the completion of a stop form. For standard checkpoints, the policy does not require retention of any driver information, and, only after reasonable suspicion is found can an officer detain a driver for a reasonable time. (Id. at 2–3.) For informational checkpoints, the policy prohibits ACSO officers from asking for drivers' licenses or any other documentation. (Id. at 3.) Finally, for special operations checkpoints, the policy does not oblige the checking of driver information, nor does it even require that every vehicle passing through the checkpoint be stopped. (Id. at 1, 3–4.) Therefore, at checkpoints, ACSO officers fill out stop forms when an officer detains an individual beyond the initial physical stop.

ACSO Captain Kimberly Wilson reviews checkpoint paperwork filled out by ACSO officers. (Doc. 150 at 151; see also Gov't Trial Ex. 113 at 2 (requiring completion of "Checking Station Authorization Form" for standard checkpoints).) She believes she would "notice" if officers conducted "checkpoints in the same or similar locations repeatedly." (Doc. 150 at 151–52.) Major Shelton Brown confirmed that she performed this review. (Id. at 181–82.)

ACSO conducted at least four hundred and thirty-five checkpoints in the County from 2009 through 2012.²² (Doc. 153 at 93.) Some of those occurred near mobile home parks with predominately Hispanic residents. (Doc. 147 at 156–57, 188–90, 213; Doc. 148 at 18; Doc. 150 at 85.) ACSO also patrolled at or near these mobile home parks on occasion. (Doc. 149 at 216–17; Doc. 153 at 9; Doc. 154 at 106–07; Def. Trial Ex. 61.)

22 The Government suggests that there were more, pointing to testimony from a few deputies that they conducted checkpoints on a "daily basis." (See Doc. 158 at 75 n. 23; see also Doc. 150 at 81–82.) Because the policy appears to require checkpoint authorization forms for only standard checkpoints, it is possible there were more non-standard checkpoints. But there was no evidence of any alleged activity at any unreported checkpoint.

There is some indication that ACSO's checkpoint policy was not uniformly followed insofar as obtaining supervisor approval. Lieutenant Allen stated that individual officers under his supervision would select the time and location of checkpoints. (Doc. 153 at 4.) Another officer, Officer Anthony, stated that his traffic unit—which consisted of supervisors Corporal Ray and Sergeant Wilkerson—would select checkpoints over dinner. (Doc. 147 at 209–10. But see Doc. 148 at 20–22 (Officer Anthony noting that he would receive written checkpoint approval after the checkpoint operation); Doc. 150 at 181 (Major Brown observing that approval for a checkpoint can be verbal and with written approval coming afterwards).)

However, the vast majority of checkpoints were conducted within the guidelines of ACSO's checkpoint policy. (Gov't Trial Ex. 113; Doc. 154 at 104–05.) Most officers testified as to their performance of standard checkpoints. For example, Lieutenant Mark Hoover—a patrol supervisor—explained that ACSO's checkpoints are "generally ... for your ... driver's licenses or registration or insurance." (Doc. 150 at 9; see also Doc. 148 at 22–24 (Officer Anthony similarly described the general purpose of standard checkpoints.).) He further testified, "The purpose of a checkpoint, [is] to check validity of driver's license, registration, insurance." (Doc. 150 at 27–28; see also id. at 10.) Officer Bobby Culler—a North Carolina highway patrolman—similarly recalled performing only a number of "traffic check, drivers check" checkpoints in conjunction with ACSO. (Doc. 152 at 171–72.) Deputy James Conklin echoed that the "purpose of setting up" standard checkpoints is "motor vehicle violations." (Doc. 150 at 82; see also id. at 90 ("We check driver's license and registrations.... We don't set them up specifically for drugs. If we stop a vehicle and either smell or see something, that would indicate to us we move forward.").) Similarly, Sergeant Crain testified that the checkpoints he conducted "ensur[ed] compliance with North Carolina motor vehicle law primarily." (Doc. 151 at 112–13.) Similarly, Sheriff Johnson described informational checkpoints as attempts "to get information on the crime" or individuals committing crime. (Doc. 154 at 149–50.)

The Government characterizes ACSO's use of and selection of location for some checkpoints as evidence of improper purpose. For example, the Government points to testimony of Deputy Conklin and Lieutenant Hoover that they considered checkpoints to be a form of "general law enforcement." (Doc. 158 at 142–44.) Such testimony was generally elicited by the Government through leading questions and was generally clarified on cross-examination as merely the deputies' shorthand for a "standard" checkpoint. (See, e.g., Doc. 148 at 22–24 (Officer Anthony clarifying that he participated in standard checkpoints—not checkpoints for drugs—but, if he saw drugs, would take further action); Doc. 150 at 33–34 (Officer Hoover clarifying that ACSO runs checkpoints for the purpose of observing "driving violations"); id. at 89–90 (Deputy Conklin stating that ACSO does not "set [checkpoints] up specifically for drugs" but rather for motor vehicle violations).) The Government also argues that evidence that ACSO located checkpoints in high crime areas indicates they were improperly used for generalized law enforcement. (See Doc. 151 at 113–14 (pointing to Sergeant Crain's deposition testimony that "[y]ou fish where the fish are"); Doc. 154 at 106 (Sheriff Johnson's testimony that he approves of placing checkpoints in high crime areas).) Apart from what is discussed immediately below, however, there was no evidence that ACSO conducted any standard checkpoint with a purpose to target Hispanics.

Of the over 400 checkpoints conducted, the Government presented evidence relating to four.²³ The first involved Paula Crotts—a county employee at Alamance County Central Communications. (Doc. 149 at 116–17.) On a Sunday in 2010, ACSO Deputy Sara Keller stopped Crotts and her husband (both of whom are Caucasian) at a checkpoint on Monroe Holt Road near Rocky Top mobile home park, which is predominantly Hispanic. (Id. at 116–19.) Crotts' husband was driving the car and, upon approaching the deputies, retrieved his wallet and began to pull out his driver's license. (Id. at 119.) Deputy Keller leaned in, observed the two of them (Deputy Keller knew Crotts because Crotts was a substitute teacher at the high school Deputy Keller attended (id. at 119–24)), said hello to Crotts, and said that she did not need to see Mr. Crotts' license because "they were there to get them some," gesturing toward the mobile home park. (Id. at 119.) Crotts testified (without objection) that she assumed that Deputy Keller meant "she was there to get as many Hispanics—or get an Hispanic person stopped," but she conceded she did not know. (Id. at 120, 125.) Deputy Keller had no recollection of seeing Crotts at a checkpoint and denied making the statement or waving them through. (Id. at 114–15.) She did admit that she had conducted a checkpoint on Monroe Holt Road as well as another checkpoint on Kernodle Lane—the road entering the mobile home park. (Id. at 110–13.) There was also evidence that ACSO conducted a small number of checkpoints on weekends, although the policy manual discourages them from taking place near a religious institution during worship hours. (Doc. 150 at 31–32 (noting that the policy manual did not prohibit checkpoints on the weekend); Doc. 153 at 10; Doc. 153 at 4; Gov't Trial Ex. 113 at 5.) The court concludes that the checkpoint occurred as Crotts testified.

23 The complaint lists five mobile home parks that ACSO allegedly targeted with checkpoints. (Doc. 1 ¶ 45.) At trial, the Government offered evidence as to only three of those mobile home parks—Rocky Top, Seamsters, and Calloway Drive.

Second was the previously-mentioned 2007 meeting at ACSO between Corporal Nicholson, Sergeant Crain, Sheriff Johnson, and the owner of the Rocky Top Mobile Home Park. This was the meeting when the Sheriff told Corporal Nicholson to "get me those Mexicans," referring to the Bandidos gang committing crimes in the park. (Doc. 147 at 77; Doc. 154 at 69–70.) As a result of the meeting, Corporal Nicholson took it upon himself to conduct a standard license checkpoint, but he performed no arrests, wrote no citations, and filled out no traffic stop forms. (Doc. 147 at 89, 92.)

Third, three former ACSO officers offered testimony concerning a checkpoint on Highway 49 one-half mile from the Seamsters mobile home park, which is predominantly Hispanic. (Doc. 147 at 29–30, 47, 155.) Sheriff Johnson asked Officer Lloyd to set it up, and it was the only checkpoint the Sheriff asked of him. (Id. at 155.) Two other ACSO officers—Kenneth Evans and Steven Perry—were among those working the checkpoint, but as van drivers to transport arrestees to the ACDC and not as interdiction officers. (Id. at 28–29; 46.) According to Officer Lloyd, the Sheriff told him that "if we had Hispanics coming through the checkpoints that was [sic] NOL [No Operators License] or driving revoked, we was [sic] to arrest them." (Id. at 155.) Officer Lloyd, however, added, "We discussed it, and we decided we was [sic] going to also arrest any whites or blacks at these same checkpoints also." (Id. ) Evans and Perry similarly testified that Sheriff Johnson told them to arrest Mexicans or Hispanics. (Id. at 30 (Perry: "If any Mexicans violate the law, lock their ass up."), 52 (Evans: "[I]f there is [sic] any Hispanics that come through here or whatever, ... you don't write them a citation, you take [them] to jail.").) It is not clear why the Sheriff would have directed such a statement to them because neither was tasked with any responsibility to actually arrest anyone at the checkpoint.²⁴ In any event, at the conclusion of the checkpoint ACSO had arrested equal numbers of whites, blacks, and Hispanics. (Id. at 155–56.) Officer Lloyd was also adamant that he never discriminated against Hispanics nor knew any ACSO officer who had. (Id. at 168.)

24 As previously noted, two of the three officers have reason for some bias because the Sheriff had fired them or their family members.

Several ACSO officers deny this account of the checkpoint. Principally, Sheriff Johnson denied he was present and testified that he never gave such an order to arrest Hispanics. (Doc. 154 at 56.) Chief Deputy Britt, who lives in the general vicinity of the checkpoint, testified that he recalled only a checkpoint that took place after a series of break-ins in the area and that it was actually informational only. (Doc. 154 at 24–25.) According to him, its purpose was simply to inform the community and ask if they knew anything about the break-ins. (Id. ) And because it was only informational, he stated, ACSO officers performed no arrests or citations. Indeed, while ACSO maintained citation and arrest forms, there was no record that either occurred at the checkpoint. (Id. at 25.) He also denied that Sheriff Johnson was present at that checkpoint. (Id. at 26.) Finally, Officer Culler, who was a North Carolina Highway Patrol officer from 1983 to 2011, credibly acknowledged conducting checkpoints with ACSO as a "multi-agency" effort, including at Highway 49, and never knew Sheriff Johnson to use racial slurs or discriminate against anyone. (Doc. 152 at 171–74.)

Fourth, Officer Lloyd testified during the Government's direct examination that, while preparing to assist at a checkpoint at the predominantly Hispanic Rocky Top mobile home park, then-Chief Deputy McPherson initially told him that Sheriff Johnson had said that any Hispanics driving without a driver's license or driving with a revoked license should be arrested. (Doc. 147 at 156–58.) However, after Officer Lloyd and a few others sought clarification, Chief Deputy McPherson later arrived at the checkpoint, having consulted with Sheriff Johnson, and reported back that the Sheriff meant the instruction for everyone and "didn't mean just Hispanics." (Id. at 158–59.)

5. ACSO's Stops and Searches

a. Particular Stops

The Government's complaint contains broad allegations that ACSO discriminatorily stops Hispanics without reasonable suspicion, alleging several "[e]xamples of such incidents." (Doc. 1 ¶ 37.) By the time of trial, however, the vast majority of these instances never materialized, with the Government citing evidence as to only one of those alleged instances—a stop that resulted in no arrest or citation. Cf. Floyd v. New York City, 959 F.Supp.2d 540, 624–58 (S.D.N.Y.2013) (finding multiple detailed instances of suspicionless stops or frisks of African–Americans in Fourteenth Amendment selective law enforcement case). Evidence of two other stops was offered at trial, but the Government makes no reference to either of these in its post- trial proposed findings of fact and conclusions of law. All three will be addressed below.

First, at trial, the Government questioned ACSO Deputy James Conklin, a law enforcement veteran of forty-one years, about his stop of a van at night on Interstate 40 "a few years ago." (Doc. 150 at 86.) The deputy stopped the van for impeding traffic because it was driving 20 miles per hour below the posted 65 mile per hour speed limit.²⁵ (Id. at 86, 91–93.) When making the stop, Deputy Conklin approached the vehicle from behind and did not know the ethnicity of the van's driver or passengers. (Id. at 92–93.) The van was driven by a Hispanic man and had several Hispanic passengers. (Id. ) When asked for his license and registration, the driver handed Deputy Conklin a sheet of paper with a list of twelve to fifteen names, phone numbers, and dollar amounts. (Id. at 93–94.) Deputy Conklin also observed "multiple fast-food bags with trash in them" and bottles filled with what appeared to be urine. (Id. at 94.) This evidence led him to contact ICE because he suspected possible human trafficking. (Id. ) The Government cites the fact that Deputy Conklin detained the van and its occupants for approximately fifty minutes or an hour, but on cross-examination it was apparent that he did so at the direction of ICE so its agents could arrive at the scene. (Id. at 95.) The deputy could not recall whether he issued a citation or a warning. (Id. at 86–87.) He was later advised that the U.S. Attorney's Office declined to prosecute the case because it had "concerns" about the validity of the stop. (Id. at 87.) There was no testimony as to what those "concerns" were or what became of the matter, yet Deputy Conklin believed some occupants were released and some were transported to ACSO's office.²⁶

25 Impeding traffic is a violation of North Carolina traffic law. See N.C. Gen.Stat. § 20–141(h).

26 It is thus unknown whether some occupants were ultimately detained by ICE and deported.

Second, the Government offered testimony from a former Assistant U.S. Attorney, Arnold Husser, who primarily handled immigration-related cases and worked on hundreds of them over his career. (Doc. 148 at 190–91, 198.) Husser handled approximately two dozen immigration cases from Alamance County. (Id. at 198.) Of those, he declined to prosecute two cases and described only one, which he characterized as a "bad stop" because the ACSO officer stopped a car after it switched lanes several times. (Id. at 191–93.) Husser remembered that the driver was an alien but could not recall whether he was Hispanic. (Id. at 193, 197–98.) Husser's concern at the time was that the stop was improper "no matter who the driver was." (Id. at 194.) He noted, however, that during his involvement with ACSO as an Assistant U.S. Attorney, he never observed evidence that its deputies racially or ethnically profiled Hispanics. (Id. at 196.)

Finally, the Government elicited testimony from Jose Luis Arzola, Jr., a Hispanic man living in Burlington, North Carolina, which lies within Alamance County. Arzola has been stopped by an ACSO deputy approximately three times in the ten years he has lived in the County. (Doc. 147 at 184.) He has received traffic citations, but he "usually" did not get an explanation for the stop reason until he examined the ticket. (Id. at 191.) He did not maintain that any citation was unwarranted. In 2009 or 2010, an unspecified ACSO deputy stopped Arzola while driving on a highway near Green Level. (Id. at 185.) Arzola was not aware that he had violated any traffic law but could not say whether he was speeding. (Id. ) Once Arzola pulled his car to the side of the road, the deputy asked him for his driver's license. (Id. at 185–86.) After that was produced, the deputy asked for his "papers." (Id. ) The deputy never clarified the question, but Arzola, who is legally present in the United States on a Green Card, assumed the question referred to immigration documents.²⁷ (Id. at 191–92.) Arzola responded that he did not have his immigration documents with him, but that they were at his house. (Id. at 187, 191.) Apparently satisfied, the officer let Arzola leave. (Id. at 191.) Arzola "didn't really see it as a big deal" and never filed a complaint with ACSO about the stop. (Id. at 192–93.) In fact, he testified that ACSO and its officers responded twice to his calls of two break-ins at his home and described his treatment as "excellent." (Id. at 194.) He explained that he called ACSO on both occasions because he believed that having them respond would reassure his wife and make her feel safe. (Id. at 194–95.)

27 Under the Fourth Amendment, the Supreme Court appears to allow the inquiry into a stopped individual's immigration status so long as the inquiry does not prolong the stop (and the stop is itself valid). See Arizona v. United States, ––– U.S. ––––, 132 S.Ct. 2492, 2509, 183 L.Ed.2d 351 (2012) ; Muehler v. Mena, 544 U.S. 93, 101, 125 S.Ct. 1465, 161 L.Ed.2d 299 (2005).

b. Statistical Evidence

i. Dr. John Lamberth

The Government's trial evidence of discriminatory targeting of Hispanic drivers was presented through an observational benchmark study involving statistical analysis performed by Dr. John Lamberth. Dr. Lamberth is the head of the Lamberth Consulting firm in West Chester, Pennsylvania. (Doc. 148 at 28.) Lamberth Consulting focuses on assisting police departments to identify potentially discriminatory targeting. (Id. at 31.) The firm's clients span several states and include twenty to twenty-five law enforcement agencies. (Id. at 32.) As part of the firm's work, Dr. Lamberth also serves as an expert witness in litigation. (Id. at 31.)

Prior to starting Lamberth Consulting in 2002, Dr. Lamberth was a tenured professor and chair of the department of psychology at Temple University. (Id. at 29–30.) He holds a master's degree and Ph.D. in social psychology. (Id. at 28.) During his time as a professor, Dr. Lamberth taught a number of psychology courses and published empirical analyses relating to the criminal justice system, specifically concentrating on traffic enforcement and racial or ethnic disparities. (Id. at 30–31.) The Government thus offered Dr. Lamberth as an expert in studying patterns of traffic enforcement. (Id. at 37.) Sheriff Johnson does not dispute Dr. Lamberth's expert qualifications but challenges admission of his opinions under Federal Rule of Evidence 702. (See Doc. 127.) Those challenges will be addressed in the court's conclusions of law.

The Government retained Dr. Lamberth to perform an analysis of ACSO's citation practices to measure for potential patterns of traffic law enforcement.²⁸ (Id.at 38.) Dr. Lamberth began by creating a "benchmark survey." (Id. ) To do so, he attempted to create a profile of drivers—a "benchmark"—by observing all drivers on selected roads in Alamance County, their ethnicity, and whether they were violating a North Carolina traffic law. (Id. at 38, 40.) Here, the benchmark sought to determine how many Hispanic drivers violate a traffic law on specific roads in Alamance County. (Id. at 40.) Dr. Lamberth then compared that benchmark to the actual citations issued by ACSO on those same roads. (Id. at 38–39; see also Gov't Trial Ex. 71 (example citation form).)

28 Dr. Lamberth did not analyze ACSO's traffic-stop database. Although ACSO's traffic stop forms contain information on the ethnicity of each driver stopped as observed by the officer, the stop forms lack information on the location of the stops. (Doc. 148 at 98–99.) Dr. Lamberth testified that, without that location information, he could not perform an accurate comparison between the traffic stops performed by ACSO officers and his observational benchmarks. (Id. ) Dr. Lamberth's testimony, however, did not explain why a citation form, which had the stop location, could not be linked to its corresponding stop form, which contained the ethnicity of the driver. (See id. at 173.)

Dr. Lamberth selected three Alamance County roads for his study: Highways 49, 70, and 87. (Id. at 45.) Highways 49 and 87 are two-lane roads, and Highway 70 is a mix of two-lane and four-lane roadways. (Id. ) The speed on those highways varies from 25 to 50 miles per hour. (Id. ) The highways also cut through both urban and rural areas. (Id. at 45–46.) In 2012, Dr. Lamberth set up a total of 22 sites at some unspecified distance off of those three highways to observe traffic.²⁹ (Id. at 54–55.) Each observation site was at least a quarter of a mile from an intersection, to allow for the identification of vehicles exceeding the speed limit. (Id. at 66.) He also located sites on both sides of the highways. (Id. at 67.)

29 The demonstrative offered by the Government ("D2") maps out a total of twenty-three observation sites.

Dr. Lamberth employed two persons—Richard Rivera and Sabino Valdez—to conduct the traffic survey. The two sat in a vehicle facing oncoming traffic for approximately three hours at a time. (Id. at 64, 67.) The three-hour sessions occurred fifteen times on each of the three highways and occurred between 7:00 a.m. and 1:00 a.m. during the spring and fall of 2012. (Id. at 67–68, 84, 87.) In total, the surveyors spent 135 hours making observations of some fifteen thousand drivers. (Id. at 67, 175.)

Rivera is a former New Jersey state police officer. (Id. at 70, 188.) No information on Valdez's background was provided, and neither surveyor testified at trial. Both are Hispanic and have worked with Dr. Lamberth in prior studies. (Id. at 70–71, 77.) Rivera made the "vast majority" of observations, with Valdez assisting him at some unknown frequency if there was "enough traffic to merit" two observers. (Id. at 61, 175–76.) Otherwise, Valdez's role was limited to recording Rivera's observations. (Id. at 61, 70–71, 175–76.) The surveyors identified drivers who "appeared to be" or "looked" Hispanic. (Id. at 181.) Dr. Lamberth provided no description, criteria, or standard used by the surveyors to identify someone they believed to be "Hispanic" other than "if they thought someone looked Hispanic."³⁰ (Id. )

30 Dr. Lamberth was unable to articulate a standard, definition, or description the surveyors may have used, even upon the court's separate questioning of him. (Doc. 148 at 181.)

In an effort to verify the accuracy of Rivera and Valdez's ethnic observations, Dr. Lamberth claims to have calibrated the surveyors' "inter-rater reliability." (Id. at 56.) Inter-rater reliability testing was presented as a method used in published, peer-reviewed journals to assess the reliability of visual observation. (Id. at 58-59.) The purpose of an inter-rater reliability test is to determine both how the surveyors identify individuals' ethnicities and the reliability of those identifications. (Id. at 56–57.) The test compares the observations of multiple observers to ascertain how often their perceptions agree. (Id. at 56, 61.) Dr. Lamberth prefers that the observers agree on driver ethnicity in at least 80% of the observations. (Id. at 56.) Here, Dr. Lamberth testified, Rivera's and Valdez's "inter-rater reliability" testing produced agreement as to the ethnicity of drivers in 100% of their observations. (Id. at 56–57, 60.) However, the surveyors only made ten test observations, and none included a Hispanic driver. (Id. at 128, 131.)

In addition to observing ethnicity, the surveyors also endeavored to identify those drivers who violated a North Carolina traffic law in any regard. The surveyors attempted to observe every traffic law violation (id. at 72, 179–80), so no distinction appeared to have been made as to the degree of the violation (e.g., more serious speeders were not distinguished from those slightly exceeding the speed limit). Nor was any testing done to confirm the accuracy of the surveyors' abilities to identify traffic law violations, but both surveyors were apparently "familiar with" North Carolina traffic laws, according to Dr. Lamberth. (Id. at 72.) Rivera also used a radar device to identify speeding violations. (Id. )

Using Rivera and Valdez's observations of ethnicity and traffic law violations, Dr. Lamberth purported to create a benchmark for each of the three highways. (Id. at 78.) After initially identifying Hispanic violators, Dr. Lamberth weighted the benchmark based on the incidence of citations at a given location. (Id. at 63–64, 79.) Put differently, Dr. Lamberth weighted the observed proportion of Hispanic violators at a given location by the number of ACSO citations at that location. (Id. at 79–80.) In doing so, he provided no testimony as to how he determined which citations occurred at which survey location. This weighting aimed to account for different rates of violations across the three surveyed highways. (Id. at 79–80.)

Following the observational study and weighting based on survey location, Dr. Lamberth calculated the benchmarks identifying the percentage of Hispanic violators on the surveyed highways. For Highway 49, he concluded, the weighted percentage of Hispanic violators was 4.17%, meaning that, of the total drivers observed on this road, 4.17% were both Hispanic and violated a traffic law. (Id. at 103–04.) On Highway 70, the weighted percentage was 4.34%. (Id. at 107.) Highway 87's weighted percentage was 2.71%. (Id. at 106.)

Dr. Lamberth used these weighted percentages of Hispanic violators as benchmarks to compare against ACSO's actual citation practices. To do this, he gathered all of the actual citations issued as a result of traffic stops on Highways 49, 70, and 87.³¹ (Id. at 38–39, 47.) The citations included the location where the citation was issued, permitting Dr. Lamberth to identify those citations issued on the three highways. (Id. at 89.) Dr. Lamberth identified citations issued on those highways from 2008 through October 2013, which totaled approximately 2,000. (Id. at 48–49, 90.) From that group he then removed those citations issued between the hours of 1:00 a.m. and 7:00 a.m. because his surveyors did not conduct observations during that time. (Id. at 69.)

31 Dr. Lamberth's analysis included non-traffic citations issued as a result of traffic stops. (Doc. 148 at 112.) The Government attempted to introduce additional analysis by Dr. Lamberth in which he removed those citations for non-traffic offenses. (Id. at 113–18.) Sheriff Johnson objected on the ground that the Government failed to amend Dr. Lamberth's expert report to include this additional analysis as required by Federal Rule of Civil Procedure 37(c)(1). (Id. at 113, 116–17.) The Government responded that, under Rule 26(e)(1)(A), no amendment was required. (Id. at 115–17.) The court took the issue under advisement (id. at 118) but need not resolve it because the additional analysis would not change the court's conclusion that the analysis is fundamentally unreliable for other reasons, as provided below.

ACSO's citation database lacks information identifying the ethnicity of the individual cited. (Id. at 90.) To attempt to address this problem, Dr. Lamberth conducted a "surname analysis." (Id. at 90–91.) The basic assumption of surname analysis for this case is that individuals with particular surnames "tend heavily to be Hispanic." (Id. at 90.) Operating on this premise, Dr. Lamberth took the names provided on ACSO citation forms and estimated the probability that those individuals would self-identify as Hispanics. (Id. at 91.) To do this, he used probabilities developed by the U.S. Census Bureau that estimated whether a person with a given surname would self-identify as Hispanic. (Id. at 91–92.) Then, he took each group of names provided in ACSO's citation database that had an associated U.S. Census Bureau probability and multiplied each group of names by its associated probability.³² (Id. at 92.)

32 Dr. Lamberth testified that he also performed a surname analysis in which he estimated the number of Hispanics using U.S. Census Bureau probabilities but using a probability cutoff of 75%. (Doc. 148 at 91, 94.) In other words, if a name on an ACSO citation had an associated probability less than 75%, that individual would not be counted as Hispanic in Dr. Lamberth's study. Dr. Lamberth testified that this alternative calculation did not significantly change his results. (Id. at 94.)

Using this information, Dr. Lamberth calculated estimates of the percentage of citations issued to Hispanics for each highway. (Id. at 94–95.) According to Dr. Lamberth, the surname analysis estimated that 20.77% of ACSO's citations on Highway 49 were issued to Hispanics. (Id. at 104.) For Highway 70, 24.45% of ACSO citations were given to Hispanics. (Id. at 107.) And, for Highway 87, 15.39% of ACSO citations were issued to Hispanics. (Id. at 106.) Dr. Lamberth calculated an error rate for these estimates between 3% and 5%, noting it was "bidirectional," meaning that the estimates may slightly undercount or overcount the number of Hispanics receiving citations. (Id. at 97–98.)

With the benchmarks from the observational study and the estimates from the surname analysis, Dr. Lamberth calculated "odds ratios" for each highway. (Id. at 33.) Dr. Lamberth's odds ratios sought to measure the likelihood that a Hispanic would receive a citation compared to a non-Hispanic. (Id. at 33, 39–40.) The odds ratios reflect Dr. Lamberth's comparison of his benchmarks to his surname analysis estimates of the citations received by Hispanics on the three Alamance County highways. For example, an odds ratio of 2.0 means that the subgroup analyzed is twice as likely to be cited as compared to an individual outside the sub-group. (Id. at 34.)

Here, the odds ratios were 6.0 for Highway 49, 7.13 for Highway 70, and 6.5 for Highway 87. (Id. at 104, 106–07.) Dr. Lamberth testified that each of those odds ratios was statistically significant. (Id. at 104–08.) He also conducted this same analysis for the intersection of Graham–Hopedale Road and Apple Street Extension in Alamance County. (Id. at 50–51, 108.) The resulting odds ratio for that intersection, which he said was statistically significant, was 12.73. (Id. at 108–09.) From these numbers, Dr. Lamberth ultimately opined that ACSO cited Hispanics at a much higher rate than the rate at which they violated traffic laws. (Id. at 39.)

ii. Dr. John MacDonald

The Government also presented expert testimony by Dr. John MacDonald as to ACSO's law enforcement practices after traffic stops. Dr. MacDonald is an associate professor of criminology and sociology at the University of Pennsylvania. (Doc. 149 at 4.) Criminology is "a social science that's focused on studying the causes of crime, the response of crime, and some aspects of [lawmaking]...." (Id. at 6.) Dr. MacDonald holds a bachelor's degree in political science and both a master's degree and Ph.D. in criminology. (Id. at 5.) Prior to joining the faculty at the University of Pennsylvania, Dr. MacDonald was a professor at both the University of South Carolina and the University of Florida and spent time as a behavioral scientist at the Rand Corporation. (Id. ) Since 2006, he has held various positions at the University of Pennsylvania. (Id. at 6.) His work at both the Rand Corporation and the University of Pennsylvania involved statistical analysis and the criminal justice system. (Id. at 5–6.) Much of his scholarly research and peer-reviewed publications also contain statistical analyses on the topics of crime control and responses to crime. (Id. at 7–8.)

Beyond his academic work, Dr. MacDonald has served as a consultant for several law enforcement agencies to analyze their policing activities. (Id. at 8.) His consulting work included research on identifying patterns of racial and ethnic disparities in stops and arrests, police use of force, and the effectiveness of certain law enforcement policies. (Id. at 9.) This research also involved statistical analysis. (Id. ) The Government offered Dr. MacDonald as an expert in criminology and statistical analysis. (Id. at 10.) Sheriff Johnson does not dispute Dr. MacDonald's expert qualifications.

In this case, Dr. MacDonald conducted two studies seeking to identify racial and ethnic disparities in the outcomes of ACSO traffic stops. (Id. at 11–12.) First, Dr. MacDonald examined the following outcomes occurring after ACSO traffic stops: (1) citation; (2) arrest; (3) written warning; (4) verbal warning; and (5) no enforcement action (hereafter "post-stop outcomes"). (Id. ) Second, he performed an analysis of ACSO's searches following a traffic stop and the rate at which those searches yielded illegal contraband (hereinafter "hit rates"). (Id. at 40.)

For both studies, Dr. MacDonald relied exclusively on ACSO's traffic stop data from approximately June 2008 to October 2013.³³ (Id. at 11–15.) For that time period, ACSO's traffic stop database contained information from 20,059 traffic stop forms based on the information provided by the deputies completing them. (Id. at 13–14.) Those stop forms include information on the following: the initial reason for the traffic stop; vehicle driver information (including the driver's race and ethnicity but not name); the enforcement action taken as a result of the stop (specifically, whether an officer issued a citation, made an arrest, issued a verbal or written warning, or made no enforcement action); whether the officer performed a search during the stop; the type of search (i.e., whether the search was based on probable cause, consented to, based on a search warrant, incident to arrest, or a protective frisk); whether a passenger was searched; and whether the officer found "contraband" (e.g., illegal drugs or weapons). (Id. at 14–16; Gov't Trial Ex. 59 (example of ACSO traffic stop form).)

33 Dr. MacDonald did not rely on the citation database that Dr. Lamberth used in his observational study. (Doc. 149 at 18.)

Sheriff Johnson does not dispute the admissibility of Dr. MacDonald's testimony on either study, but he challenges the inferences the Government seeks to draw from it. The court finds Dr. MacDonald's testimony generally credible but, for the reasons noted herein, finds many of the conclusions the Government seeks to draw from it unsupported.

(a) Post–Stop Outcome Study

In his first study examining post-stop outcomes, Dr. MacDonald performed two comparisons. First, he calculated a "raw" percentage for each of the five post-stop outcomes and separated his results between Hispanics and non-Hispanics. (Doc. 149 at 19–20.) He then performed a "controlled comparison" for each of the five post-stop outcomes, comparing Hispanics and non-Hispanics, while controlling for the initial reason for the traffic stop.³⁴ (Id. ) Dr. MacDonald used a logistic regression model, which held differences in stop reason constant between Hispanic and non-Hispanic stops. (Id at 23, 37–38.) He controlled for stop reason because he believed that differences in stop reasons may explain differences in the post-stop outcomes faced by Hispanics and non-Hispanics. (Id. at 23.)

34 ACSO's stop form provides ten reasons for traffic stops: checkpoint, driving while impaired, investigation, other motor vehicle violation, safe movement violation, seat belt violation, speed limit violation, stop light/sign violation, vehicle equipment violation, and vehicle regulatory violation. (Gov't Trial Ex. 59.)

Dr. MacDonald did not, however, control for the reason officers issued citations or made arrests following a stop. That is, his analysis did not attempt to determine whether any observed differences in the arrest and citation records for Hispanics and others could be explained by the reason a person was arrested or cited. (Doc. 155 at 41 (stating that, under his study, "you wouldn't know the type of arrest, charge, or the type of citation").) Dr. MacDonald acknowledged that, by not controlling for the reason for citation and arrest, his post-stop outcome study lacks "a layer of context[ ]." (Id. at 42.)

Dr. MacDonald's comparisons demonstrated statistically significant differences between Hispanics and non-Hispanics for each post-stop outcome. (Doc. 149 at 21–40.) First, comparing the post-stop outcome of citations, Dr. MacDonald found that 55.8% of stopped Hispanics received a citation versus 32% of stopped non-Hispanics. (Id. at 21–22.) After controlling for stop reason, Dr. MacDonald's study showed that a stopped Hispanic is 146% more likely to receive a citation relative to a stopped non-Hispanic. (Id. at 25–26.) Second, as to arrests, the post-stop outcome study revealed that 11.9% of stopped Hispanics were arrested as compared to 6.2% of stopped non-Hispanics. (Id. at 29.) Dr. MacDonald's controlled comparison of arrests indicated that a stopped Hispanic is 52% more likely to be arrested than a stopped non-Hispanic. (Id. at 30.) Third, for the post-stop outcome of written warnings, Dr. MacDonald found that 5% of stopped Hispanics received written warnings relative to 9.5% of stopped non-Hispanics. (Id. at 34–35.) Controlling for stop reason, Dr. MacDonald's comparison indicated that stopped Hispanics were 44% less likely than stopped non-Hispanics to receive a written warning. (Id. at 35–36.) Fourth, for verbal warnings, the post-stop outcome study demonstrated that 22.3% of stopped Hispanics received a verbal warning versus 43.5% of stopped non-Hispanics. (Id. at 32–33.) In his controlled comparison of verbal warnings, Dr. MacDonald's study showed that stopped Hispanics were 55% less likely to receive a verbal warning than stopped non-Hispanics. (Id. at 33.) Fifth, as to the post-stop outcome of no action taken, the post-stop outcome study revealed that 4.8% of stopped Hispanics received no action as compared to 8.6% of stopped non-Hispanics. (Id. at 36–37.) The controlled comparison indicated that stopped Hispanics were 63% less likely than stopped non-Hispanics to receive no law enforcement action. (Id. at 37.)

(b) "Hit–Rate" Study

Dr. MacDonald's second study was of "hit rates," which analyzed the rate at which searches performed subsequent to a traffic stop yielded contraband. (Id. at 40.) This analysis had two major components. First, Dr. MacDonald determined the rate at which stopped Hispanics and stopped non-Hispanics were searched. He concluded that 16% of stopped Hispanics were searched as compared to 12.9% of stopped non-Hispanics. (Id. at 41.) These figures are statistically significant. (Id. )

Second, Dr. MacDonald examined the "hit-rates" for searches performed during those traffic stops. First he limited his inquiry to the discovery of drugs, because drugs were "the most common contraband found." (Id. at 43–44.)³⁵ Controlling for stop reason but not search reason, Dr. MacDonald found that approximately 6% of searches of stopped Hispanics uncovered drugs as compared to 30% of searches of stopped non-Hispanics. (Id. at 44–45.) In other words, searches of stopped Hispanics were 85% (or he said about five times) less likely to uncover drugs relative to stopped non-Hispanics. (Id. at 45.)

35 Dr. MacDonald testified that he examined other contraband and found a "consistent pattern of hit rates being significantly lower for Latinos compared to non-Latinos." (Doc. 149 at 44.)

Because these figures included searches conducted incident to arrest (which are generally non-discretionary searches conducted as a matter of course), Dr. MacDonald testified that he controlled for that as well as for whether a passenger was searched. (Id. at 46–47.) He found that searches of stopped Hispanics were 80% less likely to uncover drugs relative to stopped non-Hispanics. (Id. ) He reported that these figures are statistically significant. (Id. at 46–48.)

In summing up his conclusions, Dr. MacDonald opined that his "hit rate" analyses "suggest that there is a different standard, a lower threshold of suspicion or probable cause [being applied] in searching Latinos compared to non-Latinos." (Id. at 48.)

Finally, Dr. MacDonald measured search hit rates specifically at ACSO checkpoints. In this study, Dr. MacDonald identified the rates at which ACSO checkpoint searches yielded drugs, alcohol, or any contraband. (Id. at 49.) He did not, however, control for whether the checkpoint search was incident to arrest. The drug hit rate for checkpoint searches of stopped Hispanics was 9.89%; the drug hit rate for stopped non-Hispanics was 48.22%. (Id. ) Dr. MacDonald also observed "a significant difference between alcohol found for those who were searched who were Latino versus non-Latino who were searched."³⁶ (Id. ) Dr. MacDonald's study indicated that the overall contraband hit rate for checkpoint searches of stopped Hispanics was 10.99%, while the overall contraband hit rate for stopped non-Hispanics was 56.8%. (Id. at 49–50.) Each of these figures is statistically significant. (Id. ) In summing up his opinion as to searches ACSO conducted at vehicle checkpoints, Dr. MacDonald testified that his results "suggested at checkpoints the searches being conducted on Latinos have a lower threshold for reasonable suspicion or probable cause." (Id. at 50.)

36 Dr. MacDonald did not say in which direction that difference lay or its numerical magnitude. A demonstrative exhibit contained some information, but it was only identified and not moved or admitted into evidence.

As to all results, Dr. MacDonald concluded that they showed "a consistent pattern" across all post-stop outcomes and "on an order of magnitude that's pretty large" compared to what he has seen in other research as well as in his own work. (Id. at 51.)

iii. Officer Mark Dockery and ACSO Data on Searches Not Incident to Arrest

Sheriff Johnson argues that because everyone arrested is automatically searched, the best measure of whether ACSO officers discriminate in searches lies in the evidence related to searches where there is officer discretion—i.e., persons whose searches are not incident to arrest. In this regard and in response to Dr. MacDonald's testimony about discretionary searches, Sheriff Johnson introduced the testimony of ACSO Officer Mark Dockery.³⁷ (Doc. 153 at 31.)

37 Before trial, the Government moved to exclude Officer Dockery on the grounds he was not properly disclosed as an expert witness. (Doc. 116.) The court reserved ruling on the motion, and at the close of evidence the Government conceded that its motion had become moot because Officer Dockery did not provide expert opinion testimony. (Doc. 155 at 44–45.)

Officer Dockery is responsible for ACSO's information technology and is the "system administrator" for ACSO's record management system. (Id. at 32.) His official duties include retrieving data from ACSO's computer database. (Id. at 35–36.) He holds an associate's degree from Rockingham Community College, has taken a course in crime analysis, and has received training from the manufacturer of ACSO's records system. (Id. at 32–33.) That training included work on Crystal Report computer software, which functions to pull data from databases. (Id. )

Using ACSO's traffic stop database, Officer Dockery retrieved data for all searches not incident to arrest³⁸ from December 20, 2008, to October 4, 2013. (Id. at 51–52.) He placed all of ACSO's traffic stop data into a spreadsheet, examined all traffic stops involving searches, and removed those marked as incident to arrest. (Id. at 51–53.) This left only traffic stop searches not incident to arrest, which Officer Dockery organized by the stop reasons provided on ACSO's stop form. (Id. at 53; see also supra note 34.) He then compared the results for Hispanics and non-Hispanics (based on ethnic information in the database) by search reason, conducting a simple mathematic calculation to determine the percentage the results represent based on the overall number of stops. (Doc. 153 at 54, 73.)³⁹ Officer Dockery's calculations, which were not disputed as to their accuracy, were as follows: (Id. at 53–58.)

38 Because most departments have a policy that requires a search of all arrestees, ACSO officer discretion should not come into play for a search incident to arrest; such searches are specifically recorded as a "type of search" option on ACSO's traffic stop form. (Doc. 149 at 63; Gov't Trial Ex. 59.)

39 Although Officer Dockery often referred to these figures as "rates," they are more accurately considered as percentages. See Federal Judicial Center, Reference Manual on Scientific Evidence 294 (3d ed.2011).

STOP REASON	HISPANIC NON– INCIDENT TO ARREST SEARCHES	PERCENTAGE OF HISPANICS SEARCHED	PERCENTAGE OF NON–HISPANICS SEARCHED40
Checkpoint	35	5.18%	22.27%
Driving While Impaired	6	20.00%	19.07%
Investigatory	61	11.42%	16.46%
Regulatory41	149	10.29%	7.97%

Other	20	9.62%	11.28%
Safe Movement	38	13.24%	10.33%
Speeding	30	10.53%	5.53%
Seatbelt Violation	5	14.71%	14.04%
Stop Sign Violation	12	10.17%	7.13%
Vehicle Equipment	21	7.42%	8.08%
Vehicle Registration	23	9.87%	7.68%

Total	251

These percentages demonstrate that in many cases ACSO searched Hispanics at levels comparable to or less than that of non-Hispanics (e.g., at checkpoints, driving while impaired, investigatory, vehicle equipment, seatbelt violations, and "other"). The largest gap was at checkpoint searches, where Hispanics were searched less than non-Hispanics by a factor of more than 4 to 1. The principal areas where ACSO searched more Hispanics were stops involving moving violations such as speeding and safe movement, as well as stop sign and vehicle registration violations.

iv. Dr. David Banks

Sheriff Johnson offered David Banks, Ph.D., as an expert in statistics and who conducted a number of statistical analyses relating to ACSO's law enforcement practices. (Doc. 153 at 90.) The Government challenges Dr. Banks' qualifications and the substance of his testimony. (Doc. 115 (Motion to Exclude Expert Report and Testimony of Dr. Banks); Doc. 153 at 90–91; Doc. 158 at 110 n. 29.) As with the challenge to Dr. Lamberth, those objections will be addressed in the conclusions of law to follow.

Dr. Banks is a Professor of the Practice of Statistics at Duke University in the Department of Statistical Science. (Doc. 153 at 76.) He received his bachelor's degree in mathematics and anthropology from the University of Virginia, master's degrees in applied mathematics and statistics from Virginia Tech, and a Ph.D. in statistics also from Virginia Tech. (Id. ) Dr. Banks accepted a post-doctoral research fellowship with the National Science Foundation. (Id. ) He then taught a number of statistical courses at the University of Cambridge and Carnegie Mellon University. (Id. at 76–77.) For a number of years he was employed by the federal government, working as a mathematical statistician at the National Institute of Standards and Technology, serving as the director of the Office of Advanced Studies and the acting chief statistician at the Department of Transportation, and performing risk analyses for the U.S. Food and Drug Administration. (Id. at 78–80.) While at the Department of Transportation, Dr. Banks reviewed a written report on racial profiling on the New Jersey Turnpike at the request of the Bureau of Justice Statistics. (Id. at 81.) He also worked with the Fatality Analysis Reporting System, which is a dataset collected by law enforcement agencies. (Id. at 81–82.) After his government work, Dr. Banks joined the faculty at Duke University, where he teaches courses on data mining and statistical inference.⁴² (Id. at 80–81.) He is not tenured at Duke but has received offers for tenured positions at Yale University and Penn State University. (Id. at 90.)

42 Dr. Banks described data mining as "a relatively modern area of computer-intensive statistics in which you typically work on very large datasets, and you try to make minimal assumptions on the data—on the model that you have for the data." (Doc. 153 at 80.)

Dr. Banks has written several monographs relating to statistics, edited a number of books connected to statistical sciences, and authored approximately seventy papers on topics including data mining and computer programing. (Id. at 82–85, 89.) He currently serves on the Board of Directors of the American Statistical Association and as the editor of the Journal of the American Statistical Association. (Id. at 86.) Throughout his career, Dr. Banks has supervised people performing surname analyses, has himself published on surname analysis, and serves on the Board of Directors of the Human Rights Data Analysis Group, which uses surname analysis to link records to identify civilian casualties in conflicts. (Id. at 139–40.) He has received numerous honors and awards, including his position as a fellow of the American Statistical Association. (Id. at 86–87.)

Dr. Banks performed analyses of the following for this case: ACSO's location of checkpoints in Alamance County; the proportion of Hispanics stopped at ACSO's checkpoints; arrests of Hispanics at ACSO's checkpoints; the number of traffic stops of Hispanics in Alamance County; and the number of citations of Hispanics in Alamance County. Each will be addressed in turn below.

(a) ACSO's Checkpoint Siting

Dr. Banks first performed a permutation test to analyze ACSO's checkpoint siting. (Id. at 91–92.) A permutation test "considers all possible reassignments of labels to the data and determines whether or not the observed assignment of labels is statistically unlikely compared to random assignment of the labels." (Id. at 92.) As applied to the current case, Dr. Bank's permutation test examined whether ACSO sited checkpoints closer to Hispanic communities than would occur given random chance. (Id. at 92–95.)

To do this, Dr. Banks analyzed a list of 305 checkpoint sites—every location at which ACSO conducted a checkpoint in the past four years.⁴³ (Id. at 92–94.) He then identified the six Hispanic communities that ACSO allegedly targeted through its checkpoint operations. (Id. at 94–96.) For each year from 2009 through 2012, Dr. Banks measured the distance from ACSO's checkpoints to the six Hispanic communities (id. at 93–96) and ran a thousand simulations of ACSO's checkpoint siting (id. at 94). Each simulation randomly selected sites from the list of 305 actual sites and, in doing so, assumed that the list of sites reflected all possible checkpoint sites.⁴⁴ (Id. at 94, 97–98.) Dr. Banks then compared the actual distance of ACSO's checkpoints for each year to the distances reflected in his simulations. (Id. at 93–94.)

43 The total number of checkpoints conducted over the four-year period studied was 435, indicating that some sites were used more than once. (Doc. 153 at 93.) The Government never challenged the total volume of checkpoints as being inappropriate or as any evidence of discriminatory intent or conduct.

44 Dr. Banks acknowledged that the 305 locations "probably [don't] consist of all possible sites." (Doc. 153 at 97.) The permutation test also assumed that the selection of checkpoints was independent, meaning that selection of one checkpoint location did not affect the next location's selection. (Id. at 97, 159–60.)

The results of Dr. Banks' permutation test showed that 40% of the simulations randomly placed checkpoints closer to Hispanic communities than ACSO's actual checkpoint sites. (Id. at 94–95.) Given these results, and by applying commonly-accepted statistical principles, Dr. Banks concluded that "there was absolutely no evidence that checkpoints were being sited closer to Hispanic communities than would have occurred if they were done just at chance." (Id. at 99.) As the Government correctly points out, this conclusion assumes that ACSO's sites reflected all reasonably available sites in the County. While the Government did not provide any credible evidence of other available sites, ACSO's evidence demonstrated that site locations could not be selected arbitrarily, as they must meet several safety and logistical standards. Therefore, the court concludes that ACSO's 305 sites were a reasonable reflection of available sites in the County.

(b) ACSO's Checkpoint Stops

Dr. Banks next examined those who were stopped at ACSO's checkpoints.⁴⁵ He found that 36% of the stops at an ACSO checkpoint involved a Hispanic driver. (Id. at 99.) Dr. Banks offered four reasons why this may be the case: (1) Hispanics may be more likely to commit certain types of offenses or engage in certain types of behavior that could lead to a detainment at a checkpoint; (2) Hispanics may be more likely to drive without licenses; (3) Hispanics may be of lower socioeconomic status and thus not register their vehicles as often or have car insurance; and (4) ACSO may place checkpoints on roads where Hispanics drive. (Id. at 172–93, 201.)

45 As noted earlier, every motorist passes through a checkpoint. Only those who are stopped for further, reportable action become the basis of a stop form.

Dr. Banks offered three studies to support his first explanation. (Id. at 174–76.) The first is a Prince William County, Maryland study that found that "illegal Hispanics were stopped at ... about three times the rate for DUI[s] compared to non-Hispanics." (Id. at 110 (referencing Thomas M. Guterbock et al., Evaluation Study of Prince William County's Illegal Immigration Enforcement Policy (2010)).) The second was a Durham County, North Carolina study—conducted by the National Highway Transportation Safety Administration—that showed that, in Durham, Hispanics comprised 9% of the population but were involved in 29% of traffic accidents. (Id. at 111 (referencing National Highway Transportation Safety Administration, Promising Practices for Addressing Alcohol–Impaired Driving Within Latino Populations: A NHTSA Demonstration Project (2010)).) The third was a Fatality Analysis Reporting System study that found that, among fatal accidents, 46 to 47% of Hispanics were intoxicated as compared to 35% of non-Hispanics. (Id. at 110 (referencing Paul L. Zador et al., Alcohol–Related Relative Risk of Driver Fatalities and Driver Involvement in Fatal Crashes in Relation to Driver Age and Gender: An Update Using 1996 Data, 61 J. Stud. on Alcohol & Drugs 387 (2000)).) Dr. Banks provided no evidence concerning his other justifications. (Id. at 188–92.)

To further explain his initial finding, Dr. Banks performed a study in which he compared the proportion of Hispanics stopped at checkpoints sites (36%) to the 2010 unadjusted Census estimate for Hispanics in Alamance County—11.6%. (Id. at 101, 106.) Dividing the 36% figure by the 11.6% Census estimate, Dr. Banks derived Alamance County's "disparity" ratio of over 3.0. (Id. at 100, 106.) He then repeated this calculation of ratios for ten other (mostly nearby) North Carolina counties and the city of Burlington (which lies within the County). (Id. at 101.) Comparing ratios across these other jurisdictions, Dr. Banks found that, of the North Carolina counties (and city) examined, Alamance County had only the fourth largest ratio of Hispanics stopped at checkpoints to the unadjusted Census percentage estimate of Hispanics.⁴⁶ (Id. at 101–02.)

46 Although not specifically testified to, Dr. Banks' demonstrative at trial showed that Caswell, Chatham, and Forsyth Counties all had higher ratios.

The Government criticizes Dr. Banks' use of unadjusted Census data. Dr. Banks testified that, while he did not know of peer-reviewed studies using unadjusted Census data, he performed a case-control methodology with that data, which is a common methodology in academic literature. (Id. at 152–53, 156–57.) Second, he stated that he did not attempt to adjust the Census data because "there was no set of adjustments that I actually thought I could actually defend." (Id. at 105.) Census adjustments are changes to the U.S. Census' population estimates meant to improve the U.S. Census' estimates' accuracy. (Id. at 102.) Adjustments account for population characteristics like literacy rates and migrant worker patterns. (Id. at 103–04.) Third, Dr. Banks testified that making adjustments of that kind would have been "problematic" in his study without first controlling for whether or not a person operates their vehicle in a safe manner—the dominant factor in his study. (Id. at 143–45.) Because he could not adjust for driving behavior, performing other "minor" adjustments would have been "misinformative." (Id. at 144–45.) Fourth, based on North Carolina's Department of Motor Vehicle data, Dr. Banks found that 9.92% of the commuting miles in Alamance County were driven by Hispanics—a figure he claimed was "consistent" with the unadjusted U.S. Census estimate for the County.⁴⁷ (Id. at 106–07, 205.) Fifth, and most importantly, he explained that his analysis "did not require that the actual census count be correct. It only required that the inaccuracy in the census count for Alamance County be about the same as the inaccuracy of the census count for [the other localities]." (Id. at 105–06, 198.) Dr. Banks noted that, through "pooling information" on Hispanic population figures, the U.S. Census makes the same or similar adjustments to all nearby counties' population estimates. (Id. at 148.) Given that U.S. Census practice, Dr. Banks believed that the lack of adjustments caused nearly identical inaccuracies across counties. (Id. at 198–99.)

47 Dr. Banks provided no testimony as to how he reached this figure, but the Government did not challenge is methodology at trial or in its pre-trial motion to exclude his testimony and expert report. (See Doc. 115.)

The court accepts Dr. Banks' use of Census data for the limited purpose it is offered: to compare Alamance County's ratio to that of other North Carolina jurisdictions. For the reasons cited by the Government, the court does so without accepting that any ratio is itself indicative of a true benchmark analysis. Rather, it is only Dr. Banks' relative comparison of Census data, whose alleged internal errors are consistent across all the examined jurisdictions, that will be considered.

(c) ACSO's Checkpoint Arrests

Dr. Banks next examined ACSO arrests at checkpoints.⁴⁸ To conduct this analysis, he randomly chose a 10% sample of ACSO's 435 checkpoints conducted between 2009 and 2012, selecting 45 different checkpoints. (Id. at 93, 111–12.) He deemed the 10% sample "adequate" because non-Hispanics had "a considerably higher arrest rate than Hispanics," which would require "a remarkable reversal" by the remaining arrest data. (Id. at 113–14.) For those 45 checkpoints, Dr. Banks "hand-matched" traffic stop forms to arrest reports to determine the number of checkpoint arrests and found that 110 stops and 21 arrests occurred there. (Id. at 112, 165–66.) His results showed that Hispanics "were arrested conspicuously less often than non-Hispanics." (Id. at 112.) However, because his confidence intervals for the two groups (Hispanic and non-Hispanics) overlapped, he determined that from a purely statistical viewpoint there cannot be said to be any difference between them. (Id. at 112–13.)

48 This is distinguished from Dr. MacDonald's arrest analysis, which focused on all arrests and not just those at checkpoints.

(d) ACSO's Traffic Stops

Dr. Banks' fourth study examined ACSO's non-checkpoint traffic stops. In this analysis, Dr. Banks compared the proportion of non-checkpoint stops of Hispanics in several North Carolina counties to that county's unadjusted Census estimate of Hispanics; he then divided that by the proportion of stops of non-Hispanics to the proportion of non-Hispanics in that county. (Id. at 114–15.) The product of this analysis was a ratio, from which Dr. Banks subtracted 1.0 to avoid double counting, to show "whether or not there was an excess or a deficiency of Hispanic traffic stops." (Id. ) Dr. Banks found that Alamance County's ratio was negative, meaning that, at non-checkpoint stops, ACSO stopped fewer Hispanics compared to their proportion of the population.⁴⁹ (Id. ) Because this calculation for Alamance County employed the unadjusted Census data as a benchmark, the court does not consider it reliable for the reasons noted earlier. However, Dr. Banks made similar calculations for fourteen other North Carolina counties. (Id. at 115–16.) In comparing the results across the counties, Dr. Banks found that eight of the counties had higher ratios than Alamance County. (Id. ) Specifically, he found that Alamance County compared "extraordinarily favorably" to Orange and Randolph Counties, which he termed "suspicious." (Id. )

49 Sheriff Johnson did not move the admission of Exhibit 79, and Dr. Banks did not testify to any numerical values.

According to Dr. Banks, the results across all jurisdictions he examined showed that "unless many, many other counties in North Carolina are engaged in aggressive ethnic profiling, Alamance County does not seem to be misbehaving" by comparison. (Id. at 117.) Thus, he concluded, his study of traffic stop ratios "shows that there is no statistical evidence of ethnic profiling in [ACSO's] traffic stops." (Id. at 116.) This is an overstatement. While Dr. Banks' analysis in this regard does not prove that no ethnic profiling is occurring in Alamance County, it does suggest that the County compares favorably to other counties about which there is no claim of discriminatory policing.

(e) ACSO's Citations

Finally, Dr. Banks analyzed what he termed a "Hispanic citation ratio." In this study, apparently using unadjusted Census data, he calculated the proportion of Hispanics receiving citations relative to the proportion of non-Hispanics receiving citations for eleven North Carolina jurisdictions.⁵⁰ (Id. at 123.) A ratio of 1.0 would reflect a completely ethnic-neutral citation rate. (Id. ) The ratio for Alamance County was "above one,"⁵¹ which "suggests that Hispanics are receiving citations at a higher rate in the population" than non-Hispanics. (Id. ) By comparison, Alamance County's ratio was less than that of three other counties he compared.⁵² (Id. at 123–24.) Orange County had a ratio "a little bit less than three" and Randolph County's ratio was "extremely elevated."⁵³ (Id. ) Based on this comparison, Dr. Banks opined that Alamance County's ratio was not "suspicious," especially compared to those other three counties with "clearly larger" ratios. (Id. at 123–25.) Again, to the extent Dr. Banks used the unadjusted Census data to compute a figure for Alamance County, it would be an improper benchmark for reaching a conclusion about the County alone; but to the extent he used the unadjusted Census data across multiple jurisdictions, the court will consider that for the limited purpose it is offered.

50 It appears that these were for citations at checkpoints only, but that is not entirely clear. (See Doc. 153 at 168–69.)

51 Sheriff Johnson never moved the admission of Def. Trial Ex. 80, but the court notes the ratio was between one and two.

52 Although not specifically testified to, Dr. Banks' demonstrative at trial showed that he had calculated the ratios for nine other North Carolina counties and the city of Burlington, making Alamance County's ratio larger than six counties.

53 While Randolph County's large ratio would give him "concern," Dr. Banks stated that it would require further analysis to "try and figure out why there was an elevated rate." (Doc. 153 at 125.) Dr. Banks did not argue that this comparison was proof of discrimination or lack of discrimination, apparently conceding that further investigation was necessary.

C. Evidence of Racial and Ethnic Bias Within ACSO

The evidence of ethnic/racial bias regarding Hispanics at ACSO was conflicting. ACSO Sergeant Dan Cubino, a Spanish-speaking Hispanic of Cuban descent who was an eleven-year veteran and worked in the ICE unit from 2007 to 2012, testified that he never knew Sheriff Johnson or any ACSO leadership to engage in, or direct, any discriminatory action against Hispanics. (Doc. 152 at 130–31, 155.) Had he ever observed it, he said, he would not have tolerated it. ACSO Detective Carlos Rossi—a Peruvian immigrant, U.S. Army veteran, and former ACSO patrol officer—echoed Sergeant Cubino's assessment of ACSO. (Doc. 153 at 212–13.) Detective Rossi worked the Hispanic neighborhoods in Alamance County and never observed any discrimination toward Hispanics. (Doc. 154 at 4.) He, too, would have felt obligated, as an immigrant to the United States, to report discrimination had he observed it. (Id. at 4–5.) Both men were highly credible.

In addition, ACSO has engaged in several forms of outreach to the County's growing Hispanic community. For example, ACSO hired Ester Benbassat, an Argentinian resident of the County, to teach Spanish to ACSO officers. (Doc. 153 at 14, 16.) And as previously mentioned, Arzola, a Hispanic resident of the County whom the Government suggests was stopped because of his ethnicity, characterized his treatment by ACSO officers as "excellent" when responding to his calls after home break-ins. (Doc. 147 at 194.) In fact, his wife was reassured by their presence. (Id. at 194–95.) This is contrary to the Government's generalized allegation that Hispanics in the County are distrustful of and fear law enforcement but further supported by Detective Rossi's testimony, admitted without objection, that many Hispanic residents within Alamance County understand the work that ACSO officers do and appreciate their help. (Doc. 154 at 13.) To the extent some Hispanics in the community hesitated to communicate with ACSO, he explained, it was his experience that much of their hesitation is attributable to cultural differences, with many coming from cultures of police distrust. (Id. at 8, 15.) Finally, according to ACSO Director of Personnel Captain Richard Longamore, Jr., who handles most citizen complaints, ACSO has never received a complaint of racial profiling or any type of complaint from an individual speaking Spanish. (Doc. 152 at 196–97.)

ACSO has repeatedly attempted to hire more Hispanic officers. (Doc. 152 at 197, 199–200; Doc. 154 at 5.) ACSO pays an individual who can speak Spanish at the time of hire a 4.5% salary increase. (Doc. 152 at 200.) In 2012, ACSO also approved sponsorship of a Hispanic officer in North Carolina's Basic Law Enforcement Training School, only to learn that the individual had already received sponsorship from the Burlington Police Department. (Id. at 197–99.) In addition to recruiting efforts, ACSO recently began to offer its officers Spanish language classes to improve language skills and better understand Spanish-speaking culture—an idea originating from Sheriff Johnson and Chief Deputy Britt. (Id. at 195; Doc. 153 at 14–16, 217.) These classes have had a positive effect on relations with the Hispanic community. ACSO officers are now better able to communicate with the County's Spanish-speaking community. (Doc. 150 at 160.) Captain Longamore is now conversant in Spanish after taking ACSO's Spanish classes and will often field calls from Spanish-speaking residents. (Doc. 152 at 195–96.)

Sheriff Johnson himself has been personally involved in ACSO's outreach programs. He started "The Sheriff's Christmas Program," which provides gifts to children, many of whom are Hispanic. (Doc. 151 at 54; Doc. 154 at 72.) He also personally delivers these presents to the homes of the children. (Doc. 154 at 72.) ACSO also provides security at soccer fields in the northern part of the County, and many of those at the fields are Hispanic. (Id. at 73.)

Other evidence at trial was troubling, however. Over the course of several years, some ACSO officers have used a number of racial and ethnic epithets, slurs, and jokes derisive of Hispanics, and there have been emails of similar pejorative character. Most of this conduct occurred within the ACDC—not the patrol division—of ACSO.

1. Racial and Ethnic Epithets and Jokes

In its complaint, the Government alleged a number of discriminatory statements made by Sheriff Johnson and ACSO officers. (Doc. 1 ¶¶ 31–33.) For example, the Government claimed that Sheriff Johnson referred to Hispanics as "taco eaters," that he "lamented the increased Latino presence in North Carolina's workforce and public schools," and that ACSO officers threatened residents with deportation on multiple occasions. (Id. ¶¶ 31, 33.) At trial, the Government presented no evidence to support those allegations or several other similar allegations made in its complaint.

There was evidence that some ACSO and federal ICE officers uttered racial and ethnic epithets and jokes. Several ACSO and ICE officers used the following terms to describe Hispanics: "wetback" (Doc. 147 at 159; Doc. 149 at 136, 217; Doc. 150 at 64; Doc. 153 at 6), "spic" (Doc. 147 at 160; Doc. 149 at 220; Doc. 150 at 64; Doc. 153 at 7), "beaner" (Doc. 147 at 160; Doc. 149 at 220; Doc. 150 at 102 (former officer Gerry Helms attributing use of word to Major Miles); Doc. 153 at 7).⁵⁴ Major Miles also heard multiple uses of the term "wets" as a descriptor for Hispanics, but those individuals using the term were mostly federal ICE agents. (Doc. 149 at 218.) Former officer Gerry Helms also heard other ACSO officers use the terms "taco" and "Mexicant," referring to Hispanics. (Doc. 150 at 103–04.) Finally, several officers heard the term "Mexican" to refer to Hispanics generally. (See, e.g., Doc. 147 at 160–61; Doc. 149 at 137.)

54 Two officers said they also heard the term "Tonk." (Doc. 149 at 136 (noting that, while ACSO officers used the term, the epithet originated with an ICE agent), 218 (commenting that this term was used only by ICE agents).) The Government, however, elicited no testimony as to what it contends this meant. (See id. at 144 (Lieutenant Denham testifying that he does not know what the word "Tonk" means).)

Beyond epithets, some ACSO officers also sometimes made jokes and derogatory comments about Hispanics. For example, Officers Anthony and Helms heard other ACSO officers tell racially and ethnically insensitive jokes. (Doc. 148 at 9; Doc. 150 at 104–05.) Former ACSO officer David Cobb recalled "a couple of times" when officers told a detainee to "go back to your country." (Doc. 150 at 64.)

The majority of the epithets, slurs, and jokes occurred within the ACDC. (See Doc. 149 at 224 (Major Miles); Doc. 150 at 67–68 (Cobb), 104 (Helms). But see Doc. 147 at 160 (Officer Lloyd heard patrol officers use epithets); Doc. 149 at 136 (Lieutenant Denham similarly heard patrol officers use derogatory terms).) Such evidence came largely from current and former officers in ACSO's detention division—not those in the traffic enforcement division. (Doc. 149 at 126–27 (Lieutenant Denham spent time both as an officer in the 287(g) program within the detention center and as a detective under Sheriff Johnson), 211 (Major Miles was major over the detention division); Doc. 150 at 57–64 (Cobb spent time as a detention officer (80% of his time in the jail and 20% in the sally port) and court bailiff, as well as spending a few years as a patrol officer), 98–101 (Helms mainly worked as a detention officer or bailiff transporting prisoners).) Only two former officers—Cobb and Helms—testified that other officers actually directed an epithet toward Hispanic individuals or that Hispanic individuals may have heard insensitive terms and jokes. (Doc. 150 at 65–66, 113.) Most accounts were that ACSO officers uttered such epithets outside the presence of a Hispanic person. (See, e.g., Doc. 147 at 160; Doc. 149 at 224.)

2. Derogatory Emails

In addition, the Government introduced evidence that certain ACSO officers sent or forwarded several racially and ethnically derogatory emails. Except for one instance that is specifically noted, none of the individuals involved was ever reprimanded for their involvement in any of the following emails.

On January 25, 2010, Lieutenant Hoover—an officer in the patrol division at the time—forwarded an email titled "Rules for Kicking Ass" to several officers assigned to him. (Doc. 150 at 14; Gov't Trial Ex. 20.) He thought the email was "amusing" and, as a military veteran, respectful to the military. (Doc. 150 at 15, 27.) The email listed a dozen "rules" that are hyperbolic pro-America or pro-military statements. For example, Rule 1 read, "The next time you see any adults talking (or wearing a hat) during the playing of the National Anthem—kick their ass"; Rule 2 read, "When you witness, firsthand, someone burning the American Flag in protest—kick their ass"; and Rule 11 stated, "Last, but not least, whether or not you become a member of the military, support our troops and their families. ... Without them, our Country would get its ass kicked." (Id. at 24–26.) The email also contained a message to "[m]ake sure you read # 12." (Gov't Trial Ex. 20.) Rule # 12 stated, "If you ever see anyone either standing for or singing the national anthem in Spanish—KICK THEIR ASS." (Id. )

On July 12, 2010, Officer Mario Wiley—a supervising ACDC detention officer at the time—emailed a video game to several of his supervisee officers. (Doc. 149 at 234–35, 238; Gov't Trial Ex. 51.) Titled "Border Patrol," the video game was a shooting game with the objective of shooting people running across a river toward a shore bearing a sign reading, "Welcome to the United States." (Doc. 149 at 235; Gov't Trial Ex. 51.) According to Officer Wiley, the people running across the screen were Mexicans. (Doc. 149 at 236.) Although Officer Wiley claims that he and any others playing the game "didn't pay a whole lot of attention of [those] running," this is not credible (id. at 236–37)—the game labelled the runners as "Mexican Nationalists," "drug smugglers," or, in the case of women, offensively as "breeders." (Id. at 239.) After the conclusion of a game, the final screen tells the player how many "wetbacks" were shot. (Id. at 240–41; Gov't Trial Ex. 51.) The use of the game appears limited to certain persons within the ACDC, and there is no evidence that Sheriff Johnson or any senior ACSO officer was aware of the game until it was discovered during Officer Wiley's deposition in this case. When use of the game was discovered, Sheriff Johnson met with Officer Wiley, who agreed to step down several ranks with a corresponding 9% pay reduction. (Doc. 149 at 245–46; Doc. 154 at 61–62.)

On June 1, 2012, Officer Randy Jones—ACSO's part-time public information officer—forwarded an email titled "Be on the lookout for red 1951 Chevy." (Doc. 150 at 54; Gov't Trial Ex. 23.) Officer Jones sent the email to several supervisors—Chief Deputy Britt, Captain Robert Wilborn, and Captain Longamore. (Doc. 150 at 54.) The email read:

The United States Border Patrol is asking citizens to keep on the lookout for a red 1951 Chevy that they suspect is being used to smuggle illegal immigrants across the border from Mexico and into points along the U.S. border. If you see the vehicle pictured below and have reason to believe that it is the suspect vehicle, you are urged to contact your local Police Department or the U.S. Border Patrol.

(Doc. 150 at 54; Gov't Trial Ex. 23.) The punchline of the email is a picture of a red 1951 Chevy outer chassis with scores of legs protruding underneath where wheels would normally be, such that the individuals are using the chassis as a not-so-clever decoy. (Gov't Trial Ex. 23.) Officer Jones intended the email to be a joke and a "good-natured poke" at another law enforcement agency—border patrol. (Doc. 150 at 55–56.)

On July 13, 2012, Officer Jones forwarded another email he had received from a non-ACSO employee titled "Slow Response" with no additional text of his own to several of his supervisors—Chief Deputy Britt, Captain Longamore, Captain Wilborn, and Major Brown. (Doc. 150 at 50–52; Gov't Trial Ex. 22.) The email explains "Sam Elliott's ^[55] Take on Slow Response" and tells of observing two men—a "Muslim extremist" carrying guns and bombs, and a "Mexican" with a "large backpack of drugs ... strapped to his back"—drowning in the Rio Grande River due to their loads. (Gov't Trial Ex. 22.) The text notes that "[b]eing a responsible Texan ... I informed the El Paso County Sheriff's Office and Homeland Security" but that, after several hours of no response, both have drowned and "I'm starting to think I wasted two stamps." (Id. ) Officer Jones said he sent the email because he thought it exemplified the irony in the groups that represented "two of the most dangerous criminal elements in our society": "armed terrorists" and "drug traffickers." (Doc. 150 at 56.) On August 23, 2012, Lieutenant Hoover—now a supervisor in ACSO's Special Services division—forwarded another email titled "Texans" to two subordinates in ACSO's Animal Control division. (Doc. 150 at 16–17; Gov't Trial Ex. 117.) The body of the email begins with the title "The Dallas Solution" and told a story of a homeowner's association embroidering the letters "INS" (ostensibly for "Inner Neighborhood Services" and not "Immigration and Naturalization Service") onto hats. (Doc. 150 at 18; Gov't Trial Ex. 117.) The story continues that members of the homeowners association, upset with construction debris left nearby and wearing the "INS" hats, approached "Mexican construction workers" in hopes of scaring them. (Doc. 150 at 18–21.) Lieutenant Hoover testified that the email was "not about scaring off Mexican workers" but rather about "stopping the trash and cleaning up the trash." (Id. at 21–22.) This explanation is not credible.

55 Sam Elliott is an American actor who plays in westerns.

On January 22, 2013, Captain Longamore forwarded an email from Victor Jeffries—whom ACSO had previously retained to teach Spanish to its officers and court interpreter—to Sheriff Johnson and Chief Deputy Britt. (Doc. 152 at 207, 209.) The forwarded email contained no additional text and was originally titled "Sex slave trafficking by the Family Values folks in the Southeast." (Id. ) The substance of the original email commented on a news article and stated in relevant part:

The clientele were most certainly 99.44% co-ethnic illegal aliens themselves. Do the "gag me with a spoon" fifth grade arithmetic—25 "polvitos" per day times 22 days per month times 12 months per year times 11 female sex slaves equals quite a few undesirable illegal aliens. And this is just ONE small subset of an operation that just happened to be busted by a largely Anglophone police intelligence network. Eliminate the young male illegal aliens and a very substantial portion of their co-ethnic prostitutes and sex slaves will self deport. Unfortunately for the American taxpayer, the 11 female sex slaves in this case (and many others) will get to stay in the U.S. under the "U" Visa program (undocumented victims and witnesses). What will our society be able to do with young Mesoamerican females who at best are semi-literate in any language and only have experience milking cows and working as captive sex slaves.

(Id. at 207–08.) According to Captain Longamore, Jefferies repeatedly sent him emails although Jefferies no longer taught classes for ACSO, and Longamore forwarded the email to Sheriff Johnson and Chief Deputy Britt "for informational purposes." (Id. at 210.) Sheriff Johnson testified that he did not open the email but, when informed about the email, told Chief Deputy Britt to make sure that Jeffries was not associated in any way with ACSO. (Doc. 154 at 155–58.)

On May 1, 2013, Corporal Darryl Meyers—a supervisor in ACSO's detention division—forwarded his supervisor, Lieutenant Wesley Anderson, an email titled "Profiling." (Doc. 150 at 36–37, 43; Gov't Trial Ex. 21.) Corporal Meyers did not add anything to the text of the original email. (Gov't Trial Ex. 21.) The original email began, "THE ORIGIN OF PROFILING." (Id. ) It then described a scene between historical figures Davy Crockett, William B. Travis, and Jim Bowie at the Alamo. (Id. ) As the group looks out over "the hordes of Mexicans moving towards the Alamo," Davy Crockett says to Jim Bowie, "Jim, are we, by any chance, having landscaping done today?" (Id. (quotation marks omitted).) Lieutenant Anderson forwarded the email, without additional text, to Greg Shattery—ACSO's maintenance director. (Doc. 150 at 42.) Corporal Meyers testified that he did not believe the email to be derogatory or a slur. (Id. at 44.) This is not credible, either.

D. ACSO's Supervision and Discipline of Its Officers

In its complaint and post-trial brief, the Government challenged ACSO's supervision and discipline of its officers. (Doc. 1 ¶¶ 52–67; Doc. 158 at 19–22, 67–75.) In addition to presenting testimony from ACSO officers at trial, the Government introduced expert testimony from Margo Frasier on standard law enforcement practices and procedures.⁵⁶ (Doc. 151 at 21.) She specifically provided testimony on (1) whether ACSO had policies and training to limit discriminatory policing, and (2) what measures ACSO took in light of complaints about discriminatory policing. (Id. at 22–23.) To reach her opinion on these issues, Frasier reviewed ACSO's entire policy book, including its harassment policy, read through witness depositions selected in part by the Government, and met with Government attorneys. (Id. at 23, 61, 73.) Frasier also visited Alamance County for about six or seven hours but met exclusively with Government witnesses—several of whom were members of the advocacy group Fairness Alamance—to further inform her opinion. (Id. at 60–61.) During her brief visit, Frasier did not meet with any current ACSO patrol officers or visit ACSO. (Id. at 58–60.) With Frasier's expert testimony and the testimony of ACSO officers, the evidence at trial established the following facts on ACSO's supervision and discipline.

56 Sheriff Johnson originally challenged Frasier's law enforcement qualifications as to her familiarity with North Carolina law. (Doc. 151 at 21–22.) This court allowed Frasier to testify subject to a later motion by Sheriff Johnson. (Id. at 22.) Sheriff Johnson never subsequently objected to Frasier's qualifications.

1. Supervision

ACSO has several policies meant to supervise the conduct of its officers.

a. ACSO's Complaint Policy

ACSO implements a citizen complaint procedure in conjunction with its disciplinary policy. (See Def. Trial Ex. 16.) The purpose of the complaint policy "is to establish guidelines and procedures for receiving, reporting, investigating, and adjudicating allegations and complaints against [ACSO] personnel." (Id. at 1.) ACSO accepts complaints from "any source." (Id. at 2.) Alamance County residents may visit ACSO's website, email, call, or walk into ACSO to report an issue. (Doc. 152 at 187–89; Doc. 154 at 46.) Many residents also contact officers personally, particularly Sheriff Johnson, with issues and complaints. (Doc. 154 at 46.) "Any non-ranking or nonsupervisory member" in ACSO receiving a complaint must report the complaint to his or her supervisor. (Def. Trial Ex. 16 at 2.) All complaints received require completion of a complaint form. (Doc. 154 at 46; Def. Trial Ex. 16 at 2.) All supervisors must report complaints to the chief deputy within twenty-four hours of their receipt. (Def. Trial Ex. 16 at 2.)

The process of investigation for a complaint depends on the type of complaint received. (Doc. 152 at 188; Doc. 154 at 46–47.) If the complaint is "minor," described as "performance issues" such as an officer being rude, the officer's supervisor handles the complaint. (Doc. 152 at 187–88; Doc. 154 at 46–48.) These investigations, known as "Supervisory Investigations," require preliminary investigation and possible discipline, if warranted, by the supervisor. (Doc. 152 at 190–91; Def. Trial Ex. 16 at 2–3.) The chief deputy then reviews the complaint, investigation, and any discipline. (Doc. 152 at 190–91; Def. Trial Ex. 16 at 2–3.) If satisfied, the chief deputy signs off on the complaint, but if not, he returns the complaint to the supervising officer for further investigation. (Doc. 152 at 190–91.)

If, however, the complaint relates to an "ethical issue," it is handled by ACSO's Office of Professional Standards. (Doc. 154 at 47; Def. Trial Ex. 16 at 1–2.) Ethical issues include civil rights violations. (Def. Trial Ex. 16 at 1–2.) Under this type of investigation, an internal investigative officer conducts the investigation and reports directly to the chief deputy. (Doc. 152 at 191; Doc. 154 at 46–47; Def. Trial Ex. 16 at 3.) If the complaint involved potential criminal exposure, ACSO also contacts the District Attorney and SBI. (Doc. 154 at 47.) After investigation, the internal investigation officer classifies the complaint as "unfounded," "unresolved," "exonerated," or "sustained." (Def. Trial Ex. 16 at 6.) Sustained complaints or allegations are handled under ACSO's "Rules of Conduct/Disciplinary Procedures" policy. (Id. ) ACSO retains all complaints, even those found to be unfounded or minor. (Doc. 152 at 191; Doc. 154 at 48.)

Around 2009, ACSO installed car video cameras in all marked front-line cars. (Doc. 154 at 44–45.) According to Chief Deputy Britt, the cameras sometimes aid in the investigation of complaints and add "another level of supervision." (Id. at 45.) Chief Deputy Britt testified that he would review video records when relevant to a complaint he reviewed. (Id. ) Sheriff Johnson sometimes reviewed video records as well. (Id. at 91.) ACSO, however, had no policy regarding the review of video records. (Id. at 50.)

b. ACSO's Review of Stops, Arrest, and Searches

Outside of the complaint process, supervisors provide some review of traffic stops, arrests, and searches. For arrests, an officer fills out an arrest report, and his supervisor "checks off" on the report. (Doc. 150 at 139–40.) After the supervisor provides this review, the report goes directly to ACSO's records clerk or its warrant division clerk. (Id. ) Captain Wilson—head of ACSO's patrol division—performs no review of arrests for probable cause or review for patterns of racial or ethnic arrest profiling. (Id. at 147.)

For traffic stops, since at least 2009, Captain Wilson collects daily traffic stop forms filled out by ACSO officers. (Id. at 141–42.) She then compares those reports to an Alamance County Central Communications Department report (a computer-aided dispatch or "CAD report"). (Id. ) The CAD report logs all stops radioed over the County's communication system. (Id. ) If a stop appears in the CAD report that does not have a corresponding stop form, Captain Wilson contacts the officer who made the stop to complete a stop form. (Id. at 142.) This type of reminder by Captain Wilson occurs approximately one to three times per week. (Id. ) This process is designed to ensure that all stops are memorialized by a stop form, which requires the recording of information, including ethnicity. Captain Wilson, however, conducts no "substantive review" of the traffic stop forms to ensure that sufficient legal justification existed for the stops. (Id. at 143.) Her review of CAD reports has never identified an unconstitutional stop. (Id. at 144–45.) Similarly, Captain Wilson conducts no substantive review of searches following traffic stops and has never found a search to lack probable cause. (Id. at 145.) Captain Wilson has never been asked to conduct a review for potential racial or ethnic profiling. (Id. at 150.)

Captain Wilson's superior, Major Brown, similarly conducted no systematic check of ACSO traffic stop forms, but he occasionally looked through traffic stop forms and arrest reports. (Id. at 175–78.) A lieutenant under Captain Wilson in the patrol division, Gerry Williams, testified that he also did not review stop forms for adequate legal justification. (Id. at 130.) Rather, he relied on Captain Wilson to do so. (Id. ) Captain Wilson has returned officers' incomplete stop forms to him but never because of a deficiency in a stop's legal justification. (Id. at 131.)

Several officers also relied on the Alamance County District Attorney's office and magistrates to review the legal justification for ACSO's stops, arrest, and searches. For example, Captain Wilson testified that "the magistrate's the one that's going to find probable cause on your arrest. If she has—she or he has a problem, then they would speak of it." (Id. at 156.) Sheriff Johnson similarly believed that challenges to the legal justification for a stop should be left to "the discretion of the officer and the discretion of the magistrate or the [District Attorney's] office or the judge." (Doc. 154 at 83.) Neither the Government nor Sheriff Johnson presented evidence that the Alamance County District Attorney's Office or any magistrate or judge ever found an unconstitutional stop, arrest, or search by ACSO. Similarly, the Alamance County District Attorney—Patrick Nadolski—and Magistrate Wortinger testified that they had never found a stop, arrest, or search lacking in the requisite legal justification. (Doc. 149 at 169; Doc. 152 at 93–94.) Sheriff Johnson also met regularly with District Attorney Nadolski and never received a report regarding an unconstitutional stop, arrest, or search conducted by an ACSO officer. (Doc. 152 at 88; Doc. 154 at 82.)

Finally, ACSO conducts performance evaluations of its officers. From about 2004 to 2009, ACSO stopped conducting formal reviews because they were no longer required by the County. (Doc. 154 at 153.) ACSO did conduct informal performance reviews during that time period, however. (Id. at 93.) Starting in 2009, ACSO—of its own accord—reinstituted formal annual performance reviews of its officers. (Id. at 151, 153–54.) Neither party has submitted evidence on the substance of these performance reviews.

According to Frasier, standard review of traffic stops, citations, arrests, and searches includes the review of the makeup of those stopped, cited, arrested, or searched. (Doc. 151 at 25–26, 79–80.) That review should also include evaluation of the standards applied by officers in making arrests, citations, stops, and searches. (Id. at 27, 79–80.)

c. ACSO's Traffic Stop Data Reporting

ACSO submits monthly reports of its traffic stop data, including race and ethnicity information, to the North Carolina SBI.⁵⁷ (Doc. 150 at 148.) Captain Wilson is responsible for submitting this data. (Id. ) For several months from 2009 to 2012, in addition to sending data to the SBI, Captain Wilson also sent a summary report of traffic stops to Sheriff Johnson and Chief Deputy Britt. (Id. at 148–49.) Her additional report organized the data sent to the SBI into percentage figures. (Id. at 148, 162–63.) However, neither Sheriff Johnson nor Chief Deputy Britt followed up with her on either the original data or Captain Wilson's additional reports. (Id. at 149.) It turns out that the data Captain Wilson sent to Sheriff Johnson and Chief Deputy Britt was the same as the data sent to the SBI, so her report was duplicative. (Id. at 158.) Eventually, Chief Deputy Britt, through Major Holland, determined that the additional report was not needed, and Captain Wilson stopped sending it. (Id. at 149–50.) All ACSO data sent to the SBI is posted by the North Carolina Attorney General on its website, along with all state law enforcement agencies' traffic stop data. (Doc. 154 at 21.) In addition, ACSO recently began posting its traffic stop data on its own website. (Id. )

57 As noted earlier, because of software changes, ACSO underreported its traffic stop data to the SBI for a time. (Doc. 154 at 20.) Once made aware of the problem, Chief Deputy Britt reported the problem to the SBI. (Id. ) To remedy the problem, ACSO implemented new software and created additional administrative review to ensure that it reported all officer stops. (Id. )

d. ACSO's Training of Its Officers

ACSO officers undergo extensive training, which includes education as to making legal stops, arrests, and searches, as well as minority sensitivity. All ACSO officers must undergo some seven hundred hours of training at Basic Law Enforcement Training. (Id. at 82.) ACSO also assigns new officers a training officer for a period of time ranging from six months to a year. (Id. ) North Carolina law further requires that officers attend twenty-four hours of training annually. (Doc. 152 at 177; Doc. 154 at 66.) From 2009 to 2012, that mandatory training included a two-hour course for "juvenile and minority sensitivity training," known as "JMST." (Doc. 152 at 177–78.) ACSO required its officers to attend that two-hour training session. (Def. Trial Exs. 21–24.) In 2012, ACSO officers also received State-mandated training in "traffic interdiction." (Doc. 152 at 182; Def. Trial Ex. 24.)

In addition to the State-mandated training, ACSO implemented significant supplemental training under Sheriff Johnson. Prior to Sheriff Johnson's tenure, ACSO officers received little training and instruction. (Doc. 152 at 186–87; Doc. 154 at 65–66.) After taking office, Sheriff Johnson secured a $50,000 training budget to supplement the training required by the State. (Doc. 154 at 66.) From 2009 to 2012, ACSO officers attended over 450 officer training schools throughout the country. (Def. Trial Ex. 14.) Hundreds of ACSO's officers received this supplemental training and spent a total of 50,849 hours in training during that time. (Id. ) Many of these supplemental training hours educated officers on the necessary legal standards for conducting searches, making stops, and performing arrests. (Doc. 154 at 83–84; Def. Trial Exs. 21–24.)

As an example of this training, Lieutenant Williams—a patrol officer—provided a nuanced description of the substance of his drug interdiction training. Lieutenant Williams testified that, using approximately twenty "indicators" learned through training, officers could more effectively make stops likely to undercover drug trafficking and other drug-related crimes. (Doc. 150 at 123–24, 132.) He made clear that "being Latino" is not an indicator. (Id. at 123–24.) If the indicators suggested possible drug trafficking, Lieutenant Williams would then look for reasonable suspicion to stop the vehicle. (Id. at 132.) He acknowledged that the indicators alone "did not give [him] the right to stop the vehicle." (Id. at 132–33.) Lieutenant Williams explained that he drew on this drug interdiction training when making stops. (Id. at 123.)

e. ACSO's Harassment Prevention Policy

ACSO's policy manual also contains a "Harassment in the Workplace" policy. (See Gov't Trial Ex. 138.) The policy states that ACSO "will not condone or tolerate, in any way, harassment of any type in the workplace." (Id. at 1.) In particular, the harassment policy bans "[h]arassment of any person because of gender, race, color, age, religion, disability, ancestry or national origin" whether that harassment is "directed at an employee, vendor, or customer." (Id. at 3–4.) The policy goes on to list examples of prohibited conduct, including "racial and ethnic slurs or offensive stereotypes and making jokes about these characteristics." (Id. at 4.) The policy requires that a violation of the policy "be reported promptly" unless the victim chooses to handle the situation on his or her own. (Id. at 5.)

f. Implementation of Email and Video Filtering Software

Since the start of the Government's pre-trial investigation in this case, ACSO began employing computer software known as "Ironport" and "Blue Coat" to monitor and filter ACSO emails. (Doc. 154 at 28–29.) The software is designed to curb inappropriate conduct on ACSO's computers and in its emails. (Id. ) The Blue Coat software blocks access to internet sites and certain information on the internet. (Id. ) Both programs now block internet games from loading and block the download of games to an ACSO computer without administrator authorization. (Id. )

2. Discipline

While ACSO has a disciplinary policy, that policy was not introduced into evidence. (See Def. Trial Ex. 16 at 6 (referring to ACSO's "Rules of Conduct/Disciplinary Procedures" policy).) ACSO has available four basic disciplinary violation levels—labelled Classes A through D. (Doc. 152 at 194.) The classes of disciplinary violations correspond with disciplinary action as follows:

A Class D violation would be basically a verbal reprimand, or it could be a plan of corrective action. A Class C reprimand would be a written reprimand, which it could be a written reprimand that would go in their personnel file. A Class B violation could be one to two days off, suspension without pay; and a Class A violation could be a combination of both, up to three days off without pay, including demotion or termination.

(Id. ) Sheriff Johnson, Chief Deputy Britt, or both sign off on disciplinary action. (Id. at 194; Doc. 154 at 48.) Use of racially or ethnically degrading terms or the sending of emails of a similar character are Class A violations, thus permitting a disciplinary action of up to three days suspension without pay, demotion, or termination. (Doc. 154 at 48, 60, 71.) Sheriff Johnson also listed five basic rules that he personally requires officers to follow. (Id. at 74–75.) Those rules prohibit "[l]ying, laziness, incompetence, running around with a married man or a married woman, and drinking in the county and showing yourself." (Id. )

Specific contours of ACSO's disciplinary policy aside, the evidence demonstrated that ACSO inconsistently applied it. In a number of circumstances, ACSO disciplined its officers for wrongdoings, which included racially and ethnically inappropriate epithets, jokes, statements, and emails. Prior to trial but after the Government began its investigation, ACSO demoted the Director of Detention—Coley Rich—to Captain and reduced his pay following an investigation in the ACDC conducted by Chief Deputy Britt. (Id. at 30–31, 62–63.) ACSO suspended Officer Anthony for failing to report another officer's racial slur. (Doc. 147 at 201; Doc. 154 at 78.) The officer using the racial slur received discipline for his statement as well. (Doc. 154 at 78–79, 143–44.) Sheriff Johnson denied a promotion to Lieutenant Anderson for his use of a racially insensitive joke. (Doc. 150 at 44–46.) ACSO demoted Corporal Nicholson for sexual harassment. (Doc. 147 at 92–93.) ACSO also took disciplinary action against several other officers for, among other reasons, misleading an assistant district attorney, insubordination, and assault and detention of a juvenile. (See Doc. 147 at 162 (Officer Lloyd); Doc. 150 at 69 (Cobb), 105, 111 (Helms); Doc. 154 at 41–44 (Officer Lloyd); see also Doc. 154 at 48 (noting discipline occurring because of citizen complaints).)

The Government and Sheriff Johnson also produced testimony on the discipline of Officer Wiley, who sent the "Quick Draw" video game. Because discovery of the video game occurred only after Officer Wiley's deposition in this case while litigation was pending, Sheriff Johnson deferred discipline until after the case is resolved. (Doc. 154 at 61–62.) In the meantime, Officer Wiley was invited to voluntarily step down three levels in rank, resulting in a pay decrease, which he accepted. (Doc. 149 at 246–47; Doc. 154 at 62.)

In a number of other instances, however, ACSO failed to discipline officers making racially and ethnically inappropriate statements, which departs from its own harassment policy. Multiple officers testified that they never observed or meted out discipline for the use of derogatory terms for Hispanics. (See, e.g., Doc. 149 at 136–37 (Lieutenant Denham stated that he has never disciplined officers for using derogatory language.); id. at 223–24 (Major Miles testified that he does not know of an officer receiving discipline for derogatory language and has not disciplined any officers for such conduct.); Doc. 150 at 66–67 (Cobb testified that he did not know of any ACSO officer receiving discipline for the use of derogatory statements.); id. at 104 (Helms stated he knew of no officer receiving discipline for making derogatory statements.).) Similarly, several officers testified that they did not receive discipline for the sending or forwarding of racially and ethnically inappropriate emails. (See, e.g., Doc. 150 at 15–16, 21 (Lieutenant Hoover never faced disciplined for forwarding the "Rules for Kicking Ass" email or the "Texans" email.); id. at 36 (Corporal Meyers was never disciplined for forwarding the derogatory "Profiling" email.); id. at 42–44 (Lieutenant Anderson did not discipline Meyers for forwarding the "Profiling" email and instead forwarded the email to another ACSO employee.).)

3. Margo Frasier

In her expert opinion, Frasier concluded that ACSO (1) lacked adequate practices and procedures to detect discriminatory policing; (2) failed to discipline its employees; and (3) failed to respond properly to complaints about discriminatory policing.

First, Frasier testified that ACSO's current practices and procedures insufficiently monitored for discriminatory policing. (Doc. 151 at 24–25.) She particularly criticized the practice of relying exclusively on a magistrate or district attorney review to detect discriminatory policing. (Id. at 27–28.) According to Frasier, magistrate and district attorney review insufficiently scrutinizes law enforcement practices.⁵⁸ (Id. at 28–29.) Frasier also critiqued Captain Wilson's matching of CAD reports to traffic stop forms and ACSO's method of data reporting as inadequate. (Id. at 32–33.) While recognizing that ACSO reported all of its data to the SBI, Frasier opined that Captain Wilson and ACSO should have conducted some sort of analysis of the data. (Id. at 32–33, 98.) Frasier, however, offered no testimony as to what ACSO should have done in its review or what other law enforcement agencies do with their stop data. (Id. at 27 (describing the analysis as "burrowing down and seeing whether or not you're treating people differently").) Finally, Frasier stated that ACSO should not leave its stop, arrest, and search practices to officer discretion. (Id. at 38–40.) Instead, Frasier opined that ACSO should have implemented systematic review of officer stops, citations, and arrests. (Id. at 25–27.) For Frasier, that systematic review should occur regardless of whether the law enforcement agency believes discriminatory policing is, in fact, occurring. (Id. at 31.) For review of traffic stops, Frasier believed that a law enforcement agency should have "a policy on traffic stops, as far as the facts of how they're to be conducted." (Id. at 34.) As to an arrest policy, she testified that arrests should require supervisor approval. (Id. at 34–35.) Frasier offered no opinion on a search policy that would adequately curb profiling. She further suggested that, to find "the general premise of policies and what sort of policies you would have, ... you can find a template for policies on the web." (Id. at 69.)

58 Frasier testified that magistrates and district attorneys "often never" report discriminatory policing. (Doc. 151 at 29.) This opinion is contrary to the testimony of District Attorney Nadolski (called by Sheriff Johnson), who testified he would report discriminatory policing if he observed it. (Doc. 152 at 93–94.) Magistrate Wortinger was never asked whether she would report discriminatory policing.

On cross-examination, Frasier could point to no specific procedure that ACSO should use to review stops, arrests, and searches. (Id. at 79–80.) She could also offer no opinion on the extent to which other sheriffs' offices in North Carolina review for potential citation patterns. (Id. ) And, although critical of Captain Wilson's and ACSO's stop data collection and reporting, Frasier opined, "[T]hose [procedures] are all good things." (Id. at 80–81.) She further characterized ACSO's installation of video cameras in its vehicles as "fortunate." (Id. at 79.) Frasier also appeared unaware of Sheriff Johnson's weekly meetings with District Attorney Nadolski, but agreed that, if they did meet, that would be "a good thing." (Id. at 86; see also Doc. 152 at 88 (District Attorney Nadolski acknowledged working with Sheriff Johnson "often."); Doc. 154 at 82 (Sheriff Johnson stating he met with District Attorney Nadolski on a weekly basis).) Lastly, she did not say whether the training ACSO officers received from the State of North Carolina satisfied her concerns or not, and it is unclear whether Frasier was aware of ACSO's training.

As to ACSO's disciplinary policies, Frasier testified that ACSO lacked adequate measures. (Doc. 151 at 44–46.) She opined that a law enforcement agency should have a "zero-tolerance" policy toward derogatory statements, jokes, and emails and should require training for those violating that policy. (Id. at 40, 42.) She advocated that those at the top of an organization should vocally condemn any violation when it occurs. (Id. at 42.) In Frasier's view, the lack of both this policy and consequences for policy violations would "create an atmosphere where that sort of behavior is tolerated." (Id. at 41, 45–46.) She acknowledged, however, that these types of comments and jokes continue to occur in law enforcement agencies, although much less today than when she started forty years ago. (Id. at 41–42.)

While Frasier testified that "nothing happened" as a result of the emails and improper slurs made within ACSO, she was unaware of ACSO's adoption of email filtering software, the demotion of Captain Rich, and the discipline of Officer Anthony. When asked about ACSO's adoption of email filtering software, she opined that this was an "appropriate response." (Id. at 91.) She also admitted to being unaware of Sheriff Johnson's discipline of Officer Anthony for failing to report a racial slur but was "glad to hear it." (Id. at 90.) Frasier also allowed that Sheriff Johnson may face limitations on his ability to discipline during the course of the present litigation. (Id. at 92–93.)

Finally, Frasier critiqued ACSO's response to complaints about law enforcement profiling. She stated that a law enforcement agency faced with complaints should not ignore them but rather address them. (Id. at 25.) More specifically, she testified that a law enforcement agency should "strengthen" its policy. (Id. at 37–38.) Frasier urged a law enforcement agency receiving complaints to "sit down with critics" and have a "conversation" with them. (Id. at 51–52.) Without this reaction, Frasier opined, the community would come to distrust law enforcement. (Id. at 55–56.) In her opinion, ACSO failed to make adequate efforts at community outreach. (Id. at 54.)

Frasier, however, again appeared to be unaware of certain efforts undertaken by ACSO. Of particular note, she acknowledged she was unaware "that the [S]heriff met with [Professor] Roselle and Fairness Alamance" over their complaint. (Id. at 63.) She allowed, though, that if Sheriff Johnson met with Professor Roselle and Fairness Alamance, then she "applaud[ed] him." (Id. ) As noted previously, Sheriff Johnson did in fact met with Professor Roselle and Fairness Alamance on more than one occasion. Frasier also acknowledged that Sheriff Johnson investigated Professor Roselle's complaint regarding the underreporting of stop data and corrected an underreporting problem in the computer data. (Id. at 66–67.) Frasier approved of Sheriff Johnson's action as "the sort of thing that you should do ... when somebody raises a legitimate concern." (Id. ) Although appearing unaware of ACSO's provision of security at soccer fields frequently attended by Hispanics, Frasier commended this and other community outreach efforts as "all good things" and "positive." (Id. at 53–54, 88.)

In short, Frasier is a credible witness who appears not to have been provided complete information about ACSO. Thus, her opinions have limited value because they were based on a partial review of the evidence and lack of awareness of key facts. Further, her opinion appears to have been informed in part by Professor Roselle, who was less than objective in her concerns about ACSO.

II. CONCLUSIONS OF LAW

Based on the above facts and any further facts noted hereafter, the court makes the following conclusions of law.

A. Motion to Exclude Dr. Lamberth's Testimony

As a preliminary matter, the court must resolve Sheriff Johnson's motion to exclude the testimony of Dr. Lamberth.⁵⁹ (Docs. 127–28, 142–44.)

59 Sheriff Johnson also moved to exclude Dr. Lamberth's expert reports which, although hearsay, were never offered at trial.

Sheriff Johnson does not challenge Dr. Lamberth's qualifications, only the admissibility of his testimony. Specifically, the Sheriff challenges Dr. Lamberth's absence of a standard for identifying Hispanics; his method of calculating a benchmark for comparison; the lack of a known error rate; the use of citation rather than traffic stop data; the use of selected observation time periods; and the non-random selection of roads. (See Doc. 128.) In a hearing on August 8, 2014, this court reserved ruling on Sheriff Johnson's motion, holding the motion in abeyance until trial. (Doc. 156 at 2.) Sheriff Johnson renewed his motion at trial during Dr. Lamberth's testimony. (Doc. 148 at 36.) The court deferred decision until after Dr. Lamberth's direct and cross-examination. (Id. at 36–37.) Counsel elected to withhold argument on the motion until the close of the Government's case. (Id. at 189–90.) After hearing argument on the motion following the close of the Government's case, the court reserved ruling. (Doc. 151 at 230–31; Doc. 152 at 4–14, 57.) With the entirety of Dr. Lamberth's testimony now before it and after careful consideration, this court concludes that the testimony is unreliable as applied to this case and will grant Sheriff Johnson's motion to exclude it.

1. Admissibility

Federal Rule of Evidence 702 governs the admissibility of expert testimony. This rule requires that trial judges "ensure that any and all scientific testimony or evidence admitted is not only relevant, but reliable." Daubert v. Merrell Dow Pharm., Inc., 509 U.S. 579, 589, 113 S.Ct. 2786, 125 L.Ed.2d 469 (1993) ; see also Kumho Tire Co. v. Carmichael, 526 U.S. 137, 147, 119 S.Ct. 1167, 143 L.Ed.2d 238 (1999). "A reliable expert opinion must be based on scientific, technical, or other specialized knowledge and not on belief or speculation, and inferences must be derived using scientific or other valid methods." Cooper v. Smith & Nephew, Inc., 259 F.3d 194, 200 (4th Cir.2001) (quoting Oglesby v. Gen. Motors Corp., 190 F.3d 244, 250 (4th Cir.1999) ). To guide this "gatekeeping" function, the Supreme Court has identified several non-exhaustive factors useful for evaluating the reliability of proposed expert testimony, which include:

(1) whether a theory or technique can be or has been tested; (2) whether it has been subjected to peer review and publication; (3) whether a technique has a high known or potential rate of error and whether there are standards controlling its operation; and (4) whether the theory or technique enjoys general acceptance within a relevant scientific community.

Id. at 199 (citing Daubert, 509 U.S. at 592–94, 113 S.Ct. 2786 ). The Government must establish the admissibility of Dr. Lamberth's testimony by "a preponderance of proof." Id.

a. Testability

"Ordinarily, a key question to be answered in determining whether a theory or technique is scientific knowledge that will assist the trier of fact will be whether it can be (and has been) tested." Daubert, 509 U.S. at 593, 113 S.Ct. 2786. In this case, the most evident problem with Dr. Lamberth's testimony is that his methodology is seriously flawed and cannot be replicated. Although citing prior studies by Dr. Lamberth and others, the Government has not shown that the methods used in Dr. Lamberth's study have been independently tested. See Ruffin v. Shaw Indus., Inc., 149 F.3d 294, 297–99 (4th Cir.1998) (finding no evidence that expert witness's methodology had been independently replicated "under the same conditions as used in" the expert's study); Newman v. Motorola, Inc., 218 F.Supp.2d 769, 777–83 (D.Md.2002) (excluding experts' testimony in part because their studies had not been replicated). More specifically, the studies relied on by the Government as proof of reliability do not involve observational studies of Hispanics, like the one performed by Dr. Lamberth. See Fabien Jobard et al., Measuring Appearance–Based Discrimination: An Analysis of Identity Checks in Paris, 67 Population 349, 358 (English ed.2012) (observational study not involving Hispanics and identifying individuals while walking at railway stations); Joseph B. Kadane & John Lamberth, Are Blacks Egregious Speeding Violators at Extraordinary Rates in New Jersey?, 8 Law, Probability & Risk 139, 142–44 (2009) (mobile observational study comparing African–American drivers, not Hispanic drivers, to non-African-American drivers); Katherine Beckett et al., Race, Drugs, and Policing: Understanding Disparities in Drug Delivery Arrests, 44 Criminology 105, 110–15 (2006) (observational study involving Hispanics but only involving individuals either self-reporting their ethnicity or observed while standing in "outdoor drug market"). The cases cited by the Government to suggest the testability of Dr. Lamberth's study suffer from similar shortcomings. See Chavez v. Illinois State Police, 251 F.3d 612, 644 (7th Cir.2001) (citing article discussing Dr. Lamberth's work observing African–American drivers); Md. State Conf. of NAACP Branches v. Md. Dep't of State Police, 72 F.Supp.2d 560, 566 (D.Md.1999) (referencing "statistics" without any mention of how those statistics were calculated or what they measured and applying no evidentiary standard to those statistics); Commonwealth v. Lora, 451 Mass. 425, 886 N.E.2d 688, 700–02 (2008) (mentioning Dr. Lamberth's study observing African–American drivers); State v. Ballard, 331 N.J.Super. 529, 752 A.2d 735, 742–45 (N.J.Super.Ct.App.Div.2000) (accepting, on a request for discovery and not on the merits, an observational study by Dr. Lamberth involving Hispanics but offering no information on the study's features, making it difficult to derive much persuasive value from the decision); State v. Soto, 324 N.J.Super. 66, 734 A.2d 350, 352–57 (N.J.Super.Ct. Law Div.1996) (analyzing Dr. Lamberth's observational benchmark work regarding African–American, but not Hispanic, drivers).

Not only has Dr. Lamberth's methodology in this case not been independently tested under the same or similar conditions, but, more importantly, it cannot be. See Daubert, 509 U.S. at 593, 113 S.Ct. 2786. Critical to Dr. Lamberth's study was his identification of Hispanics to set a "benchmark" by which to judge ACSO's data. Yet, Dr. Lamberth conceded that no control, standard, or description was used to identify Hispanics in his study. (Doc. 148 at 181.) Instead, Rivera—the surveyor observing virtually all of the drivers—simply identified people as Hispanic if he thought they "appeared to be" Hispanic. (Id. ) Dr. Lamberth offered no information on what, if any, standard Rivera used.⁶⁰ Dr. Lamberth's study thus relies entirely on the subjective views of Rivera and Valdez and their personal, totally subjective say-so of who should be considered "Hispanic." As a result, it not only cannot be determined who was considered "Hispanic," but the method used cannot be independently tested. These flaws render the study unreliable. See Brown v. Burlington N. Santa Fe Ry. Co., 765 F.3d 765, 772–73 (7th Cir.2014) (upholding district court's exclusion of expert testimony because that testimony relied on "subjective experience instead of a proper scientific approach"); Durkin v. Equifax Check Servs., Inc., 406 F.3d 410, 420–22 (7th Cir.2005) (excluding expert testimony as "untestable say-so"); R.F.M.A.S., Inc. v. So, 748 F.Supp.2d 244, 282–83 (S.D.N.Y.2010) (excluding expert testimony that was "little more than conclusory say-so").

60 When asked for his own definition of Hispanic, Dr. Lamberth responded, "[T]he Census Bureau basically indicates that people who self-identify as Hispanics are Hispanics." (Doc. 148 at 129–30.) Yet, as he admits, it is virtually impossible to track down those in the observational study to have those individuals self-identify as Hispanic. (Id. at 59 (stating that he "[a]lmost never" has the opportunity to ask drivers their race or ethnicity).)

The Government attempts to downplay this failing by pointing to Dr. Lamberth's opinion that the errors in his benchmark would have to be wrong by "about 400 percent or more" to destroy the statistical significance of his conclusions. (Doc. 148 at 109–10.) In other words, he urges, he would have to be wrong in about three out of every four cases. (Id. ) As noted in the discussion of "error rates," infra, the problem with this argument is that there is no reliable way of knowing just how wrong his benchmark may be. See BASF Corp. v. Sublime Restorations, Inc., 880 F.Supp.2d 205, 212–14 (D.Mass.2012) (holding that expert's subjective "eyeballing" produced "an unknown error rate" and lacked reliability); KW Plastics v. U.S. Can Co., 131 F.Supp.2d 1289, 1294 (M.D.Ala.2001) (concluding that "an unknown error rate" rendered an expert's methodology unreliable). Thus, the court declines to accept this beguiling invitation to ignore the fundamental flaw in Dr. Lamberth's methodology.

b. Peer Review

Similarly, Dr. Lamberth has not demonstrated adequate peer review and publication supporting the application of his methodology in this case. See Daubert, 509 U.S. at 592–94, 113 S.Ct. 2786.

The Government contends that Dr. Lamberth's methodology has been used and cited in other studies (several of which include Dr. Lamberth as an author). But, none of those studies attempted to observe Hispanics in passing vehicles from the roadside, as in this case. Rather, most involved comparing African–Americans to non-African-Americans, or utilized more reliable methods of observation. See Jobard et al., supra, at 358 (observational study not involving Hispanics and identifying individuals while walking at railway stations); Kadane & Lamberth, supra, at 142–44 (mobile observational study comparing African–American drivers, not Hispanic drivers, to non-African-American drivers); Beckett et al., supra, at 110–15 (observational study involving Hispanics but only involving individuals either self-reporting their ethnicity or observed while standing in "outdoor drug market"); see also David A. Harris, The Stories, the Statistics, and the Law: Why "Driving While Black" Matters, 84 Minn. L.Rev. 265, 277–88 (1999) (referring to Dr. Lamberth's mobile observational study involving African–Americans).⁶¹ Those articles, therefore, offer little support for the reliability of the stationary observation of Hispanics from the roadside during the day and night undertaken in this case.

61 Both the Government and Sheriff Johnson cite and quote favorably from Lorie Fridell, By the Numbers: A Guide for Analyzing Race Data from Vehicle Stops (2004). None of the expert witnesses testified that the book was authoritative, as required by Federal Rule of Evidence 803(18). In fact, Dr. Banks stated that the book was not authoritative, and Dr. Lamberth conceded the same, even though he authored one of the appendices. (Doc. 148 at 185; Doc. 153 at 121–22.)

c. Error Rates

The known or potential rate of error of a technique is a factor that bears on its reliability and thus the proffered conclusions. See Daubert, 509 U.S. at 592–94, 113 S.Ct. 2786. Here, Dr. Lamberth's methodology contains several unknown rates of error without any reliable method to control for them.

Most obvious, the rate at which the surveyors misidentify Hispanic drivers is unknown, as the surveyors used no controlling standard. Dr. Lamberth admitted there was no control, template, or description used to identify Hispanics in his study. (Doc. 148 at 181.) Instead, surveyors simply identified people as Hispanic if either (almost always Rivera) thought the driver "appeared to be" Hispanic based on their subjective evaluation. (Id. )

Dr. Lamberth's attempted use of an "inter-rater reliability" test in an effort to control for the unknown error rate provides no assistance. As described earlier, this test, when conducted according to design, independently compares the observations of multiple observers to ascertain how often they agree on an individual's race or ethnicity.⁶² (Doc. 148 at 56–57, 61.) The more the observers agree, the more "reliable" their reported observations. In this case, while Dr. Lamberth reports that Rivera and Valdez claim to have reached "100% agreement" during their inter-rater reliability test (id. at 60), this conclusion is essentially meaningless. Once probed, their test group was revealed to be only ten drivers—a statistically unreliable sample size.⁶³ (Id. at 128.) And, even further, none of those drivers was identified as Hispanic. (Id. at 131.) The test therefore provides no information as to whether the surveyors would agree as to the identification of a single Hispanic driver.⁶⁴ In short, the error rate of Hispanic driver identification in Dr. Lamberth's study is completely unknown because the surveyors followed no objective standard. See EEOC v. Kaplan Higher Educ. Corp., 748 F.3d 749, 753 (6th Cir.2014) (finding no abuse of discretion in district court's exclusion, as unreliable, of expert who relied on "raters" who "had no particular standard in classifying each applicant" into one of five races and who "instead, ... just eyeballed ... DMV photos"); Valente v. Textron, Inc., 931 F.Supp.2d 409, 425–26 (E.D.N.Y.2013) (finding expert testimony unreliable because "a different expert [could not] verify [the testifying expert's] results"), aff'd, 559 Fed.Appx. 11 (2d Cir.2014).

62 The court notes, but does not rely on, the fact that By the Numbers, to which Dr. Lamberth contributed, suggests that observational benchmarking should use "two or three observers," with the observers' reported findings tested for inter-rater reliability. Fridell, supra note 61, at 174. Whatever the interpretation of this standard, the court simply observes that Dr. Lamberth essentially used only one observer, Rivera, and one recorder, Valdez.

63 From a statistical point of view, Dr. Lamberth failed to ensure accurate identification because of his test's lack of reliable sensitivity and specificity. Sensitivity is "the true positive rate" of a test (i.e., the rate at which the surveyors correctly identified Hispanics), and specificity is "the true negative rate" (i.e., the rate of correctly identifying non-Hispanics). (Doc. 153 at 118–19.) Dr. Lamberth's test produced unreliable information as to the surveyors' sensitivity and specificity by using only ten drivers. (Id. at 118–20.) According to Dr. Banks, the test would have needed at least two hundred non-Hispanics to ensure accurate and reliable identification. (Id. at 119–20.)

64 Of course, even if there was agreement, Dr. Lamberth's test provides no information about whether any identified driver was actually Hispanic. But because ACSO could have stopped drivers whom they thought looked Hispanic but were not, and because practical difficulties may exist in making that determination, the court does not consider this additional problem.

As noted, the Government nevertheless contends that Dr. Lamberth's team "would have to be wrong by over 400 percent" of the time or, put differently, would have to have "misidentified 3 out of every 4 Latino drivers" for the error rate to impact the study's statistical significance. (Doc. 158 at 51.) The Government further relies on Dr. Lamberth's previous studies reporting high levels of inter-rater reliability. (Id. at 44.) Neither of these arguments is persuasive.

As to the Government's first argument regarding the degree of error, there is no proof that Dr. Lamberth's team could agree on an assessment of any Hispanic driver. Yet, it is the Government's burden to prove admissibility. See Cooper, 259 F.3d at 199. Given the lack of any proof as to Rivera's error rate, the Government cannot rely on the degree of error necessary to impact statistical significance or prior studies to attempt to shore it up. On this record, an unknown figure raised to any power remains unknown.

The Government's reliance on past studies is equally unconvincing. The Government offers no evidence as to how Dr. Lamberth conducted past inter-rater reliability tests, rendering them no more trustworthy than the one conducted here involving only ten drivers. Moreover, even assuming Rivera's inter-rater agreement was reliable in previous studies, past reliability provides no more than the possibility that Rivera is reliable in this case. That is not the standard for determining testability and rate of error.

Further, the surveyors' error rate in identifying North Carolina law violations is completely unknown.⁶⁵ Dr. Lamberth testified that Rivera is a former New Jersey police officer but that he was familiar with North Carolina's traffic laws. (Doc. 148 at 72, 188.) No information was provided as to Valdez. According to Dr. Lamberth, Rivera was looking for all violations of Chapter 20 of the North Carolina General Statutes. (Id. at 179.) It is questionable whether Rivera—as a former New Jersey police officer—or Valdez could in fact know and identify all North Carolina traffic violations. But even if they could, it is doubtful that they would have had proper opportunity to identify many traffic violations while seated on the side of the road facing oncoming traffic. (Id. at 64.) Identifying certain traffic violations, such as a failure to display a registration plate, N.C. Gen.Stat. § 20–63(b), and a failure to equip a vehicle with rear lamps in good working order, N.C. Gen.Stat. § 20–129(d), would require observation of a vehicle as it passes and drives away, which would then limit a surveyor's ability to identify ethnicity and traffic violations for other oncoming vehicles. See United States v. Mesa–Roche, 288 F.Supp.2d 1172, 1190 (D.Kan.2003) (noting "an obvious concern" with Dr. Lamberth's study "is the accuracy of the spotters' recorded data, in light of the short period of time they have to record this information as a car passes them by"). The Government bears the burden of demonstrating the admissibility of Dr. Lamberth's study and has failed to persuade the court in demonstrating the surveyors' ability to identify North Carolina traffic violations. See Cooper, 259 F.3d at 199.

65 An additional unknown error rate is the error rate in Dr. Lamberth's Hispanic surname analysis. Drs. Lamberth and Banks offered competing estimates of the error rate for this surname analysis, 3% to 5% versus 10% to 20%, respectively. (See Doc. 148 at 95–97; Doc. 153 at 140–41.) Because of the problems already identified, the court need not address this issue.

d. General Acceptance

This court also questions whether the observational methodology Dr. Lamberth used in this case is generally accepted in the relevant scientific field. See Daubert, 509 U.S. at 592–94, 113 S.Ct. 2786. The Government offers Dr. Lamberth as an expert in studying patterns of traffic enforcement. (Doc. 148 at 37.) The Government, however, offers little to demonstrate the general acceptance of Dr. Lamberth's observational benchmarking techniques used to measure traffic enforcement patterns as applied to Hispanics.

Although never testifying that his methodology used in this case is generally-accepted within any scientific field, Dr. Lamberth testified that other researchers have used a similar benchmarking methodology in North Carolina. (Id. at 169.) He also opined that certain aspects of his study, such as the inter-rater reliability test and surname analysis, have been accepted. (Id. at 58, 91.)

Critically, though, the Government offers nothing to show that Dr. Lamberth's roadside observational method for identifying Hispanic violators is generally accepted in a scientific field. As noted earlier, the articles the Government relies on to demonstrate the reliability of Dr. Lamberth's study do not involve Hispanics or the same observational techniques conducted here.

In fact, one of the articles that Dr. Lamberth and the Government cite directly undermines the methodology Dr. Lamberth used in this case. Both Dr. Lamberth (id. at 169) and the Government (Doc. 158 at 40) point to William R. Smith et al., The North Carolina Highway Traffic Study (Jan.2004) ( "Smith article"), as supportive of Dr. Lamberth's observational benchmarking. That article actually supports, and applied, a different observational method, known as the "carousel method." Smith, supra, at 6, 251, 254. That method is one "in which a research vehicle is driven at the speed limit, and vehicles that pass the research vehicle can be examined from the vantage point of the research vehicle, wherein the researchers can identify the race and other demographic features of the driver." Id. at 254. This was the methodology used by Dr. Lamberth in a prior case. See Soto, 734 A.2d at 352–57.

When discussing the roadside observation methodology used in this case, however, the Smith article reaches a different conclusion, undermining Dr. Lamberth's opinions:

One might think that data collection would be as easy as standing by the side of the road and observing. Armed with a radar gun and a pad of paper, researchers seemingly could record the race, as well as perhaps the gender, age, and speed of drivers as they drive by, and then would be able to compare such data with the rates of stops and citations for a given highway. Researchers might even be able to observe vehicles weaving unsafely, or vehicles with expired license plate "stickers," or perhaps identify drivers they thought were "driving while intoxicated." One could conceivably estimate speed of the vehicles on a video tape by measuring the time during which a known distance is traveled.

These first thoughts on how to collect data on the behavior of drivers, however, have several problems that make them impractical. One problem is that the glare on windshields and side windows of vehicles makes it difficult to see clearly the motorist's race, as well as his or her gender or age, from a safe distance at the side of the road. The glare is in part due to the tinting done on most, if not all, windshields and to the angle of light from the sky (even on cloudy days). Video cameras suffer from the same problem. The skeptic is encouraged to simply try to stand (from a safe distance) near a highway with fast-moving traffic, and see how often they can successfully identify the demographic characteristics of drivers. We experimented with this method and found that we were frequently unable to do so. Compounding this difficulty is the high rate of speed of the passing vehicles, as there is little time to assess demographic characteristics, much less their speed. More importantly, when we tried this technique (roadside viewing), it did not seem likely that our failure to identify demographic characteristics was random (else one could argue that misses were "random error" and could be safely ignored). Rather, some types of vehicles or conditions, such as an open side window, seem to permit greater visibility than others. In short, we suspect that it is simply too difficult to reliably ascertain demographic characteristics from the sides of busy roads (also, it is rather unsafe for the observer, unless he or she is well removed from the highway, but then it is even harder to see into passing vehicles).

Id. at 252–53 (emphasis added). While certainly not dispositive on the issue of general acceptance, this article highlights the lack of support for the particular method of observation used in this case.

Finally, although the Government cites case law accepting Dr. Lamberth's work, a more extensive search reveals quite the opposite, with many decisions voicing concerns similar to those in the Smith article. As noted earlier, this court has identified four cases supporting some form of Dr. Lamberth's observational methodology, but each case either did not involve Hispanics or failed to provide sufficient information to assess the relevant studies' comparability to this case.

Several decisions analyzing the exact methodology performed by Dr. Lamberth in this case, however, have questioned that methodology. For example, in United States v. Alcaraz–Arellano, 302 F.Supp.2d 1217 (D.Kan.2004), aff'd, 441 F.3d 1252 (10th Cir.2006), the court found Dr. Lamberth's study to be unreliable, noting:

[Dr. Lamberth's] study relies upon subjective and swift, if not split-second observations of Dr. Lamberth's surveyors for its collection of benchmark data regarding the race and ethnicity of drivers. In most cases, one person looked at a car and wrote down the race he or she believed the driver to be, and that conclusion served as the ultimate benchmark data. Yet one person's perception of what another's race is may be vastly different than someone else's perception of the same individual.

Id. at 1229–30 ; see also id. (noting further that "a better method of gathering benchmark data is to have two or three surveyors look at the same car and compare results to measure the extent to which surveyors uniformly perceived race"). Another decision observed, "[G]iven that race is merely a social construct, the Lamberth study ... ha[s] not resolved the inherent difficulties in identifying the race or ethnicity of a particular person." United States v. Parada, 289 F.Supp.2d 1291, 1305–06 (D.Kan.2003). Two other decisions found Dr. Lamberth's methodology "suspect," noting "flaws" such as the "short period of time" surveyors had to record information as vehicles passed. United States v. Duque–Nava, 315 F.Supp.2d 1144, 1158 (D.Kan.2004) ; Mesa–Roche, 288 F.Supp.2d at 1190.

Most surprising, in these cases critiquing Dr. Lamberth's methodology, is the revelation that the Government was the party opposing his methodology as unreliable. See Alcaraz–Arellano, 302 F.Supp.2d at 1230 (agreeing with Government's expert that it would be "difficult" to conduct Lamberth's observational methodology); Mesa–Roche, 288 F.Supp.2d at 1190 (rejecting "government's criticism that the entire Lamberth study should be disregarded because it rests on an invalid presumption").

To summarize, the Government points to only one study and one case accepting observations of Hispanics and no studies or cases approving of the same observational methodology used in this litigation. This showing is far from demonstrating general acceptance of the methodology Dr. Lamberth employed in this case. See Cooper, 259 F.3d at 199.

Considering all factors related to Dr. Lamberth's testimony, the court is not persuaded that Dr. Lamberth's opinions are sufficiently reliable and valid to be admissible under Rule 702 of the Federal Rules of Evidence. Dr. Lamberth's observations and analysis may be helpful on a consulting basis to encourage a law enforcement agency to further examine its practices, but they simply fail to be reliable and to produce valid results to be admissible in this case. Therefore, Sheriff Johnson's motion to exclude the testimony of Dr. Lamberth will be granted.⁶⁶2. Credibility

66 The exclusion of Dr. Lamberth's study in this case should not be read to suggest that observational studies of Hispanics could never be admissible in federal court. Here, the multitude of errors, when combined, simply render Dr. Lamberth's study unreliable and thus inadmissible.

Even if Dr. Lamberth's study were admissible, this court would not accept his conclusions as credible for several reasons.

First, Dr. Lamberth's testimony directly contradicts his prior writings. At trial, he testified that the most accurate benchmark for determining disparities in traffic enforcement was law violators rather than all traffic. (Doc. 148 at 170.) Yet, in the book By the Numbers, Dr. Lamberth authored an appendix offering the exact opposite conclusion. Lamberth et al., "Making the Case for Measuring 'Who Is Driving' Instead of 'Who Is Violating,' " in Lorie Fridell, By the Numbers: A Guide for Analyzing Race Data from Vehicle Stops 411 (2004). In that appendix, Dr. Lamberth concludes, "We believe that it is not necessary or possible to develop a benchmark that adequately measures the factors that influence a police officer to stop a particular vehicle." Id. at 416 (emphasis added); see also id. at 415 ("[C]ategorizing hundreds of possible violations is an insurmountable task."). Dr. Lamberth reached this conclusion, in part, by noting the "important argument ... that there are literally hundreds of traffic violations for which a motorist can be legally stopped." Id. at 414–15 (observing the "subjective" nature of measuring traffic violations).

Dr. Lamberth's methodology and testimony further conflicts with his prior opinions in the By the Numbers appendix on the effectiveness of stationary observation sites. Here, Dr. Lamberth used stationary observational sites to measure traffic violations, and he opined that those sites allowed for "a very complete set of" traffic violation observations. (Doc. 148 at 64–68.) In his prior work, however, Dr. Lamberth unequivocally states,

[D]etecting the vast majority of traffic violations for which a motorist can be stopped from a stationary point or even from a moving vehicle is either not possible or prohibitively time consuming. That is, while many of the violations are always present (equipment violations), they may not be obvious until the vehicle is observed from several angles. Stationary observations do not allow the necessary views.

Lamberth, supra, at 415 (emphasis added); see also Kadane & Lamberth, supra, at 143 ("Observing the race/ethnicity of a motorist from a vehicle is easier to do than from a stationary point on the roadway because the observer is closer to the observed individual ... and has more time to make the observation.").⁶⁷

67 Dr. Lamberth also contradicted his prior writings concerning the relevant data to compare with an observational benchmark. He designed his methodology in this case to compare his observed benchmarks to citations issued by ACSO. (Doc. 148 at 38–39.) He previously wrote, however, that traffic stop data, rather than citation data, is "the more reliable measure of officer behavior." Lamberth, supra, at 416.

In addition to the contradictions in the appendix he authored in By the Numbers, Dr. Lamberth also offered differing opinions on surveyors' ability to identify Hispanic drivers. At trial, Dr. Lamberth downplayed difficulties in identifying Hispanic drivers. (Doc. 148 at 127.) Counsel for Sheriff Johnson, however, impeached this testimony by citing Dr. Lamberth's prior work, in which he states "it is easier to identify black motorists visually than it is for Hispanic motorists." (Id. ) Moreover, Frasier, a former sheriff and the Government's expert in law enforcement practices and procedures, confirmed the unreliability of Dr. Lamberth's methods in the context of this case:

Q: How do you recognize an Hispanic?

A: I don't think that you can assume that you can recognize an Hispanic.

(Doc. 151 at 92.)

In sum, Dr. Lamberth repeatedly contradicts himself and abandons previously-held (and commonsensical) views to bolster the methodology he used in this litigation. This type of self-serving testimony seriously undermines his credibility and leads the court to reject his expert testimony.

B. Motion to Exclude Dr. Banks' Testimony

The Government moves to exclude the testimony of Dr. Banks.⁶⁸ (Doc. 115.) Sheriff Johnson responded to the motion (Doc. 133), and the Government replied (Doc. 141). The Government challenges both Dr. Banks' qualifications and the admissibility of his testimony. (Doc. 115 at 3–21; Doc. 153 at 90–91; Doc. 158 at 110–15.) Both challenges will be addressed in turn.

68 The Government also moved to exclude Dr. Banks' expert reports, but Sheriff Johnson never offered them at trial.

1. Dr. Banks' Qualifications

The Government contests Dr. Banks' qualifications to critique the work of Drs. Lamberth and MacDonald. (Doc. 115 at 16–21; Doc. 153 at 90–91.) A witness "may testify in the form of an opinion" when that witness "is qualified as an expert by knowledge, skill, experience, training, or education." Fed.R.Evid. 702. "[A] witness may be qualified as an expert on any one of the five listed grounds." Friendship Heights Assocs. v. Vlastimil Koubek, A.I.A., 785 F.2d 1154, 1159 (4th Cir.1986). According to the Fourth Circuit,

Where the expert's qualifications are challenged, the test for exclusion is a strict one, and the purported expert must have neither satisfactory knowledge, skill, experience, training nor education on the issue for which the opinion is proffered. One knowledgeable about a particular subject need not be precisely informed about all details of the issues raised in order to offer an opinion.

Kopf v. Skyrm, 993 F.2d 374, 377 (4th Cir.1993) (quoting Thomas J. Kline, Inc. v. Lorillard, Inc., 878 F.2d 791, 799 (4th Cir.1989) ); see also Martin v. Fleissner GmbH, 741 F.2d 61, 64 (4th Cir.1984) (holding that a "lack of direct experience is not a sufficient basis to reject [a proposed expert's] testimony, but may affect the weight that testimony is given, a decision properly made by the [fact finder]"). "The witness' qualifications to render an expert opinion are also liberally judged by Rule 702." Kopf, 993 F.2d at 377.

First, the Government contends that Dr. Banks lacks the expertise to testify on Dr. Lamberth's use of surname analysis and observational benchmarking. (Doc. 115 at 16–18.) The court finds Dr. Banks qualified to testify on those subjects. Throughout his career, Dr. Banks has supervised individuals performing surname analyses, and he has also published on surname analysis. (Doc. 153 at 139.) Dr. Banks further sits on the Human Rights Data Analysis Group, which utilizes surname analysis when performing records linkage to identify civilian casualties in areas of conflict. (Id. at 140.) As to observational benchmarking, while acting as the chief statistician at the U.S. Department of Transportation, Dr. Banks analyzed a written report on racial profiling on the New Jersey Turnpike. (Id. at 81.) At trial, Dr. Banks also demonstrated his familiarity with the observational method employed by Dr. Lamberth. When asked about his experience with observational studies, Dr. Banks cited several papers and a book on the subject. (Id. at 135.) In sum, Dr. Banks is qualified to analyze Dr. Lamberth's surname analysis and observational study.

Second, the Government argues that Dr. Banks lacks the qualifications necessary to critique Dr. MacDonald's statistical analysis because Dr. Banks is not a criminologist. (Doc. 115 at 18–21; Doc. 153 at 90–91.) The Government's argument is unpersuasive. Without belaboring the obvious, Dr. Banks is a well-qualified statistician. During his career as a statistician, he has worked with the Fatality Analysis Reporting System—a law enforcement dataset. (Doc. 153 at 81–82.) At trial, Dr. Banks provided statistical and mathematical opinions on Dr. MacDonald's study. He offered critiques of Dr. MacDonald's studies, including the use of a linear regression analysis (id. at 128–29); the methods by which Dr. MacDonald reached his statistical conclusions regarding hit rates (id. at 125–26); Dr. MacDonald's performance of a specific statistical test (id. at 130–31); and the use of variable controls in Dr. MacDonald's post-stop outcome analyses (id. at 129–30). Dr. Banks is undoubtedly qualified to offer those opinions on the statistical methods of the Government's experts.

2. The Admissibility of Dr. Banks' Testimony

The Government also contends that Dr. Banks' testimony on ACSO's traffic stops, checkpoint stops, checkpoint arrests, and checkpoint placement is inadmissible. (Doc. 115 at 3–16; Doc. 158 at 110–15.)

As noted earlier, Rule 702 imposes a "gatekeeping obligation" on trial judges to "ensure that any and all scientific testimony or evidence admitted is not only relevant, but reliable." Kumho Tire Co., 526 U.S. at 147, 119 S.Ct. 1167 (quoting Daubert, 509 U.S. at 589, 113 S.Ct. 2786 ). The same factors addressed in connection with the analysis of Dr. Lamberth's testimony apply here. As the proponent of the evidence, Sheriff Johnson bears the burden of establishing the admissibility of Dr. Banks' expert testimony by "a preponderance of proof." See Cooper, 259 F.3d at 199. Applying these factors and for the reasons noted below, the court finds Dr. Banks' expert testimony admissible.

a. Dr. Banks' Study of ACSO Traffic Stops

First, the Government argues that Dr. Banks' analysis of ACSO's traffic stops lacks reliability. The contention is based on a twofold argument: (1) Dr. Banks used unadjusted Census data rather than "risk adjusted" data as a benchmark; and (2) Dr. Banks failed to account for county differences when making comparisons across counties. (Doc. 115 at 3–10; Doc. 158 at 112–13.) Neither argument is persuasive, and the court finds the analysis admissible.

As to Dr. Banks' use of unadjusted Census data, the Government's reliability contention misunderstands the purpose for which Sheriff Johnson offered Dr. Banks' traffic stop study as well as the study's use of Census data. In his traffic stop analysis, Dr. Banks compared, across several North Carolina counties, the ratios of traffic stops in each county with that county's unadjusted Census data as to Hispanics and non-Hispanics.⁶⁹ His use of unadjusted Census data in this fashion was not to determine a benchmark by which to assess whether ACSO disparately stopped Hispanics. Cf. Chavez, 251 F.3d at 640–45 (examining unadjusted Census data as an actual benchmark to prove discriminatory effect of policing under the Fourteenth Amendment). Rather, he used the unadjusted Census data as a baseline denominator in a case-control study—a common methodology in academic literature—to compare the relative difference between each jurisdiction. (Doc. 153 at 152–57.) His testimony was offered to cast doubt on the Government's allegations of discrimination by noting similarities and differences in stop rates across counties.

69 As noted earlier in the court's findings of fact, Census adjustments are changes to the U.S. Census' population estimates meant to improve the accuracy of U.S. Census estimates. (Doc. 153 at 102.) Adjustments account for population characteristics like literacy rates and migrant worker patterns. (Id. at 103–04.) "Unadjusted" Census data is a set of population estimates without any adjustments.

Second, the Government charges that Dr. Banks failed to "account for any differences among the selected counties to ensure their comparability to Alamance." (Doc. 158 at 113.) Dr. Banks acknowledged that unadjusted Census data somewhat imprecisely gauged population totals. (Doc. 153 at 105.) However, he explained that his analysis "did not require that the actual census count be correct. It only required that the inaccuracy in the census count for Alamance County be about the same as the inaccuracy of the census count for [the other counties]." (Id. at 105–06, 198.) Based on his experience working with U.S. Census data and its adjustments, Dr. Banks testified that he was aware that, through "pooling information" on Hispanic population figures, the U.S. Census makes the same or similar adjustments to all population estimates in nearby counties. (Id. at 148.) Consequently, any lack of adjustment caused nearly identical inaccuracies across the counties in his study. (Id. at 148, 198–99.) Further, he testified, there was no set of adjustments a statistician could make to the Census data. (Id. at 105.) The dominant factor in his analysis is whether a motorist operated his vehicle in a safe way, yet there was no way to adjust for driving behavior. (Id. at 143–45.) So, performing other "minor" adjustments would have been "misinformative." (Id. at 143–45 (noting that "it is perilous to adjust for second-order effects when you have first-order effects that are outstanding").)⁷⁰ Thus, because the unadjusted Census data was not used here as a benchmark but was rather kept constant across all other jurisdictions compared, the court accepts its use for that purpose.

70 Dr. Banks found that 9.92% of the commuting miles in Alamance County were driven by Hispanics—a figure "consistent" with the unadjusted population estimate for the County that he used. (Id. at 106–07, 205.) Of course, it is unlikely that all Hispanics are of driving age.

b. Dr. Banks' Analysis of ACSO Checkpoint Stops

The Government next maintains that Dr. Banks' study of ACSO's checkpoint stops suffers from reliability issues. Here, too, the Government challenges Dr. Banks' use of unadjusted Census data to make comparisons across counties. For the sake of brevity, the court will not repeat its previous analysis and, for the reasons stated above, finds the Government's challenge equally unpersuasive.

The Government also challenges Dr. Banks' explanations for why his finding that 36% of those stopped at ACSO checkpoints were Hispanic (id. at 99, 106) is not necessarily evidence of discrimination. (Doc. 115 at 10–12; Doc. 158 at 110–12.) In addressing the 36% figure, Dr. Banks cited literature supporting one of his possible explanations. (Doc. 153 at 99, 110–11.) Generally, the Government argues that two of Dr. Banks' proffered explanations lack evidentiary support in the record. The Government's argument, however, is directed toward the weight and persuasiveness of Dr. Banks' explanations rather than their admissibility. See In re Prempro Prods. Liab. Litig., No. 4:03CV1507–WRW, 2008 WL 3285183, at *1 (E.D.Ark. Aug. 7, 2008) (observing that expert's reliance on certain studies goes to the weight of the evidence rather than its admissibility); Atkinson Warehousing & Distrib., Inc. v. Ecolab, Inc., 99 F.Supp.2d 665, 670 (D.Md.2000) ("If the jury finds that the facts upon which a particular expert relied are not sufficient to support the opinion or that the facts relied upon are erroneous, the jury may reject the opinion. If the jury finds that the reasons supporting the opinion of a particular expert are sound and that the facts relied upon do support such opinion, then the jury may give weight to the opinion and consider it in reaching its verdict."); Kenneda v. United States, 815 F.Supp. 926, 936 (S.D.W.Va.1993) ("Any weakness in the factual underpinnings of [the expert's] opinion go to the weight and credibility of his testimony, not to its admissibility." (quoting S. Cent. Petroleum, Inc. v. Long Bros. Oil Co., 974 F.2d 1015, 1019 (8th Cir.1992) )). Thus, as with the traffic stop study, Dr. Banks' checkpoint analysis is sufficiently reliable to be admissible under Rule 702.

c. Dr. Banks' Study of ACSO Checkpoint Arrests

The Government also challenges the reliability of Dr. Banks' study of ACSO's checkpoint arrests. Dr. Banks analyzed the outcomes from checkpoint stops for 10% of the checkpoints conducted by ACSO between 2009 and 2012. (Doc. 153 at 112.) The Government argues that Dr. Banks' analysis of 10% of checkpoints "is less probative than Dr. MacDonald's analysis of all arrests from checkpoints."⁷¹ (Doc. 158 at 114.) The Government also cites Federal Rule of Evidence 702(b) which allows an expert to provide opinion testimony if that testimony "is based on sufficient facts or data." While the Government's point is well-taken, this argument again addresses the weight rather than the reliability of Dr. Banks' arrest study. See Marvin Lumber & Cedar Co. v. PPG Indus., Inc., 401 F.3d 901, 915–16 (8th Cir.2005) (holding that argument concerning statistician's sample size goes toward the weight of the testimony "not to its evidentiary admissibility"). The Government does not contend that Dr. Banks' study is statistically or mathematically unreliable. Dr. Banks himself testified that the 10% sample "was adequate" for statistical purposes. (Doc. 153 at 113.) Therefore, the Government's criticisms of Dr. Banks' arrest study will be considered as to the weight, not the admissibility, of Dr. Banks' testimony on this point.

71 Technically, Dr. MacDonald testified that he examined all arrests and controlled for the stop reason, which presumably included checkpoints. (Doc. 149 at 23, 38.)

d. Dr. Banks' Permutation Test

The Government finally argues that the permutation test that Dr. Banks performed analyzing ACSO's checkpoint placement is unreliable. The Government argues that Dr. Banks assumes that "the checkpoint locations actually used by ACSO constitute the universe of all possible checkpoint locations."⁷² (Doc. 158 at 115; see also Doc. 115 at 15–16; Doc. 153 at 158–63.) As a result, the Government continues, Dr. Banks' test is "meaningless" and "merely shuffles the timing of checkpoint locations and does not provide information about the proximity of checkpoints to Latino neighborhoods." (Doc. 115 at 14–15; Doc 158 at 114–15.)

72 In its original motion to this court (Doc. 115), the Government also argued that another of the permutation test's assumptions—that the checkpoint locations were chosen independently of each other—rendered the study unreliable. (Doc. 115 at 16 (citing Doc. 86).) The Government elicited no testimony on this assumption on cross-examination at trial and made no mention of the assumption in its post-trial filing. At trial, Dr. Banks stated that this assumption was a "weak" one and did not change the result of his analysis. (Doc. 153 at 98–99.) The Government offers the court no persuasive reason to conclude that the assumption on the independence of checkpoint selection renders the permutation study unreliable.

To be sure, Dr. Banks acknowledged the assumptions he made in his permutation study. (Doc. 153 at 97–99.) Yet, the Government provided no evidence of other available sites to contradict Dr. Banks' methodological assumption that the checkpoint data represented all possible checkpoint sites in the County. In fact, ACSO's evidence demonstrated that its checkpoint locations must meet certain safety and logistical standards, limiting "the universe" of permissible locations. The court accepts the 305 sites used in the permutation test—every location at which ACSO conducted a checkpoint in the past four years—as a reasonable reflection of available sites in the County. Dr. Banks' analysis using those 305 sites does not render his study unreliable and thus inadmissible.

Importantly, the Government makes no argument that Dr. Banks' permutation test is statistically unreliable. Rather, the Government contends that his study is "meaningless" and of little probative value. (Doc. 115 at 14–15; Doc. 158 at 114–15.) Those arguments concern the relative import and persuasiveness of Dr. Banks' study rather than its admissibility as a matter of statistical science. See United States v. Cavely, 318 F.3d 987, 997–98 (10th Cir.2003) ; Covic v. Berk, No. 11–2571–STA–DKV, 2014 WL 4546806, at *4 (W.D.Tenn. Sept. 12, 2014) ; Wilspec Technologies, Inc. v. Dunan Holding Grp. Co., No. CIV–06–818–L, 2008 WL 7254328, at *1 (W.D.Okla. June 20, 2008). Thus, the Government's motion to strike Dr. Banks' permutation test as unreliable within the meaning of Rule 702 is denied.

C. Motion for Adverse Inference

Prior to trial, the Government moved the court to draw an adverse inference against Sheriff Johnson based on alleged spoliation of evidence. (See Doc. 131.) Sheriff Johnson responded (Doc. 139), and the Government replied (Doc. 145). On August 8, 2014, the court heard argument on the Government's motion and took it under advisement. (Doc. 156 at 51–52.) At the close of evidence, noting that the motion was still outstanding, the court made no final ruling. (Doc. 155 at 44.) In its post-trial filing, the Government renews its request for an adverse inference (Doc. 158 at 121 & n. 31), and Sheriff Johnson argues that the motion be denied (Doc. 157 at 16–20). For the reasons set forth below, the Government's motion for an adverse inference will be denied.

The Government contends that on June 28, 2011, before the filing of the present action, a former ACSO employee called an attorney for the Government and informed the attorney that, around June 15, 2011, Sheriff Johnson had directed deputies at a checkpoint in Green Level to "arrest any 'Mexicans' who did not have driver's licenses." (Doc. 131–2 ¶¶ 4–5.) Whoever allegedly heard this alleged statement never testified, and no evidence regarding it was presented at trial.⁷³ The former employee was also said to have informed the Government's attorney that Sheriff Johnson gave the alleged order via ACSO's central communications system ("C–COMM") and that communications on C–COMM were retained for only a short time period. (Id. ¶ 6.) Without a request for preservation, the recordings remain in the system for one year. (Doc. 156 at 34.) The Government attorney sent a letter by mail and email dated June 30, 2011, to Clyde Albright, the Alamance County Attorney. (Doc. 131–3 at 2; Doc. 156 at 38–39.) Up until that time, Albright was counsel of record for the County and Sheriff Johnson in the DOJ's then-pending declaratory judgment lawsuit seeking unfettered access to ACSO deputies as part of the DOJ's investigation of discriminatory policing that preceded the present action. See United States v. Alamance Cnty., No. 1:11–cv–507 (M.D.N.C.). The letter requested that Albright preserve all communications over the C–COMM system during the week of June 12–18, 2011. (Doc. 131–3 at 2; Doc. 156 at 38–39.) The letter further stated that the communications "may contain information relevant to" the Government's investigation. (Doc. 131–3 at 2.) At 12:46 p.m. that same day, however, S.C. Kitchen—ACSO's current attorney in this case—served a notice of appearance for ACSO in the declaratory judgment action, and at 2:11 p.m., Albright filed a notice of appearance on behalf of Alamance County. (Docs. 139–1; 139–2 at 1; 139–3; 156 at 38.) On July 5, 2011, Kitchen filed a notice of appearance in that same action on behalf of Sheriff Johnson in his official capacity. (Doc. 139–4.) The entry of separate counsel for the County appears to have been a result of (1) the Government's litigation position that it was entitled to interview ACSO deputies without ACSO or County counsel being present, and (2) the fact that, under North Carolina law, ACSO is a separate entity from the County.

73 The Government attempted to offer Deputy Randleman's inadmissible hearsay testimony that he had heard rumors of such a statement. (Doc. 149 at 194.)

On July 15, 2011, Kitchen explained to the Government attorney that he was appearing in order to represent the ACSO deputies during the DOJ's investigation. (Doc. 139–5.) The Government attorney acknowledged that, after the appearances by counsel, Kitchen—not Albright—represented Sheriff Johnson and ACSO, but it disputed Kitchen's authority to appear during interviews of deputies as their individual counsel. (Id. ) On July 20, 2011, the Government repeated its preservation request and included Kitchen on the email. (Doc. 131–5.)

The C–COMM system used by ACSO is under the control of Alamance County—not ACSO. (Doc. 156 at 41–42.) Employees controlling the system work for Alamance County—not ACSO. (Id. at 42.) ACSO employees have neither the knowledge nor ability to preserve C–COMM recordings. (Id. at 41.)

Thus, after receiving the June 30 email, Albright, as the Alamance County Attorney, sought to have the requested C–COMM recordings preserved. (Id. at 39.) He testified that he met with the C–COMM director, Dexter Bower, and Paula Crotts (the same individual involved in the previously mentioned checkpoint), informing them of the Government's request.⁷⁴ (Id. at 35, 39.) Bower and Crotts indicated they would keep the recordings. (Id. at 39.) On July 6, 2011, Crotts sent an email (with the subject "June 12–June 18 2011 ACSO CFS") to Albright with a 240–page computer-assisted dispatch report. (Id. at 39–40; see also Doc. 139–7.) Albright further testified that, after informing Bower and Crotts that the Government wanted the actual recordings and not the dispatch report, he believed that the recordings would be preserved.⁷⁵ (Doc. 156 at 40.) According to Albright, he had no reason to believe the recordings were not preserved until 2013. (Id. at 40–41.) In April 2013, Albright asked Crotts for the C–COMM recordings from June 12–18, 2011, but, by that time, the recordings were no longer in the system. (Id. at 35, 40–41.)

74 Crotts has no recollection of this meeting but does not deny that it occurred. (Doc. 156 at 37.) Her subsequent July 6, 2011 email (Doc. 139–7) is strong evidence that Albright's account is correct.

75 Crotts again said she has no recollection of meeting with Albright but does not deny that she did. (Doc. 156 at 37.)

The Government argues that, because the C–COMM recordings were lost, it is entitled to an adverse inference that a particular "recording—had it been preserved—would have contained evidence of Defendant's discriminatory intent," specifically that "Sheriff Johnson stated 'If he's Hispanic, take him to jail.' " (Doc. 158 at 121 & n. 31; see also Doc. 131.) The Government, however, has failed to demonstrate that the loss of the C–COMM recording merits an adverse inference against Sheriff Johnson.

"Spoliation refers to the destruction or material alteration of evidence or to the failure to preserve property for another's use as evidence in pending or reasonably foreseeable litigation." Silvestri v. Gen. Motors Corp., 271 F.3d 583, 590 (4th Cir.2001). If a finding of evidence spoliation is made, "the trial court has broad discretion to permit a jury to draw adverse inferences from a party's failure to present evidence, the loss of evidence, or the destruction of evidence." Vodusek v. Bayliner Marine Corp., 71 F.3d 148, 156 (4th Cir.1995). Application of an adverse inference requires a showing that a party (1) knew that certain evidence was relevant to some issue at trial; (2) had that evidence within its control; and (3) caused the loss of the evidence through its willful conduct. See Vulcan Materials Co. v. Massiah, 645 F.3d 249, 259–60 (4th Cir.2011) ; Victor Stanley, Inc. v. Creative Pipe, Inc., 269 F.R.D. 497, 526 (D.Md.2010) ("[A]n adverse inference instruction makes little logical sense if given as a sanction for negligent breach of the duty to preserve, because the inference that a party failed to preserve evidence because it believed that the evidence was harmful to its case does not flow from mere negligence.").

Sheriff Johnson did know that the requested information was relevant. In its June 30 letter to Albright, the Government clearly states that the recordings "may contain information relevant to" the Government's investigation. (Doc. 131–3 at 2.) Moreover, the Government repeated this request in its July 20 email, and Kitchen—acting as attorney for Sheriff Johnson—received a copy of that request. (Doc. 131–5 at 2.) Although counsel for Sheriff Johnson claims he did not know why the recordings were relevant, he did know that the Government believed they were. That is sufficient to trigger a duty to preserve. See Silvestri, 271 F.3d at 591 (observing that the duty to preserve arises "not only during litigation but also extends to that period before the litigation when a party reasonably should know that the evidence may be relevant to anticipated litigation").

That being said, the Government has not shown that Sheriff Johnson had control of the recordings or that the loss of the recordings was the result of willful conduct. First, Sheriff Johnson had no control over the recordings because the C–COMM system operates under the direction and control of Alamance County and not ACSO. (Doc. 156 at 41–42.) Sheriff Johnson and Alamance County are different legal entities. Neither he nor his deputies is an employee or agent of the county. See Capers v. Durham Cnty. Sheriff Dep't, 1:07cv825 (M.D.N.C. July 17, 2009) (Doc. 85), adopting recommendation, 2009 WL 798924, at *6 (M.D.N.C. March 23, 2009) ; McLaughlin v. Bailey, ––– N.C.App. ––––, 771 S.E.2d 570, 576 (2015) ; Clark, 450 S.E.2d at 749. Therefore, because the C–COMM system and its recordings are under the control of Alamance County, Sheriff Johnson did not have control over the C–COMM recordings and will not be held responsible for their loss. See Vulcan, 645 F.3d at 260 ; Hodge v. Wal–Mart Stores, Inc., 360 F.3d 446, 450–51 (4th Cir.2004) (concluding that a party that "did not have control of" evidence could not be blamed for its loss).

In addition, no evidence suggests that the loss of the recordings was the result of either willful destruction or loss by Sheriff Johnson. See Turner v. United States, 736 F.3d 274, 282 (4th Cir.2013) ("[S]poliation does not result merely from the 'negligent loss or destruction of evidence.' " (quoting Vodusek, 71 F.3d at 156 )). If any misconduct occurred, it was by C–COMM staff and Alamance County, which operate outside of Sheriff Johnson's control. At best, this leaves the Government with the assertion that Kitchen somehow conspired to allow the recordings to be destroyed. But, there is no evidence of that. While there remains a factual dispute between County Attorney Albright and County/C–COMM employee Crotts about the nature of Albright's preservation requests, the presence of Crotts' July 6, 2011 email to Albright is strong evidence that he made the preservation request. And Albright's testimony that he forwarded Crotts' responsive email and dispatch attachments to Kitchen (Doc. 156 at 39–40) is evidence that Kitchen—as attorney for Sheriff Johnson—had a reasonable belief that Albright had made the necessary requests to C–COMM to preserve the recordings. Thus, the lack of willful conduct evidence also precludes an adverse inference against Sheriff Johnson, and the motion will be denied.

D. Motion to Exclude Kenneth Evans' Testimony

At trial, counsel for Sheriff Johnson moved to strike the testimony of former lieutenant Kenneth Evans based on an alleged violation of Rule 4.2 of North Carolina's Professional Rules of Conduct. See L.R. 83.10e(b) (adopting North Carolina's ethical rules). Rule 4.2 states,

During the representation of a client, a lawyer shall not communicate about the subject of the representation with a person the lawyer knows to be represented by another lawyer in the matter, unless the lawyer has the consent of the other lawyer or is authorized to do so by law or a court order.

In its post-trial filing, the Sheriff claims that Government counsel violated Rule 4.2 by speaking with Evans without the counsel for the Sheriff present. (See Doc. 147 at 65 (containing Evans' testimony that he met with the Government prior to litigation)). For several reasons, Sheriff Johnson's motion to strike Evans' testimony will be denied.

First, Sheriff Johnson argues that "the issue" of whether the Government could "interview deputies without the presence of the attorney for the Sheriff" has been "previously determined, and is therefore res judicata. " (Doc. 157 at 89.) In making this argument, Sheriff Johnson references the prior lawsuit by the DOJ seeking a declaratory judgment against Sheriff Johnson, ACSO, and Alamance County. (Id. ) More specifically, that lawsuit sought a declaration that Rule 4.2 permitted DOJ's attorneys to interview certain ACSO non-command staff and all former ACSO employees outside the presence of Defendants' counsel. United States v. Alamance Cnty., No. 1:11–cv–507 (M.D.N.C. June 23, 2011) (Doc. 1). After Alamance County moved to dismiss the Government's complaint and the Government moved for summary judgment, the Government filed an unopposed motion to dismiss its action as moot. United States v. Alamance Cnty., No. 11–cv–507 (M.D.N.C. Sept. 18, 2012) (Doc. 51). This court entered an Order dismissing the action with prejudice. United States v. Alamance

Cnty., No. 11–cv–507 (M.D.N.C. Sept. 18, 2012) (Doc. 53).

Sheriff Johnson, however, cites no authority explaining how this court's Order granting the Government's voluntary dismissal operates as res judicata in this case.⁷⁶See, e.g., Ohio Valley Envtl. Coal. v. Aracoma Coal Co., 556 F.3d 177, 210 (4th Cir.2009) (listing the basic requirements for res judicata).

76 Sheriff Johnson cites this court's order dismissing his motion for judgment on the pleadings (Doc. 19), in which this court assumed, without deciding, that its prior Order granting the Government voluntary dismissal was a final judgment. (Id. at 7.) The court, however, did not hold that its prior order in the declaratory judgment action was a final judgment on the merits. (Id. at 6–9 (holding that, assuming that its prior order was a final judgment, that order did not bar § 14141 claims under the declaratory judgment exception).)

Also, in addition to the absence of argument on the applicability of res judicata, Sheriff Johnson offers only the conclusion that the Government violated Rule 4.2 without citing any case law or legal rule explaining how the Government violated the rule. See, e.g., N.C. R. Prof'l Conduct 4.2, cmt. 9 (prohibiting communications with only certain employees within an organization); United States v. Joseph Binder Schweizer Emblem Co., 167 F.Supp.2d 862, 866 (E.D.N.C.2001) (holding, in a criminal case, that the federal government could meet with persons unrepresented by counsel pre-indictment because it was "authorized by law to do so"). Sheriff Johnson's unsupported legal conclusions fail to comply with this court's Local Rules. See Local Rule 7.2(a)(4) (requiring that briefs contain "all statutes, rules and authorities relied upon"). The court need not, and will not, devise arguments or scour case law to support a party's legal conclusions. See Hayes v. Self–Help Credit Union, No. 1:13–CV–880, 2014 WL 4198412, at *2 (M.D.N.C. Aug. 22, 2014) ("It is not the role or the responsibility of the Court to undertake the legal research needed to support or rebut a perfunctory argument."). Thus, Sheriff Johnson's motion to strike Evans' testimony will be denied.

Moreover, even assuming without deciding that counsel for the Government violated Rule 4.2, there is no evidence that the assumed violation prejudiced Sheriff Johnson at trial. See United States v. Quest Diagnostics Inc., 734 F.3d 154, 167 (2d Cir.2013) (noting that the "relevant inquiry" for determining a remedy to an ethical violation is the "possibility of prejudice at trial" (quoting U.S. ex rel. Fair Lab. Practices Assocs. v. Quest Diagnostics Inc., No. 05 CIV. 5393, 2011 WL 1330542, at *11 (S.D.N.Y. Apr. 5, 2011) (internal quotation marks omitted))); McCallum v. CSX Transp., Inc., 149 F.R.D. 104, 113 (M.D.N.C.1993) ("The nature and extent of prejudice suffered or likely to be suffered in the future by the other parties is also a relevant consideration. While the sports based aphorism 'No harm; No foul' cannot be wholly transplanted to matters involving ethical violations, the extent of harm cannot be ignored."). Sheriff Johnson certainly points to none. There is also no evidence that Evans' statements to the Government's attorney were used at trial. Rather, Sheriff Johnson simply argues that a violation of Rule 4.2 occurred and that Evans' testimony should therefore be struck. (See Doc. 157 at 89–90.) Thus, even if an ethical violation had occurred, the lack of any apparent harm or prejudice resulting from the allegedly unethical contact with Evans would similarly doom Sheriff Johnson's motion to strike the lieutenant's testimony. See Reliable Money Order, Inc. v. McKnight Sales Co., 704 F.3d 489, 502 (7th Cir.2013) ("[W]hen an ethical breach neither prejudices an attorney's client nor undermines the integrity of the judicial proceedings, state bar authorities are generally better positioned to address the matter through disciplinary proceedings, rather than the courts through substantive sanction in the underlying lawsuit."). Although the court denies the Sheriff's motion, the Government's pre-litigation contact with Evans can be considered as it would be with any other fact witness in assessing the credibility and weight of the testimony.

E. 42 U.S.C. § 14141 Claims

On the merits, the Government brings two claims under 42 U.S.C. § 14141. First, it contends that the evidence offered at trial establishes that ACSO engaged in a "pattern or practice" of Fourteenth Amendment violations. Second, it maintains that ACSO committed a "pattern or practice" of Fourth Amendment violations.

Section 14141 of Title 42 —titled "Cause of Action"—reads in full:

(a) Unlawful conduct

It shall be unlawful for any governmental authority, or any agent thereof, or any person acting on behalf of a governmental authority, to engage in a pattern or practice of conduct by law enforcement officers or by officials or employees of any governmental agency with responsibility for the administration of juvenile justice or the incarceration of juveniles that deprives persons of rights, privileges, or immunities secured or protected by the Constitution or laws of the United States.

(b) Civil action by Attorney General

Whenever the Attorney General has reasonable cause to believe that a violation of paragraph (1) has occurred, the Attorney General, for or in the name of the United States, may in a civil action obtain appropriate equitable and declaratory relief to eliminate the pattern or practice.⁷⁷

77 The phrase "paragraph (1)" appears to be a textual error in the statute and should likely read "paragraph (a)."

The statute does not define what constitutes a "pattern or practice" of constitutional rights deprivations. Both the Government and Sheriff Johnson provide only a single citation to the Supreme Court's decision in Int'l Bhd. of Teamsters v. United States, 431 U.S. 324, 97 S.Ct. 1843, 52 L.Ed.2d 396 (1977), an employment discrimination case under Title VII of the Civil Rights Act of 1964, as the interpretative source for § 14141's "pattern or practice" requirement. (See Doc. 157 at 90; Doc. 158 at 82.) Section 14141's legislative history similarly sheds no interpretive light on the "pattern or practice" standard. See Marshall Miller, Police Brutality, 17 Yale L. & Pol'y Rev. 149, 165–66 & n. 109 (1998) (observing that "the direct legislative history of § 14141 is virtually non-existent").

That being said, the legislative history of § 14141's predecessor bill, the Police Accountability Act of 1991 ("PAA"), H.R. 2972, 102nd Cong. (1991), offers some, albeit limited, assistance in interpreting § 14141. See H.R. 3371, 102d Cong. § 1202 (1991); H.R.Rep. No. 102–242, pt. 1 (1991). The predecessor bill contained language nearly identical to § 14141 and particularly the "pattern or practice" phrasing. See H.R. 3371, 102d Cong. § 1202(a)(1) (1991). In a 1991 report on the PAA, the House Judiciary Committee ("the Committee") included its interpretation of the "pattern or practice" expression. H.R.Rep. No. 102–242, pt. 1, at 137–39 (1991). It began by noting that "[t]he Justice Department currently lacks the authority to address systemic patterns or practices of police misconduct." Id. at 137. The Committee then drew a parallel to the "pattern or practice authority" of the Attorney General "under eight civil rights statutes including those governing voting, housing, employment, education, public accommodations and access to public facilities." Id. Another comparison was made to 42 U.S.C. § 1983, with the Committee observing, "The Act does not increase the responsibilities of police departments or impose any new standards of conduct on police officers. The standards of conduct under the Act are the same as those under the Constitution, presently enforced in damage actions under section 1983." Id. at 138; see also id. (citing the Federal Government's present inadequate ability to curb "patterns or practices such as the lack of training or the routine use of deadly techniques like chokeholds, or the absence of a monitoring and disciplinary system" as impetus for the bill); id. at 138–39 (citing two lawsuits against police departments "illustrat[ing] both the need for [the Attorney General's] authority and how it will work").

Given the limited background provided by the Committee on a bill other than § 14141, other judicial constructions of the phrase "pattern or practice" provide further guidance. The parties have not identified any federal decision that has applied § 14141's "pattern or practice" standard. The court has found one recent decision applying § 14141's standard on a motion for summary judgment, but that decision offers little interpretive guidance.⁷⁸See United States v. Town of Colorado City, Ariz., No. 3:12–CV–8123–HRH, 2015 WL 3774315, at *4–6 (D.Ariz. June 17, 2015). The phrase "pattern or practice," however, is not a concept foreign to federal courts. Congress repeatedly uses the expression in federal statutes when authorizing the Attorney General to bring suit. See, e.g., 42 U.S.C. § 2000e–6(a) (authorizing the Attorney General to pursue injunctive relief in cases concerning a "pattern or practice" of employment discrimination); 42 U.S.C. § 3614(a) (authorizing the Attorney General to bring civil action in cases involving a "pattern or practice" of Fair Housing Act violations); 52 U.S.C. § 10101(e) (authorizing courts to find a "pattern or practice" of voting rights deprivations).

78 Also, in Nat'l Cong. for Puerto Rican Rights v. New York City, 75 F.Supp.2d 154 (S.D.N.Y.1999), the district court suggested an overlap between § 14141 and § 1983. Id. at 165–66. In that case, the court denied a request to stay a class action bringing § 1983 claims pending resolution of a DOJ investigation pursuant to § 14141. Id. Among other reasons for denying the stay request, the court observed that any legal rulings in the § 1983 class action would be res judicata and have collateral estoppel effect in any potential § 14141 litigation. Id.

Two areas of law—employment discrimination and § 1983 litigation—prove particularly useful. First, as both parties urge, in the employment context the Supreme Court has observed, "[T]he 'pattern or practice' language ... of Title VII ... was not intended as a term of art, and the words reflect only their usual meaning." Int'l Bhd. of Teamsters, 431 U.S. at 336 n. 16, 97 S.Ct. 1843. To demonstrate a "pattern or practice" claim under Title VII, the Government must "prove more than the mere occurrence of isolated ... or sporadic discriminatory acts"; it must demonstrate that the acts were "standard operating procedure—the regular rather than the unusual practice." Id. at 336, 97 S.Ct. 1843 ; see also id. at 336 n. 16, 97 S.Ct. 1843 ("[A] pattern or practice would be present only where the denial of rights consists of something more than an isolated, sporadic incident, but is repeated, routine, or of a generalized nature." (quoting 110 Cong. Rec. 14,270 (1964))); Cooper v. Fed. Reserve Bank of Richmond, 467 U.S. 867, 877–78, 104 S.Ct. 2794, 81 L.Ed.2d 718 (1984) (concluding that an attempt to prove a "pattern or practice" claim "may fail even though discrimination against one or two individuals has been proved"); EEOC v. Am. Nat. Bank, 652 F.2d 1176, 1188 (4th Cir.1981) (concluding that government's "pattern or practice" burden may be met "by statistics alone, ... or by a cumulation of evidence, including statistics, patterns, practices, general policies, or specific instances of discrimination" (citations omitted)); Stastny v. S. Bell Tel. & Tel. Co., 628 F.2d 267, 278 (4th Cir.1980) (holding that a "pattern or practice" of employment discrimination can be shown through a "sufficient number of instances of similar discriminatory treatment" or statistical data). Legal commentators also suggest that the definition of "pattern or practice" supplied by employment law should persuasively guide interpretations of § 14141. See Miller, supra, at 166–67, 169–72; Eugene Kim, Vindicating Civil Rights Under 42 U.S.C. § 14141: Guidance from Procedures in Complex Litigation, 29 Hastings Const. L.Q. 767, 778–80 (2002).

Second, the Supreme Court has required "a pervasive pattern," "persistent and widespread discriminatory practices of state officials," or "systematic maladministration" to establish a governmental "custom" of constitutional rights deprivations under 42 U.S.C. § 1983. Monell v. Dep't of Soc. Servs., 436 U.S. 658, 685 n. 45, 690–91, 98 S.Ct. 2018, 56 L.Ed.2d 611 (1978) ; Rizzo v. Goode, 423 U.S. 362, 375–76, 96 S.Ct. 598, 46 L.Ed.2d 561 (1976) (reversing lower court's findings of a pattern of police misconduct); see also Owens v. Baltimore City State's Attorneys Office, 767 F.3d 379, 403 (4th Cir.2014) ("Sporadic or isolated violations of rights will not give rise to Monell liability; only 'widespread or flagrant' violations will." (quoting Spell v. McDaniel, 824 F.2d 1380, 1390 (4th Cir.1987) )). As other commentators have observed, the federal government's inability to bring a § 1983 action appears to be a motivating force behind § 14141's enactment. See, e.g., Miller, supra, at 166–69.

With this background in mind, the court turns to each of the Government's § 14141 claims.

1. Fourteenth Amendment Claim

The Government first alleges that ACSO violated § 14141 by engaging in a pattern or practice of discriminatory law enforcement directed against Hispanics in violation of the Fourteenth Amendment. The Fourteenth Amendment's Equal Protection Clause bars any state from "deny[ing] to any person within its jurisdiction the equal protection of the laws." U.S. Const. amend. XIV, § 1. The Clause "is essentially a direction that all persons similarly situated should be treated alike." City of Cleburne, Tex. v. Cleburne Living Ctr., Inc., 473 U.S. 432, 439, 105 S.Ct. 3249, 87 L.Ed.2d 313 (1985).

Proving an equal-protection claim involves a two-step analysis. First, the Government must demonstrate the unequal treatment of a person or group of persons as compared to similarly situated individuals. See id.; Morrison v. Garraghty, 239 F.3d 648, 654 (4th Cir.2001). According to the Fourth Circuit, unequal treatment constituting an equal protection violation "occurs in one of two ways: (1) when the government explicitly classifies people based on race, or (2) when a law is facially neutral, but its administration or enforcement disproportionately affects one class of persons over another and a discriminatory intent or animus is shown." Monroe v. City of Charlottesville, 579 F.3d 380, 388 (4th Cir.2009). Express classifications are those that are "explicitly stated on the face of a statute or in the reasons given for its administration or enforcement." Sylvia Dev. Corp. v. Calvert Cnty., 48 F.3d 810, 819 (4th Cir.1995). Suspect facially neutral classifications are simply those that are not explicitly stated but are nevertheless applied. Id. at 818–19.

Second, this court must "determine whether the disparity in treatment can be justified under the requisite level of scrutiny." Morrison, 239 F.3d at 654. "The level of scrutiny depends on the type of classification." Sansotta v. Town of Nags Head, 724 F.3d 533, 542 (4th Cir.2013).

Here, the Government contends that ACSO's law enforcement policies reflect both an express classification and a facially neutral, but nevertheless discriminatory, classification. (See Doc. 158 at 84.) The Government argues that ACSO bases those classifications on ethnicity.⁷⁹ (Id. at 83–84.) If proven, any unequal treatment based on ethnicity is subject to strict scrutiny. See Fisher v. Univ. of Tex. at Austin, ––– U.S. ––––, 133 S.Ct. 2411, 2417, 186 L.Ed.2d 474 (2013) ; Regents of Univ. of Cal. v. Bakke, 438 U.S. 265, 291, 98 S.Ct. 2733, 57 L.Ed.2d 750 (1978) ("Racial and ethnic distinctions of any sort are inherently suspect and thus call for the most exacting judicial examination."); see also City of Cleburne, 473 U.S. at 440, 105 S.Ct. 3249 (applying strict scrutiny to unequal treatment based on national origin). To survive strict scrutiny, a classification based on ethnicity "must (1) be narrowly tailored and (2) further a compelling government interest." Monroe, 579 F.3d at 390.

79 The Government does not argue that any ACSO policy or conduct was irrational, if not proven to be based on a suspect or quasi-suspect classification. See Sansotta, 724 F.3d at 542–43.

a. Express Classification

The Government first claims that ACSO's law enforcement practices and policies expressly classify individuals based on their ethnicity. (See Doc. 158 at 84.) Laws, policies, and practices "that explicitly distinguish between individuals" on the basis of ethnicity "fall within the core of" the Equal Protection Clause. Shaw v. Reno, 509 U.S. 630, 642, 113 S.Ct. 2816, 125 L.Ed.2d 511 (1993). While oral statements of government officials have rarely been held to be express classifications, one court has held that such statements expressly classify but "only ... at the managerial level, when commanders and supervisors acting in accordance with the policy instruct officers to target racially defined groups for stops." Floyd, 959 F.Supp.2d at 663 n. 768.

In support of its contention, the Government argues that two sets of facts demonstrate a pattern or practice of express classifications:⁸⁰ (1) Sheriff Johnson's alleged order(s) to "arrest Hispanics" during an ACSO meeting; and (2) his alleged direction to "arrest Hispanics" at a checkpoint near a predominantly Hispanic mobile home park.⁸¹ (Doc. 158 at 86–87.)

80 The Government points to two other incidents—both involving ACSO's alleged use of "discriminatory criteria to select Alamance County employees for investigation"—as demonstrating a purported express classification. (Doc. 158 at 88 (citing Deputy Randleman's investigation of the driver who hit Commissioner Vaughan and the investigation of a Hispanic library employee requested by another Commissioner).) To be sure, a law enforcement agency cannot be protected from complicity in another's discrimination simply by claiming it was requested or directed to do so. But the two supporting examples provided by the Government show that individuals outside of ACSO not only initiated those investigations but identified the alleged suspects as Hispanic. The Government fails to provide any citation or explanation for why those incidents are express classifications under the Fourteenth Amendment. See Monroe, 579 F.3d at 388–89 (holding that use of ethnic information—originating outside the police department—in description of crime's perpetrator is not an express classification); Brown v. City of Oneonta, N.Y., 221 F.3d 329, 337–38 (2d Cir.2000) (finding no express racial classification where police used race in description of suspect).

81 Although not cited by the Government as evidence of an express classification, the court notes one other alleged incident: the Government offered Corporal Nicholson, who testified that Sheriff Johnson once said, "Go get the Mexicans" and "arrest every chili shitter in the park." (Doc. 147 at 77, 81.) As stated in its Findings of Fact, the court questions the veracity of this allegation, as Sheriff Johnson and the other officer attending the meeting deny those statements. Moreover, the persuasive evidence is that Sheriff Johnson stated "Go get those Mexicans" or "the Mexicans" in reference to a specific Mexican gang then under ACSO investigation for criminal activity. Corporal Nicholson also testified that he made no stops or arrests as a result of the alleged order. (Doc. 147 at 89.)

To be sure, the record reveals no evidence that ACSO had a policy—written or unwritten—that expressly classified Hispanics. The Government points to no ACSO document containing any policy, and no witness testified to any. Rather, the Government relies on testimony as to Sheriff Johnson's verbal directives to arrest Hispanics on these limited occasions and evidence that all deputies are duty-bound to carry out all of the Sheriff's orders. It is a weak claim, however. There was no evidence that even a single person—including those claiming they heard such an alleged order—implemented or executed any such directive.⁸² This total lack of implementation, particularly when viewed with the failure to provide any context for such statements, casts serious doubt on both the veracity of the Government's witnesses' characterization of them and the claim that they rise to the level of a practice or policy. Only a very few officers claimed to have heard such a statement. And, had any officer regarded such a statement to be a directive, like those from the Sheriff that he or she was duty-bound to obey, as the Government suggests, it is odd that not a single officer or employee ever complained about it, challenged it, or was concerned that he or she would risk any adverse action at all for failing to follow it. Whatever Sheriff Johnson may have said, it is unlikely that it was as portrayed, singling out Hispanics expressly.

82 The parties have not addressed the extent to which an express classification must be implemented in order to create a Fourteenth Amendment violation. The Supreme Court and lower courts have suggested ethnic classifications may trigger strict scrutiny only when official state action subjects persons to that classification or when the government actor utilizes that classification. See, e.g., Adarand Constructors, Inc. v. Pena, 515 U.S. 200, 224, 115 S.Ct. 2097, 132 L.Ed.2d 158 (1995) ("[A]ny person, of whatever race, has the right to demand that any governmental actor subject to the Constitution justify any racial classification subjecting that person to unequal treatment under the strictest judicial scrutiny." (emphasis added)); Wayte v. United States, 470 U.S. 598, 610 n. 10, 105 S.Ct. 1524, 84 L.Ed.2d 547 (1985) (concluding that "[a] showing of discriminatory intent is not necessary when the equal protection claim is based on an overtly discriminatory classification," suggesting that a showing of discriminatory effect is still required); Sylvia Dev. Corp., 48 F.3d at 819 (stating that strict scrutiny applies when "the classification utilized is explicitly stated"); see also Monell, 436 U.S. at 694, 98 S.Ct. 2018 ("[I]t is when execution of a government's policy or custom, whether made by its lawmakers or by those whose edicts or acts may fairly be said to represent official policy, inflicts the injury that the government as an entity is responsible under § 1983."). But see Doe ex rel. Doe v. Lower Merion Sch. Dist., 665 F.3d 524, 545 (3d Cir.2011) ("A statute or policy utilizes a 'racial classification' when, on its face, it explicitly distinguishes between people on the basis of some protected category." (quoting Hayden v. Cnty. of Nassau, 180 F.3d 42, 48 (2d Cir.1999) )).

While not controlling, it is notable that the Government's express classification evidence falls far short of that found sufficient in two recent, analogous cases. In Floyd v. New York City, the district court found that the New York Police Department's stated policy that "targeted young blacks and Hispanics for stops" explicitly classified on the basis of race and ethnicity, citing testimony by police officers that they targeted "the right people" and the chief of police admitted that the "right people" are "young men of color in their late teens, early 20s." 959 F.Supp.2d at 602–03 & n. 280, 663 & n. 767. Similarly, in Melendres v. Arpaio, 989 F.Supp.2d 822 (D.Ariz.2013), aff'd in part, vacated in part, 784 F.3d 1254 (9th Cir.2015), the district court held that an express classification existed where a police department's written policies "expressly permitted officers to make racial classifications" as a factor in developing probable cause or reasonable suspicion. Id. at 846–47 (noting that ICE training manual "expressly allows for consideration of race" by permitting consideration of "apparent Mexican ancestry" as relevant to reasonable suspicion).

What little context there is for the statements in this case fails to support the Government's characterization. Officer Lloyd testified that, while assisting at a checkpoint at the predominantly Hispanic Rocky Top mobile home park, then-Chief Deputy McPherson initially reported that Sheriff Johnson had said that any Hispanics driving without a driver's license or driving with a revoked license should be arrested.⁸³ (Doc. 147 at 156–57.) However, after several officers, including Officer Lloyd, sought some clarification, Chief Deputy McPherson spoke again with Sheriff Johnson and then reported back that the Sheriff "didn't mean [to arrest] just Hispanics" but rather also "Hispanics, whites, and blacks." (Doc. 147 at 158–59.)⁸⁴ While the clarification was inartful at best (there being no need to refer to race at all), it demonstrates that the directive was either misstated or misunderstood and was meant to apply to everyone. This is consistent with Officer Lloyd's testimony that he never understood Sheriff Johnson to give him an order on how to perform his traffic duties. (Id. at 150–51.)

83 Chief Deputy McPherson died in 2012. (Doc. 154 at 41.)

84 The Government never questioned Sheriff Johnson about this incident.

The Government's theory also requires a strained interpretation of other trial evidence. The Government contends that Sheriff Johnson made these statements because he wanted to move Hispanics through the 287(g) program and reap the financial benefit from housing federal detainees. But, as noted, the Government offered no evidence as to how many (or few) Hispanics arrested by ACSO were booked into the ACDC and subjected to 287(g) questioning—figures almost certainly available to the Government.⁸⁵ There was evidence, however, that in reality a relatively low number of ACDC detainees were processed by ACSO's 287(g) officers—approximately one detainee per week, and most of those housed in the ACDC came from other counties and ICE. (Doc. 152 at 163–64; Doc. 154 at 22–23.) Thus, ACSO officers contributed infrequently to the number of ICE detainees housed in the ACDC, undermining the Government's suggestion that ACSO significantly contributed arrestees to ACDC's ICE detainee population to receive financial benefits. Moreover, the Government's statistical evidence, on which it relies very heavily, contradicts its theory of the case—whether articulated as being that the Sheriff directed that all Hispanics be arrested, or only those without licenses and identification.⁸⁶ Dr. MacDonald's testimony made clear that, in 87.9% of the cases, the opposite occurred: 55.8% of Hispanics stopped were merely issued a citation; 22.3% were given only a verbal warning; 5.0% were given only a written warning, and no action at all was taken as to 4.8%. (Doc. 149 at 21, 29–37.)

85 To be sure, there was no evidence that ACSO ever held any arrestee solely for the purpose of determining his or her immigration status.

86 The Government never provided any evidence as to why anyone was arrested or cited, including whether or not they were able to provide identification, to substantiate the theory that Hispanics were arrested because they lacked identification.

Finally, Sheriff Johnson also denies that he ever gave any such order either at a staff meeting or checkpoint, and he denies even having been present at the checkpoint testified to by Officers Lloyd, Perry, and Evans. The Sheriff's testimony was corroborated by that of nine ACSO employees—some of whom were offered by the Government—who also testified that they had never heard him give an instruction to arrest Hispanics at an ACSO meeting. As to the alleged statement at the checkpoint, Chief Deputy Britt stated that ACSO only conducted an informational checkpoint following a series of break-ins in the area where Sheriff Johnson allegedly made the statement. No checkpoint data corroborating the Government's witnesses' testimony was presented. And, while the court relies on the quality of the evidence and not the number of witnesses on any particular issue, it notes that each person testifying to having heard the statements carried some credibility burden, affecting how they attempted to characterize what they say they heard.

Thus, in light of this conflicting and limited evidence, the court is not persuaded that a directive classifying Hispanics existed, as the Government suggests.

But, beyond this difficulty is the Government's failure to prove that the alleged statements, even assuming they were made, would establish a "pattern or practice" of Fourteenth Amendment violations as required by § 14141.

To demonstrate a pattern or practice, the Government must "prove more than the mere occurrence of isolated ... or sporadic discriminatory acts." Int'l Bhd. of Teamsters, 431 U.S. at 336, 97 S.Ct. 1843 (concluding that discriminatory actions must have been the "standard operating procedure [-] the regular rather than the unusual practice"); see also Monell, 436 U.S. at 690–91, 98 S.Ct. 2018 (requiring proof of "persistent and widespread discriminatory practices of state officials"). This inquiry is a fact-specific one. See United States v. Cochran, 39 F.Supp.3d 719, 730–31 (E.D.N.C.2014).

Here, the Government cites two or three alleged instances approximately five to seven years ago, each with little context for Sheriff Johnson's alleged oral instruction to arrest Hispanics and with no individual being shown to have carried it out or have been subjected to it. That evidence is insufficient to establish a pattern or practice under prevailing case law. See Cooper, 467 U.S. at 878–79, 104 S.Ct. 2794 (concluding that "the statistical evidence, buttressed by expert testimony and anecdotal evidence by three individual [s] ... was not sufficient to support the finding of a pattern of ... discrimination"); Ste. Marie v. E. R.R. Ass'n, 650 F.2d 395, 405 (2d Cir.1981) (holding that seven acts of discrimination, consisting of three denials of promotions, two failures to promote, and two rejections of applications, could not, without more, support a finding of a pattern or practice); cf. Int'l Bhd. of Teamsters, 431 U.S. at 337–39, 97 S.Ct. 1843 (observing the bolstering of statistical evidence of a pattern or practice with "over 40 specific instances of discrimination" demonstrated pattern or practice); Chisholm v. U.S. Postal Serv., 665 F.2d 482, 494–96 (4th Cir.1981) (concluding that class action demonstrated claim of discrimination following anecdotal evidence from twenty employees in addition to statistical data). This is true even when taking into account the statistical and other evidence offered by the Government, as explained below.

Given all the evidence presented at trial and considering the totality of the circumstances, the Government therefore fails to demonstrate a pattern or practice of express, ethnic classifications in violation of the Fourteenth Amendment and § 14141.

b. Facially Neutral Classification

The Government alternatively argues that ACSO's law enforcement practices and policies have a discriminatory effect on Hispanics and are motivated by discriminatory animus toward them. (Doc. 158 at 84, 89.) When challenging a facially neutral classification, a plaintiff must prove both the discriminatory effect of and discriminatory purpose behind that classification. See Pers. Adm'r of Mass. v. Feeney, 442 U.S. 256, 272, 99 S.Ct. 2282, 60 L.Ed.2d 870 (1979) ("[E]ven if a neutral law has a disproportionately adverse effect upon a racial minority, it is unconstitutional under the Equal Protection Clause only if that impact can be traced to a discriminatory purpose."); Vill. of Arlington Heights v. Metro. Hous. Dev. Corp., 429 U.S. 252, 264–68, 97 S.Ct. 555, 50 L.Ed.2d 450 (1977) ; Washington v. Davis, 426 U.S. 229, 238–42, 96 S.Ct. 2040, 48 L.Ed.2d 597 (1976).

To demonstrate a Fourteenth Amendment claim, a plaintiff must demonstrate discriminatory effect by showing the unequal treatment of a person or persons as compared to other similarly situated individuals. See City of Cleburne, 473 U.S. at 439, 105 S.Ct. 3249 ; Morrison, 239 F.3d at 654. When considering discriminatory effect in the context of selective law enforcement under the Fourteenth Amendment, courts generally rely on the Supreme Court's decision in United States v. Armstrong, 517 U.S. 456, 116 S.Ct. 1480, 134 L.Ed.2d 687 (1996). See, e.g., Farm Labor Org. Comm. v. Ohio State Highway Patrol, 308 F.3d 523, 533–34 (6th Cir.2002) (obligating plaintiffs to demonstrate that "similarly situated individual[s]" received different treatment from them to prove selective enforcement claim); Chavez, 251 F.3d at 636 ; United States v. Dixon, 486 F.Supp.2d 40, 44–47 (D.D.C.2007).

In Armstrong, the Supreme Court addressed the evidence necessary to prove discriminatory effect for a claim of selective prosecution rather than selective law enforcement. 517 U.S. at 465–71, 116 S.Ct. 1480. Attempting to obtain discovery, the Armstrong defendant had presented evidence that, "in every one" of twenty-four cocaine base cases closed by the public defender's office, the defendant was black. Id. at 459, 116 S.Ct. 1480. The appeals court found this evidence sufficient to permit the plaintiff to obtain discovery, concluding that, as a general rule, a defendant need not show that the government failed to prosecute individuals similarly situated to him. Id. at 469, 116 S.Ct. 1480. The Supreme Court disagreed, rejecting the appeals court's omission of the similarly situated requirement and finding instead that "[t]he requirements for a selective-prosecution claim draw on 'ordinary equal protection standards.' " Id. at 465, 116 S.Ct. 1480 (quoting Wayte v. United States, 470 U.S. 598, 608, 105 S.Ct. 1524, 84 L.Ed.2d 547 (1985) ). Applying these "ordinary equal protection standards," the Supreme Court reiterated, "To establish a discriminatory effect in a race case, the claimant must show that similarly situated individuals of a different race were not prosecuted." Id. In doing so, "[t]he Supreme Court made clear that the similarly situated requirement could not be discarded, reaffirming that the requirement of showing discriminatory effect is a long established requirement in [federal] jurisprudence." Chavez, 251 F.3d at 638.

The Fourth Circuit appears to align with those courts looking to Armstrong for interpretative guidance, as it relied on Armstrong to deny a defendant's claim of selective law enforcement (as opposed to selective prosecution). See United States v. Bullock, 94 F.3d 896, 899 (4th Cir.1996) (noting Armstrong's "rigorous standard for proving such a violation"); see also Farm Labor Org. Comm., 308 F.3d at 533–34 ; Chavez, 251 F.3d at 636 ; Dixon, 486 F.Supp.2d at 44–45. The Government offers no argument that Armstrong or the Equal Protection Clause's similarly situated requirement do not apply in this case, and, in fact, both the Government's complaint and brief acknowledge and apply the "similarly situated" requirement.⁸⁷ (See Doc. 1 ¶¶ 4–5, 34–36, 38; Doc. 158 at 81, 117.) The court concludes, therefore, that Armstrong is appropriate and persuasive guidance for the consideration of the selective law enforcement claims raised here.

87 The Government also offers no argument that Armstrong or the Fourteenth Amendment's similarly situated requirement apply any differently in the context of a selective law enforcement claim.

In this case, the Government aims to prove discriminatory effect largely by relying on statistical evidence of disparities in ACSO's law enforcement practices.⁸⁸ Courts analyzing Armstrong in the context of selective law enforcement claims, however, differ as to whether a plaintiff must identify particular similarly situated individuals or can instead use statistical evidence of aggregated similarly situated individuals to show the discriminatory effect of selective law enforcement. Compare United States v. Viezca, 555 F.Supp.2d 1254, 1267 & n. 7 (M.D.Ala.2008) (interpreting the Eleventh Circuit to require identification of specific "similarly situated individuals in selective traffic enforcement claims"), aff'd sub nom. United States v. Ubaldo–Viezca, 398 Fed.Appx. 573 (11th Cir.2010), with Farm Labor Org. Comm., 308 F.3d at 534 ("A claimant can demonstrate discriminatory effect by naming a similarly situated individual who was not investigated or through the use of statistical or other evidence which 'address[es] the crucial question of whether one class is being treated differently from another class that is otherwise similarly situated.' " (quoting Chavez, 251 F.3d at 638 ) (alteration in original)), and Duque–Nava, 315 F.Supp.2d at 1153–56 (holding that statistical evidence is sufficient to prove a selective law enforcement claim). The distinction bears importance here because, while the Government's complaint alleged several specific instances of alleged discriminatory policing (see Doc. 1 ¶ 37), the Government has abandoned virtually all of them following discovery, leaving it with proof of discriminatory effect consisting almost wholly of statistical evidence.

88 The Government contends that its statistical evidence, alone, can establish discriminatory purpose as well as effect. (Doc. 158 at 119.) The Supreme Court has, in other contexts, recognized that in "rare" cases statistics alone may be able to demonstrate a constitutional violation. See Arlington Heights, 429 U.S. at 266, 97 S.Ct. 555 (observing that statistical evidence must present a "stark" pattern to be accepted as the sole proof of discriminatory intent); see also McCleskey v. Kemp, 481 U.S. 279, 294 n. 12, 107 S.Ct. 1756, 95 L.Ed.2d 262 (1987) (noting that only in "rare cases" will "a statistical pattern of discriminatory impact demonstrate[ ] a constitutional violation" and citing selection of the jury venire and Title VII violations as the two types of cases in which that demonstration is acceptable). In situations similar to the present case, however, courts have held that "statistics may not be the sole proof of a constitutional violation." Chavez, 251 F.3d at 647–48 ; accord Duque–Nava, 315 F.Supp.2d at 1164 ; Mesa–Roche, 288 F.Supp.2d at 1196. Here, because of the multiple deficiencies noted as to the statistical evidence, the court does not find it to be sufficient, alone or in combination with all credible record evidence, to demonstrate discriminatory intent to establish a § 14141 claim.

For purposes of this case, the court accepts that the discriminatory effect of selective law enforcement can be demonstrated through the use of statistical evidence of similarly situated individuals. There are several reasons for doing so. None of the cases appearing to require identification of specific, similarly situated individuals in a selective law enforcement case was actually presented with statistical evidence of discriminatory effect. See, e.g., Viezca, 555 F.Supp.2d at 1267 & n. 7. Moreover, the Supreme Court has never held that statistical evidence is insufficient to prove a claim of selective prosecution. See Chavez, 251 F.3d at 638 (explaining that the Armstrong Court rejected statistical evidence "not because plaintiffs can never use statistics to prove discriminatory effect, but because the particular statistics presented to the [Armstrong ] Court did not address the relevant issue" of whether similarly situated individuals received dissimilar treatment); see also United States v. Hastings, 126 F.3d 310, 316 (4th Cir.1997) (examining, in dicta, a defendant's statistical evidence in the context of a selective prosecution case). Rather, the Supreme Court has repeatedly held that the use of statistical evidence alone is usually an acceptable and sufficient method of proving discriminatory effect in its equal protection analysis. See, e.g., Hunter v. Underwood, 471 U.S. 222, 227, 105 S.Ct. 1916, 85 L.Ed.2d 222 (1985) (citing with approval the use of statistical evidence to show discriminatory effect in equal protection analysis); Int'l Bhd. of Teamsters, 431 U.S. 324 at 339, 97 S.Ct. 1843 (observing that, in jury selection context, statistical analyses play a valuable role "in cases in which the existence of discrimination is a disputed issue"); cf. Arlington Heights, 429 U.S. at 266, 97 S.Ct. 555 ("Sometimes a clear pattern, unexplainable on grounds other than race, emerges from the effect of the state action...."). Finally, as several courts have noted, requiring the identification of specific, similarly situated individuals is sometimes impractical in selective law enforcement cases. See, e.g., Chavez, 251 F.3d at 637–40 ; Bradley v. United States, 299 F.3d 197, 206 & n. 11 (3d Cir.2002). That is, "[b]ecause law enforcement agencies do not make or keep records on individuals they do not stop, and certainly not on 'similarly situated' individuals they do not stop, imposing such a requirement on ... any defendant who challenges a traffic stop as selective enforcement, effectively denies them any ability to discover or prove such a claim." Duque–Nava, 315 F.Supp.2d at 1155.

In addition to demonstrating discriminatory effect, a plaintiff must also demonstrate discriminatory purpose. This requires more than proof that an adverse effect was merely "foreseeable." Feeney, 442 U.S. at 278, 99 S.Ct. 2282. Establishing discriminatory purpose instead requires a sufficient showing that "the decisionmaker ... selected or reaffirmed a particular course of action at least in part because of, not in spite of, its adverse effects on an identifiable group." Sylvia Dev. Corp., 48 F.3d at 819 n. 2 (quoting Feeney, 442 U.S. at 279, 99 S.Ct. 2282 ) (internal quotation marks omitted). In other words, a plaintiff must show that racial discrimination was a "motivating factor" in the governing body's decision. Arlington Heights, 429 U.S. at 265–66, 97 S.Ct. 555 ; see also Talbert v. City of Richmond, 648 F.2d 925, 929 (4th Cir.1981). This "demands a sensitive inquiry into such circumstantial and direct evidence of intent as may be available." Arlington Heights, 429 U.S. at 266, 97 S.Ct. 555. The Fourth Circuit has highlighted several factors "probative of whether a decisionmaking body was motivated by a discriminatory intent." Sylvia Dev. Corp., 48 F.3d at 819. Those factors include the following:

(1) evidence of a "consistent pattern" of actions by the decisionmaking body disparately impacting members of a particular class of persons; (2) historical background of the decision, which may take into account any history of discrimination by the decisionmaking body or the jurisdiction it represents; (3) the specific sequence of events leading up to the particular decision being challenged, including any significant departures from normal procedures; and (4) contemporary statements by decisionmakers on the record or in minutes of their meetings.

Id.; see also Arlington Heights, 429 U.S. at 266–68, 97 S.Ct. 555. This list is by no means exhaustive. See Arlington Heights, 429 U.S. at 268, 97 S.Ct. 555 (identifying, "without purporting to be exhaustive, subjects of proper inquiry in determining whether racially discriminatory intent existed"); Talbert, 648 F.2d at 929. "In the end, the plaintiff has the burden of establishing that a classification introduced through administrative action was 'clear and intentional.' " Sylvia Dev. Corp., 48 F.3d at 819 (quoting Snowden v. Hughes, 321 U.S. 1, 8, 64 S.Ct. 397, 88 L.Ed. 497 (1944) ).

The Government's evidence of a facially neutral classification broadly covers six areas of ACSO's law enforcement practices: traffic stops; checkpoint placement and stops; post-stop outcomes; searches after stops; the 287(g) program; and internal procedures. While the court will address each area in turn, it also considers all of them collectively in the context of the entire evidentiary record for purposes of the constitutional inquiry. In sum, given the totality of circumstances adduced from all the trial evidence, the Government fails to demonstrate a pattern or practice of facially neutral but discriminatory classifications.

i. Traffic Stops

In its complaint and at trial, the Government contended that ACSO targeted Hispanics for traffic stops. The Government presented Dr. Lamberth's testimony, which has been excluded (and thus need not be addressed further), and evidence of three individual traffic stops performed by ACSO officers. However, none evinces a discriminatory intent or effect.

One traffic stop was Deputy Conklin's stop of a van (which unbeknownst to the deputy contained Hispanic passengers) for impeding traffic on Interstate 40. Deputy Conklin's investigation quickly revealed evidence of human trafficking (i.e., "multiple fast-food bags with trash in them" and bottles filled with urine), a criminal offense (N.C. Gen. Stat. § 14–43.11 ), which prompted the deputy to contact ICE, who in turn asked him to detain the van's occupants for fifty minutes to an hour until ICE agents arrived. (Doc. 150 at 94–95.) Nothing about this stop suggests a discriminatory intent on the part of Deputy Conklin. The Supreme Court has acknowledged that state law enforcement may legitimately cooperate with federal immigration officials, Arizona v. United States, ––– U.S. ––––, 132 S.Ct. 2492, 2507–09, 183 L.Ed.2d 351 (2012), and no evidence was presented that the deputy delayed the van (for ICE's arrival) beyond that which was required for him to resolve the State criminal inquiry.

Another stop—the lane-changing driver testified to by Assistant U.S. Attorney Husser—fails to show a discriminatory purpose or effect. The stop was simply described as a "bad stop," and Husser was unclear whether the driver was even Hispanic. (Doc. 148 at 192–93, 198.) Importantly, Husser testified that no ACSO stop he reviewed indicated the profiling of Hispanics.

Finally, the stop of Jose Luis Arzola, Jr., fails to demonstrate discriminatory purpose or effect. The Government provided no evidence indicating a lack of probable cause for the stop. Upon being stopped, Arzola was asked for his "papers," but it is unclear to what that term refers, as Arzola also originally failed to provide the officer his vehicle registration. (Doc. 147 at 185–86, 191–92.) Even assuming the officer meant immigration documents, that request does not appear to be unlawful in and of itself absent evidence that the request delayed the legitimate purpose of the traffic stop. See Arizona, 132 S.Ct. at 2509 ; Muehler v. Mena, 544 U.S. 93, 101, 125 S.Ct. 1465, 161 L.Ed.2d 299 (2005).

The Government's evidence of discriminatory profiling in ACSO's traffic stop practices thus fails to demonstrate a Fourteenth Amendment violation.

ii. Checkpoint Placement and Stops

The next area of challenge involves ACSO's checkpoint operations. The Government charges that ACSO targeted Hispanic drivers by siting checkpoints in predominantly Hispanic neighborhoods and stopping Hispanics discriminatorily at checkpoints. In the absence of Dr. Lamberth's inadmissible testimony, the Government relies on four incidents occurring at ACSO checkpoints and Dr. Banks' finding that 36% of checkpoint stops involved Hispanics.⁸⁹

89 Elsewhere, the Government notes Dr. MacDonald's analysis of searches at checkpoints and post-stop outcomes, which include those at checkpoints. For the reasons explained infra, however, that analysis is unpersuasive evidence of a discriminatory effect at ACSO checkpoints.

The first incident offered by the Government is the checkpoint interaction between Deputy Keller and Paula Crotts, during which the deputy (who knew Crotts personally) told her husband that she did not need to see his license, adding they were "there to get them some" and motioning toward a predominately Hispanic mobile home park. (Doc. 149 at 119.) Crotts said she assumed the word "some" alluded to "as many Hispanics" or "an Hispanic person" but admitted, "I don't know what [Deputy Keller] meant." (Id. at 120, 125.)

What Deputy Keller may have meant is at best murky, and Crotts' opinion, which came in without objection, is nevertheless speculative. It would support the Government's theory if she meant they were there to "get them some Hispanics." There is no evidence of the type of checkpoint being conducted, however. If it was an informational checkpoint, it would be entirely consistent for Deputy Keller to have waived off Mr. Crotts' effort to volunteer his license because the deputies would have been there to gather information to apprehend a specific suspect. And a comment in that regard could mean "some suspects" or "some criminals," or, as Crotts surmised, "an Hispanic person" in particular. (Id. at 120.) In fact, Crotts volunteered at trial that as a dispatcher for the County, "we have had a lot of calls at that mobile home park" such that she knew the address "by heart." (Id. at 117.) If it was a standard checkpoint, however, it does appear that Deputy Keller's failure to check Crotts' husband's license may have contravened ACSO's checkpoint policy (raising a potential Fourth Amendment concern), but that any departure was most likely based on the fact that she personally knew the Crottses. Under that scenario and even attributing the Government's gloss to Keller's statement, the probative value is weak because it is saddled with such vagueness. Next is the checkpoint set up by Corporal Nicholson near a predominately Hispanic mobile home park. The evidence at trial, however, showed that that the checkpoint followed Sheriff Johnson's direction to conduct some law enforcement activity in the area due to the presence of a specific Mexican gang then under ACSO investigation for criminal activity.

Third, the Government points to Sheriff Johnson's direction to "arrest Hispanics" at a checkpoint on Highway 49 one-half mile from the Seamsters mobile home park, which is predominantly Hispanic. (Doc. 147 at 29–30, 47, 155.) As noted earlier, there are problems with the Government's characterization of the directive. Nevertheless, even assuming the Sheriff gave such a directive, the Government's own evidence reveals that the officers claiming they heard the order did not carry it out. Thus, there was never any discriminatory effect.

Fourth is Officer Lloyd's testimony that then-Chief Deputy McPherson had reported at a checkpoint that the Sheriff had said that any Hispanics driving without a driver's license or driving with a revoked license should be arrested. Upon follow-up with Sheriff Johnson, however, Deputy McPherson made clear that he "didn't mean [arrest] just Hispanics" but "Hispanics, whites, and blacks." (Id. at 158–59.) Again, there was no evidence that anyone carried out any allegedly improper order.

Finally, the Government points to testimony by Sheriff Johnson's expert witness, Dr. Banks, that from 2009 to 2012, 36.8% of ACSO's stops at checkpoints were of Hispanics. (Doc. 153 at 99–100, 170.) The Government seeks to compare this figure to unadjusted U.S. Census data reporting that Hispanics constituted only 11.6% of Alamance County's population in 2010 and 8.6% of the County's driving-age population. (Id. at 170–71.) The Government contends this shows discriminatory effect. (Doc. 155 at 55–56; Doc. 158 at 110–11.)

The Government's argument here—comparing Dr. Banks' figure to unadjusted Census data—contravenes its steadfast position throughout trial that unadjusted U.S. Census data is unreliable for use as a benchmark and leads to erroneous results. (Doc. 153 at 145–46, 157.) Indeed, in its closing argument, the Government contended that U.S. Census data provided an "unreliable" benchmark and thus was not "probative" evidence of driving patterns in Alamance County. (Doc. 155 at 58–59; see also id. at 59 ("[U]nadjusted census data is widely discredited as a benchmark in social scientific literature.").) Dr. Banks confirmed the Government's criticisms by agreeing that use of U.S. Census data as the benchmark for drivers is not supported by academic literature and would not provide reliable information. (Doc. 153 at 145–46.) Thus, the Government's argument offers little on the issue of discriminatory effect. See Chavez, 251 F.3d at 643–44 (observing that use of U.S. Census data as a benchmark "tell[s] ... very little about the numbers of Hispanics" on the roads).

Without an appropriate benchmark, the Government's reliance on Dr. Banks' finding that 36.8% of the stops at checkpoints in Alamance County were of Hispanics is unreliable proof of discrimination. (Doc. 153 at 99–100.) The Government argues that this figure demonstrates that ACSO sites checkpoints disproportionately where Hispanics drive and stops them disproportionately. (Doc. 158 at 110–11.) On this record, however, Dr. Banks' figure demonstrates neither.

As to alleged discriminatory checkpoint siting, it bears reminding what Dr. Banks' figure allegedly reflects. The figure does not reflect all persons ACSO questioned at a checkpoint, or even those who drove through one. Rather, ACSO officers completed a stop form only for those actually detained beyond the initial questioning at the checkpoint. Consequently, without a legitimate benchmark, Dr. Banks' figure says nothing about the number of persons passing through ACSO checkpoints, or their ethnicity, and thus is not proof that ACSO sites checkpoints discriminately near Hispanic areas. This conclusion is buttressed by the fact that, while there was evidence that ACSO occasionally sited checkpoints near Hispanic neighborhoods, there were 435 total checkpoints over four years positioned over much of the County. The Government has all but ignored these other checkpoints and has not shown that there was any effort to focus checkpoints on areas more densely populated by Hispanics than any other area of the County. Rather, the persuasive evidence is that ACSO sited checkpoints throughout the County and conducted them based on legitimate law enforcement reasons.

Second, regarding alleged discriminatory checkpoint stopping, the question is whether Dr. Banks' figure is evidence that persons similarly situated to Hispanics were not detained (and thus not issued stop forms) based on ethnicity. This begs the question of what percentage of Hispanics engage in conduct that prompted the detainment and issuance of a stop form. ACSO officers can detain a driver only after they determine there is reasonable suspicion to believe that the driver or vehicle passenger has violated the law. (See Gov't Trial Ex. 113 at 2.) Thus, to prove discriminatory decision-making in checkpoint detainment, the Government should have provided evidence allowing for a comparison of violating and non-violating Hispanics to violating and non-violating non-Hispanics who passed through the checkpoints. Dr. Banks' figure itself says nothing about why any person was detained.

It is possible, of course, that ACSO was stopping Hispanics discriminatorily. But it is at least equally possible that ACSO had reasonable suspicion or probable cause to detain proportionately more Hispanics than non-Hispanics without regard to ethnicity. Cf. Armstrong, 517 U.S. at 469–70, 116 S.Ct. 1480 (rejecting court of appeals' presumption that people of all races commit all types of crimes at equal rates). If true, drivers' ethnicity may merely correlate with chances of being detained without actually being a cause of detainment. It is unlikely anyone knows who in fact passed through a checkpoint without causing the production of a stop form, and the court recognizes that such information is not reasonably available to the Government. To show that ACSO discriminatorily detained Hispanics at checkpoints, however, the Government could have, for example, created a legitimate benchmark measuring the drivers as if passing through a checkpoint, categorizing and distinguishing them based on ethnicity and whether they were in fact violating the law. But the Government's benchmark study by Dr. Lamberth was fatally flawed. In the absence of some further proof, the court is unable to determine whether ACSO discriminately detained Hispanics more frequently than similarly situated non-Hispanics. Therefore, the Government's attempted use of Dr. Banks' figure—even considered against the entire record—is unpersuasive as evidence of discriminatory checkpoint stopping.

iii. Post–Stop Outcomes

The Government next contends that ACSO's post-stop decision-making evidences Fourteenth Amendment violations. Here, the Government relies principally on Dr. MacDonald's testimony, as well as on testimony that Sheriff Johnson ordered deputies to arrest Hispanics and the balance of the record. The court has already addressed the lay witness testimony, so the expert testimony will be examined below.

Dr. MacDonald examined the outcomes from ACSO traffic stops and opined that, when he controlled for the reason for the traffic stop (by way of regression analysis), Hispanics had higher arrest and citation rates but lower rates of written warnings, verbal warnings, or non-enforcement action. The Government argues that such "disparities" in post-stop outcomes between Hispanics and non-Hispanics proves discriminatory effect. (Doc. 158 at 90–94.) Whatever the superficial appeal of this argument, it fails under closer examination because the underlying analysis is too abstract and does not properly compare Hispanics to similarly situated non-Hispanics for what it claimed to measure. That is, the Government's post-stop outcome analysis fails to demonstrate a Fourteenth Amendment violation because it does not show that Hispanics were treated more harshly than non-Hispanics for similar offenses. See Armstrong, 517 U.S. at 465, 116 S.Ct. 1480 ("The requirements for a selective-prosecution claim draw on 'ordinary equal protection standards.' "); United States v. Timms, 664 F.3d 436, 447 (4th Cir.2012) (denying a Fourteenth Amendment claim because alleged dissimilar treatment was not between similarly situated persons); Williams v. Hansen, 326 F.3d 569, 576 (4th Cir.2003) (same).

"[I]n determining whether persons are similarly situated for equal protection purposes, a court must examine all relevant factors." United States v. Olvis, 97 F.3d 739, 744 (4th Cir.1996) ; see also Ah Sin v. Wittman, 198 U.S. 500, 507–08, 25 S.Ct. 756, 49 L.Ed. 1142 (1905) ( "Plaintiff in error seeks to set aside a criminal law of the state ... [on the ground] that it was made [unconstitutional] by the manner of its administration. This is a matter of proof; and no fact should be omitted to make it out completely, when the power of a Federal court is invoked to interfere with the course of criminal justice of a state." (emphasis added)). In this case, the Government claims to measure differences in stop outcome between Hispanics and non-Hispanics. But its evidence overlooks not just a relevant factor in ACSO's post-stop outcome decision-making, but the relevant factor—the basis for ACSO's decisions to arrest, cite, warn or take no action at all, which are the very actions the Government contends constitute the discriminatory policing. See Attorney Gen. of U.S. v. Irish People, Inc., 684 F.2d 928, 946 (D.C.Cir.1982) ("Discrimination cannot exist in a vacuum; it can be found only in the unequal treatment of people in similar circumstances.").

Most fundamentally, Dr. MacDonald's analysis requires the court to assume two major propositions: (1) similarity in generic stop reason means similarity in the severity of the conduct resulting in the stop; and (2) the stop reason (which Dr. MacDonald does not purport to measure) equates causally with the reason for the stop outcome (which he claims to measure). The trial evidence offers no proof of either of those assumptions, and Dr. MacDonald himself never attempted to say so at trial. Indeed, for several obvious reasons, these assumptions cannot be made. Some simple examples illustrate why.

First, "controlling for" one of the listed stop reasons noted on an ACSO stop form fails to account for the stop conduct—that is, the nature and degree of severity of the conduct resulting in the stop. Differences in the severity of stop conduct may likely explain differences in how the suspect is treated. For example, a common stop reason is "speed limit violation." (Gov't Trial Ex. 59.) However, a driver clocked at 70 m.p.h. in a 65 m.p.h. zone is much more likely to be warned, cited, or receive no action than a driver going 80 m.p.h., who is more likely to be cited or even arrested. See N.C. Gen. Stat. § 15A–401(b)(2) (stating that an officer can arrest an individual for any criminal offense without a warrant when the offense is committed in the presence of an officer, such as reckless driving, N.C. Gen.Stat. § 20–140 (defining reckless driving as operating a vehicle "carelessly and heedlessly in willful or wanton disregard of the rights or safety of others")). Similarly, other stop reasons provided on ACSO's stop forms, such as "safe movement violation" vary in degree and may therefore explain differences in post-stop outcomes. (See Gov't Trial Ex. 59.) Other stop reasons are too generic to assume they involve similar conduct and actually vary both in kind and in degree. An obvious example of the latter is "vehicle equipment violations." Compare N.C. Gen.Stat. § 20123.2 (requiring that vehicles contain a speedometer), with N.C. Gen.Stat. § 20–130.1 (prohibiting installation of red and blue lights on a vehicle). Thus, controlling for stop reasons generically, as Dr. MacDonald has done, provides little evidence that the persons stopped are in fact similarly situated.

Second, suppose two drivers, one non-Hispanic Caucasian and one Hispanic, are each stopped for speeding 10 m.p.h. over the speed limit. So far, (assuming all things else being equal, which they rarely are) the drivers are similarly situated based on the stop reason. Yet in conversing with the Caucasian driver, the officer smells alcohol and determines probable cause for driving while impaired.⁹⁰ So, the driver is arrested for that offense. However, the other officer who stopped the Hispanic driver observes that the driver was only speeding as a first offense and issues a verbal warning. Therefore, as to their actual law violations, the drivers are not similarly situated, which explains and justifies their dissimilar post-stop outcomes. This does not even take into account the offenders' prior criminal and traffic record, which may inform the officers' discretion. Dr. MacDonald's study ignores these distinctions between drivers in purporting to control for stop reason. Yet, when confronted with a comparable hypothetical at trial, Dr. MacDonald conceded that, in scenarios like the driving-while-impaired speeder, "the ultimate charge was not a consequence of the [stop reason]."⁹¹ (Doc. 155 at 43.)

90 The court uses this offense simply because Dr. MacDonald testified that a driver who was impaired is more likely to be arrested while a driver who was speeding is more likely to be cited. (Doc. 149 at 39.) Other reasons for driver arrests undoubtedly exist, but, because Dr. MacDonald's study omitted that information, no proper comparison can be made. Dr. MacDonald's failure to include the conduct underlying the ultimate outcome is particularly problematic as to his arrest comparison because, as he conceded, an arrest for mere speeding is unlikely.

91 Margo Frasier also opined that, when examining an officer's decision to arrest versus cite, one must consider the "offense[ ] for which the law allows either a citation or an arrest." (Doc. 151 at 27.)

The Government's evidence as to outcomes other than arrests and citations—written warnings, verbal warnings, and no action—fares no better when "controlled for" stop reason. While one might argue that the presence or absence of a warning might be more closely related to stop reason (although the Government provided no evidence this is so), the differences between Hispanics and non-Hispanics as to written warnings, verbal warnings, and no action could just as easily be driven by whether or not they were arrested or cited for similar conduct. That is, a decision to arrest or cite is likely made in lieu of issuing a warning. This is demonstrated by the fact that, if one were to look only at the post-stop outcomes for written warnings, verbal warnings, and no action, Dr. MacDonald's study would show that, at least as to these, Hispanics are better off as compared to non-Hispanics.

It is therefore unknown just how often (1) similarity in generic stop reason means sufficient similarity in the severity of the conduct resulting in the stop and (2) the stop reason is causally related to the basis for the stop outcome. This is particularly problematic in light of Whren v. United States, 517 U.S. 806, 116 S.Ct. 1769, 135 L.Ed.2d 89 (1996), which allows (nondiscriminatory) pretextual traffic stops. Without controlling for these obvious, nondiscriminatory reasons for post-stop outcomes, Dr. MacDonald's statistical evidence does not prove dissimilar treatment between Hispanics and similarly situated non-Hispanics as to stop outcome. See Armstrong, 517 U.S. at 465, 116 S.Ct. 1480 ; Timms, 664 F.3d at 447 ; Attorney Gen. of U.S., 684 F.2d at 946.

The Government must do more than it did here, citing superficial statistical disparities to show discriminatory effect. As the Seventh Circuit cautioned, "Of course, parties may not prove discrimination merely by providing the court with statistical analyses. The statistics proffered must address the crucial question of whether one class is being treated differently from another class that is otherwise similarly situated." Chavez, 251 F.3d at 638 ; see also Cordi–Allen v. Conlon, 494 F.3d 245, 251 (1st Cir.2007) ("[The similarly situated] requirement demands more than lip service.... It is inadequate merely to point to [disparities] in a vacuum and leave it to the [defendant] to disprove conclusory allegations that [persons] are similarly situated."); Olvis, 97 F.3d at 744 ("The goal of identifying a similarly situated class of law breakers is to isolate the factor allegedly subject to impermissible discrimination." (quoting United States v. Aguilar, 883 F.2d 662, 706 (9th Cir.1989) )); Moore v. City of Charlotte, N.C., 754 F.2d 1100, 1110 (4th Cir.1985) ("[A] conclusion that individuals have received discriminatory disparate treatment may be supported only where the notion of comparability is informed by sound, articulated principles ...."). By only citing differences in post-stop outcomes without also showing that Hispanics were treated dissimilarly from those similarly situated to them, the Government thus fails to demonstrate discriminatory effect under the Fourteenth Amendment. See Ah Sin, 198 U.S. at 507–08, 25 S.Ct. 756 (rejecting plaintiff's claim that ordinance was enforced "solely and exclusively against persons of the Chinese race" because it did not allege "that the conditions and practices to which the ordinance was directed did not exist exclusively among the Chinese, or that there were other offenders against the ordinance than the Chinese, as to whom it was not enforced"); United States v. Bass, 536 U.S. 862, 863–64, 122 S.Ct. 2389, 153 L.Ed.2d 769 (2002) ("[R]aw statistics regarding overall charges say nothing about charges brought against similarly situated defendants. " (emphasis added)); Hemsworth v. Quotesmith.com, Inc., 476 F.3d 487, 492 (7th Cir.2007) ("Statistical evidence is only helpful when the plaintiff faithfully compares one apple to another without being clouded by thoughts of Apple Pie ala Mode or Apple iPods."); Olvis, 97 F.3d at 745 ("Without an appropriate basis for comparison, raw data about the percentage of black crack cocaine defendants proves nothing. " (emphasis added)); Cooper v. S. Co., 260 F.Supp.2d 1305, 1317 (N.D.Ga.2003) ("While Plaintiff's expert reports an over-all statistical disparity between black and white employees in compensation, Plaintiff's evidence generally fails to compare similarly situated individuals, significantly diminishing the probative value of any disparity."), aff'd,390 F.3d 695 (11th Cir.2004), overruled on other grounds, Ash v. Tyson Foods, Inc., 546 U.S. 454, 457, 126 S.Ct. 1195, 163 L.Ed.2d 1053 (2006).

The court reaches this conclusion with the appreciation that the type of information necessary to determine whether persons are similarly situated is simply unavailable or practically difficult to obtain in some instances. See, e.g., Chavez, 251 F.3d at 639–40 (noting that, "[i]n a civil racial profiling case, however, the similarly situated requirement might be impossible to prove"). But see Armstrong, 517 U.S. at 466, 116 S.Ct. 1480 ("The similarly situated requirement does not make a selective-prosecution claim impossible to prove."). While the Government never made this argument, it is plain that this case is not one of them.

For one, there is evidence that Dr. MacDonald could have controlled for the differences in severity of driver stop conduct, but did not. While not documented on ACSO's stop forms and therefore not accounted for in that data, differences in severity of stop conduct are reflected in ACSO's citation data, at least as to speed. (See Gov't Trial Ex. 71 (observing on citation form that driver was going approximately 45 mph in a 35 mph speed zone).) The Government made no effort to control for those differences here. This failure is particularly important where the Government attempts to distinguish between gradations of outcome without considering gradations of conduct that are likely causally associated with outcome.

The Government also had access to ACSO's arrest and citation forms. Dr. MacDonald could therefore have matched ACSO's citation and arrests forms, which contain the reasons for citations and arrest, with ACSO's stop forms to account for differences between stop reason and the basis for the post-stop outcome.⁹² (Doc. 155 at 35–36 (Dr. MacDonald conceding that one "could try" to match the citation and arrest forms to the stop forms); see also Doc. 153 at 112, 165–66 (Dr. Banks testifying that he "hand-matched" traffic stop forms to arrest reports to determine the number of checkpoint arrests); Gov't Trial Ex. 71 (example citation form).) Dr. MacDonald testified, however, that he elected not to do so. (Doc. 155 at 42.) He acknowledged further that, without the reason for citations and arrests, "you're missing the context." (Id. at 36; see also id. at 41 (agreeing that, under his study, "you wouldn't know the type of arrest, charge, or the type of citation"); id. at 42 (admitting that, without providing the reason for citations and arrests, "a layer of contextual information" is missing).) As Dr. MacDonald candidly allowed, it is important to have the reasons for citations and arrests as "it's always better to have more data to try to explain the disparities, to see what's driving them. " (Id. at 35 (emphasis added).)

92 Importantly, this does not mean that the Government needed to match by hand all 20,059 traffic stop forms it analyzed. For example, while controlling for the issues noted, it could have instead matched a sufficiently sized sample of stop forms with citation or arrest data, like Dr. Banks did. (See Doc. 153 at 112–13, 165–66.)

Thus, by ignoring this evidence, Dr. MacDonald's analysis fails to preclude other equally plausible, non-discriminatory reasons for the inference the Government seeks to draw from his post-stop outcome figures. The thrust of the Government's claim is that ACSO's leadership instructs officers to arrest (rather than cite) Hispanics and bring them to jail (where they would be checked for immigration status in the 287(g) program and detained for deportation). (See Doc. 158 at 1–3, 10–14.) Consequently, if the theory held true, one might logically expect to find not only higher rates of arrest for Hispanics but lower rates of citations. Yet, Dr. MacDonald finds that Hispanic drivers were not only arrested but cited at rates higher than that of non-Hispanic drivers, and in significant numbers: 55.8% of Hispanics cited versus 32.0% of non-Hispanics. (Doc. 149 at 21–22.) Because Dr. MacDonald fails to examine any of the reasons for citations or arrests, as well as the differences in the stop conduct, he does not rule out (or at least make less plausible) benign motives. For example, were the Hispanics stopped for the same stop reason as non-Hispanics in fact similarly violating the law that formed the basis for their stop? Were the Hispanics arrested not in fact committing more serious crimes deserving of arrest, or were they committing lesser offenses than non-Hispanics? And, were Hispanics in reality only cited for lacking any driver's license (or when lacking any identification) or were they arrested?⁹³ Rather than rule out these plausible, non-discriminatory explanations, Dr. MacDonald's analysis instead requires the court to assume police misconduct was afoot.

93 Such questions are critical, even if the Government's theory is only that, by directing that Hispanics be arrested, ACSO necessarily cited more Hispanics because bases for their arrest did not exist.

The Government contends that its showing is sufficient to shift the burden to Sheriff Johnson to demonstrate that the disparities were caused by factors other than ethnicity. (See, e.g., Doc. 158 at 97–104.) For example, at trial and in its post-trial filing, the Government repeatedly attacked Sheriff Johnson's attempts to show that Hispanics and non-Hispanics are not similarly situated.⁹⁴ (See Doc. 155 at 54–55; Doc. 158 at 97–104.) In making these criticisms, however, the Government attempts to avoid its own obligation. Sheriff Johnson need not establish dissimilarities between Hispanics and non-Hispanics in Alamance County. The Government—not Sheriff Johnson—bears the burden of proving that ACSO treated Hispanics in a dissimilar fashion from similarly situated non-Hispanics. See Sansotta, 724 F.3d at 542 (describing the "burden" plaintiffs must carry to prove an equal protection claim); Cordi–Allen, 494 F.3d at 250 ("[C]ase law makes clear that the burdens of production and persuasion must be shouldered by the party asserting the equal protection violation.").

94 Similarly, Dr. MacDonald stated on several occasions that he did not "need" to look at the reason for citation and arrest for his study to show discriminatory effect. (See, e.g., Doc. 149 at 18.) Instead, Dr. MacDonald relied on his "belief" that the size of the disparities excused him from analyzing the reason for citations and arrests. (Doc. 155 at 34.)

In sum, Dr. MacDonald's post-stop outcome study fails to show discriminatory effect.

iv. Searches After Stops

The Government also presented a second statistical analysis by Dr. MacDonald assessing searches performed by ACSO officers as evidence of discriminatory effect. Dr. MacDonald's search analysis made three findings: (1) Hispanics were searched more frequently than non-Hispanics; (2) at checkpoints, searches of stopped Hispanics produced less drug and overall contraband (his checkpoint "hit rate"); and (3) searches of Hispanics yielded less drug contraband than searches of non-Hispanics (his overall "hit rate"). In sum, Dr. MacDonald opined that his analysis "suggest[s] that there is a different standard, a lower threshold of suspicion or probable cause [being applied] in searching Latinos compared to non-Latinos." (Doc. 149 at 48; see also id. at 42, 45.) The Government argues that, because Dr. MacDonald's analysis "suggest[s]" differences in search standards between Hispanics and non-Hispanics, it proves discriminatory effect. (Doc. 158 at 94–96.) The Government's conclusion, however, is unpersuasive.

An initial weakness with all three of Dr. MacDonald's search findings is that, although in possession of ACSO's search data, the Government neither alleged nor put on evidence of a single instance in which an ACSO officer failed to meet the requisite legal standard necessary to perform a search of any Hispanic. Such evidence would have suggested the application of a different search standard for Hispanics, and its omission stands in contrast to other cases identifying discriminatory policing.See, e.g., Floyd, 959 F.Supp.2d at 627–28, 636–37, 640–42, 652–55 (finding multiple instances of unconstitutional searches). Although the Government was certainly not required to cite such evidence to demonstrate discriminatory effect, its lack of this evidence burdens its suggested inference. See Int'l Bhd. of Teamsters, 431 U.S. at 339, 97 S.Ct. 1843 (observing that anecdotal evidence serves to bring "the cold numbers convincingly to life"); EEOC v. Joe's Stone Crab, Inc., 220 F.3d 1263, 1274 (11th Cir.2000) ("[I]n a pattern and practice case, the plaintiff must prove, normally through a combination of statistics and anecdotes, that discrimination is the [entity's] 'standard operating procedure.' " (quoting Int'l Bhd. of Teamsters, 431 U.S. at 335–36, 97 S.Ct. 1843 )); cf. Brown v. Nucor Corp., 785 F.3d 895, 914 (4th Cir.2015) (noting that "statistical and anecdotal evidence, especially when combined, thus provide precisely the 'glue' of commonality" necessary in the class certification context).

Second, according to the Government, these three findings "suggest" that ACSO uses a different standard when searching stopped Hispanics. A suggestion of different search standards, however, is insufficient to meet the Government's burden of proof. Sakaria v. Trans World Airlines, 8 F.3d 164, 172–73 (4th Cir.1993) ("In a long line of decisions in this circuit, we have emphasized that proof of causation must be such as to suggest 'probability' rather than mere 'possibility,' precisely to guard against raw speculation by the fact-finder." (citing cases)); see also Mckoy v. Charlotte–Mecklenburg Sch., No. 3:10CV494, 2011 WL 1869958, at *6 (W.D.N.C. May 16, 2011) ("The case law is clear in the Fourth Circuit that speculation and conjecture raise a mere possibility of discrimination rather than the reasonable probability which is necessary to support an inference of discrimination.").

Third, all three of Dr. MacDonald's search findings suffer from omitted variable bias—the failure to control for an important causal factor. (Doc. 153 at 132–33.) The Government attempts to use Dr. MacDonald's analysis to support the inference that ACSO applies a different search standard to Hispanics. In reaching these findings, however, Dr. MacDonald did not control for a critical variable—the type of search—that may easily explain search differences. This omission seriously undermines the probative value of those findings. See Reed Const. Data Inc. v. McGraw–Hill Cos., Inc., 49 F.Supp.3d 385, 403–04 (S.D.N.Y.2014) ("Omitted-variable problems—as the name suggests—arise when important control variables are left out of the model."). Without controlling for search type, the Government's attempt to demonstrate that ACSO discriminatorily searches Hispanics fails to show whether ACSO's search standards—or rather the type of search—is driving Dr. MacDonald's results. See People Who Care v. Rockford Bd. of Educ., Sch. Dist. No. 205, 111 F.3d 528, 537–38 (7th Cir.1997) ("[A] statistical study that fails to correct for salient explanatory variables, or even to make the most elementary comparisons, has no value as causal explanation."). The ACSO stop form lists several types of searches: consent; search warrant; probable cause; search incident to arrest; and protective frisk. (Gov't Trial Ex. 59.) While Dr. MacDonald's finding on drug contraband hit rate analysis controlled for one of these—searches incident to arrest—it failed to control for any other. Most problematic is the first among the search types—"consent." A consent search applies no search standard at all and merely requires an individual's voluntary consent. See United States v. Robertson, 736 F.3d 677, 679 (4th Cir.2013) ("[I]f an individual consents to a search, probable cause is unnecessary."). Consent may in fact explain a significant portion of searches. See United States v. Worley, 193 F.3d 380, 386 (6th Cir.1999) (describing a search conducted pursuant to voluntary and valid consent as "[o]ne of the most common" exceptions to the Fourth Amendment's warrant requirement); Ric Simmons, Not "Voluntary" but Still Reasonable: A New Paradigm for Understanding the Consent Searches Doctrine, 80 Ind. L.J. 773, 773 (2005) ("Over 90% of warrantless police searches are accomplished through the use of the consent exception to the Fourth Amendment."). Yet, Dr. MacDonald's analysis makes no distinction between a search based on consent and a search based on probable cause.⁹⁵

95 Dr. MacDonald presented an analysis controlling for searches incident to arrest only as to drug contraband hit rates. In an unsupported statement, he stated that he looked at "other indicators of contraband" and found a "consistent pattern of hit rates being significantly lower for Latinos compared to non-Latinos." (Doc. 149 at 44.) This conclusory statement is difficult to interpret. Without providing hit rates for other types of contraband, Dr. MacDonald's observed drug contraband hit rates may also be explainable by contraband type. For example, one would not expect "protective frisks"—a type of "search" identified on the stop form—to yield high numbers of drug contraband. See United States v. Hernandez–Mendez, 626 F.3d 203, 211 (4th Cir.2010) ("A frisk for weapons is permissible when an officer reasonably believes that the person being stopped 'may be armed and presently dangerous.' " (quoting Terry v. Ohio, 392 U.S. 1, 24, 88 S.Ct. 1868, 20 L.Ed.2d 889 (1968) )). Yet, Dr. MacDonald did not address this.

Moreover, while Dr. MacDonald controlled for searches incident to arrest as to overall drug contraband, his findings that (1) Hispanics were searched at a higher percentage than non-Hispanics and (2) searches of stopped Hispanics at checkpoints produced less drug and overall contraband both fail to do so. Searches conducted incident to (i.e., after) arrest are non-discretionary for ACSO officers. (Doc. 149 at 63–64; Doc. 154 at 21–22); see also United States v. Robinson, 414 U.S. 218, 235, 94 S.Ct. 467, 38 L.Ed.2d 427 (1973) ("A custodial arrest of a suspect based on probable cause is a reasonable intrusion under the Fourth Amendment; that intrusion being lawful, a search incident to the arrest requires no additional justification."). If an ACSO officer arrests an individual, he must search the individual for contraband and weapons under ACSO policy. Dr. MacDonald's overall search percentages and checkpoint search findings nevertheless included those uncontrolled-for, non-discretionary searches. Including those non-discretionary searches severely undermines the probative value of those two findings.⁹⁶In sum, beyond controlling for searches incident to arrest with respect to Dr. MacDonald's overall drug contraband finding, the Government makes no effort to address the effect of search type on the analysis. The Government's suggested inference that ACSO applies a different search standard to Hispanics, even in light of the Government's other evidence, is thus nothing more than that—a suggestion—and is too weak to show discriminatory effect by a preponderance of the evidence. See Bazemore v. Friday, 478 U.S. 385, 400, 106 S.Ct. 3000, 92 L.Ed.2d 315 (1986) (Brennan, J., concurring) ("[F]ailure to include variables will affect the analysis' probativeness."); see also EEOC v. Int'l Bus. Machs. Corp., 583 F.Supp. 875, 906 (D.Md.1984) ("Ignoring key variables may render the results of a multiple regression analysis useless.").

96 Notably, the Government cites Nicola Persico & Petra Todd, Generalising the Hit Rates Test for Racial Bias in Law Enforcement, with an Application to Vehicle Searches in Witchita, 116 Econ. J. F351 (Nov.2006), in support of its argument that ACSO conducts discriminatory searches. That article, however, observes,

A key assumption of the [hit rate] model ... is that police choose whom to search so as to maximize successful searches. Police presumably have little discretion in cases where they pull over a driver because they have a warrant for the driver's arrest or when the search is incident to an arrest. Therefore, we limit our analysis sample to observations on police-motorist encounters where police have discretion over whether to initiate the search.

Id. at F357 (emphasis added). The Government's cited article thus bolsters the conclusion that Dr. MacDonald's failure to control for search incident to arrest undermines his study's probative value.

The Government's proposed inference is further undermined by the testimony of Officer Dockery. Officer Dockery testified that, once searches incident to arrest are removed from the calculus, little disparity existed between the percentage of Hispanics and non-Hispanics searched. (Doc. 153 at 53–58.) In other words, once the analysis focuses only on searches where deputies have discretion to search, the purported blanket disparity between Hispanics and non-Hispanics that Dr. MacDonald urges becomes much more muddled and complicated, with the disparities now running both ways depending on the stop reason.

Officer Dockery's testimony that ACSO searched Hispanics and non-Hispanics in approximately equal percentages, moreover, suggests that Dr. MacDonald's search findings are an even more attenuated attempt to show discriminatory effect than his post-stop outcome study. For example, Officer Dockery testified that 11.42% of stopped Hispanics were searched for investigatory reasons, as compared to 16.46% of stopped non-Hispanics searched for the same reason. (Id. at 55.) And, Officer Dockery's testimony regarding searches at checkpoints is directly at odds with Dr. MacDonald's checkpoint hit-rate findings. Officer Dockery found that, after removing searches incident to arrest, Hispanics were searched less often than non-Hispanics, and by a factor of more than 4 to 1 (5.18% of stopped Hispanics versus 22.27% of stopped non-Hispanics). (Id. at 54.) The Government has thus not persuasively shown that ACSO conducts discretionary searches of Hispanics and non-Hispanics in different percentages, yet it seeks to argue that ACSO uses different search standards based on ethnicity.

The Government's suggested inference thus requires the court to make a significant, unsupported assumption. That is, even assuming that the type of search had been controlled for, the Government asks the court to accept the subtle proposition that ACSO discretionarily searches Hispanics and non-Hispanics in apparently equal proportions but actually applies different search standards to the two groups, such that ACSO should really search Hispanics relatively less frequently than non-Hispanics. Yet, the Government provided no reason why ACSO should be searching Hispanics less often than non-Hispanics. And again, while not required to do so to demonstrate discriminatory effect, the Government produced no instance in which an ACSO officer failed to meet the requisite legal standard when performing a search of a Hispanic person, which would have suggested the differing application of search standards. Cf. Floyd, 959 F.Supp.2d at 627–28, 636–37, 640–42, 652–55 (finding multiple instances of unconstitutional searches). Officer Dockery's testimony thus reinforces the conclusion that Dr. MacDonald's search findings lack probative value as to the existence of discriminatory effect resulting from ACSO's search practices.

For all these reasons, the Government has not carried its burden of proof that ACSO engages in discriminatory law enforcement in post-stop searches.

v. 287(g) Practices

The evidence presented at trial also emphasized the role of ACSO's 287(g) MOA with ICE. The Government emphasized three specific aspects of the 287(g) program as evidence of unlawful discrimination of Hispanics.

As an initial matter, it is not clear whether a case or controversy remains as to this claim. The 287(g) program seems to have been a driving force for the Government's case, yet all agree that ACSO's 287(g) authority was revoked in 2012. Therefore, there is no realistic probability that ACSO would discriminate unlawfully in any manner involving any 287(g) authority. See Clapper v. Amnesty Int'l USA, ––– U.S. ––––, 133 S.Ct. 1138, 1147, 185 L.Ed.2d 264 (2013) (holding that "threatened injury must be certainly impending to constitute injury in fact" and that "allegations of possible future injury" are not sufficient, (quoting Whitmore v. Arkansas, 495 U.S. 149, 158, 110 S.Ct. 1717, 109 L.Ed.2d 135 (1990) ) (emphasis, internal quotation marks, and brackets omitted)); Babbitt v. United Farm Workers Nat. Union, 442 U.S. 289, 298, 99 S.Ct. 2301, 60 L.Ed.2d 895 (1979) (holding that plaintiff must show "realistic danger of sustaining direct injury as a result of [a] statute's operation or enforcement"). ACSO raised this argument pre-trial (Doc. 87 at 18 (citing Doc. 11 at 7–8)), but the court declined to dismiss that part of the Government's claim on the grounds that ACSO's involvement in 287(g) was nevertheless admissible at trial as to ACSO's discriminatory intent and conduct. United States v. Johnson, 28 F.Supp.3d 499, 506–07 (M.D.N.C.2014). With this in mind, the court turns to the specific allegations and proof.

First, the 287(g) program's TFO position and investigations were cited as discriminatory. The impetus for the TFO position itself, however, came from ICE, not ACSO. ICE also told Sheriff Johnson that his deputies could serve as TFOs under the MOA. Until an ICE audit, neither ACSO nor ICE officials questioned ACSO's authority to retain a TFO position under the MOA. One former ICE agent believed "it was purely an ICE issue as far as ICE," adding, "Nobody did this on purpose. ICE was giving conflicting information; and once it was discovered in the management audit, as it was supposed to, it was fixed." (Doc. 152 at 166.) The TFO position was not created for a discriminatory purpose.

The Government also argues that several investigations carried out by the TFO provide evidence of Fourteenth Amendment violations. None of those investigations, however, demonstrates an intent to discriminate on the part of ACSO. Many were instigated by or organized outside of ACSO. For example, ICE and the DEA organized and coordinated the gang roundups in the County. (Doc. 147 at 73, 75; Doc. 149 at 174; Doc. 151 at 139–40; see also Doc. 152 at 102–03.) There is no persuasive indication that the roundups targeted anyone based on ethnicity. Cf. Melendres, 989 F.Supp.2d at 895–905 (enjoining policy using ethnicity as factor for roundups). The investigation of Marxavi Angel–Martinez occurred only after an Alamance County employee reported to ACSO that a Hispanic employee at the Alamance County Library was receiving maternity benefits and food stamps illegally through another person's Social Security number. (Doc. 149 at 180, 202; Doc. 154 at 86.) The investigation of newly-hired County employees followed a request from Alamance County Manager, David Smith, who forwarded ACSO a specific list of 17 County employees. (Doc. 149 at 182–83; Doc. 154 at 26–28.) Only after a citizen complaint of identity theft did ACSO investigate and eventually arrest Juan Ariano Vazquez for illegally using a citizen's Social Security number. (Doc. 149 at 183–85.) Lastly, the very limited investigation of a Hispanic man following a traffic accident (assuming this even involved 287(g) authority) was entirely orchestrated by Commissioner Ann Vaughn, who "rais[ed] Cain" about an accident in which she was involved. (Doc. 154 at 76.)

The second 287(g) practice noted at trial were those investigations connected to background checks for gun permits. ACSO's processing of gun permit applications requires review by an ACSO clerk, ACSO captains, and then approval by an ACSO major. (Id. at 38–39.) While the 287(g) program was in place, Majors Brown and Holland referred some unknown number of names to Lieutenant Denham for criminal history checks to fulfill ACSO's legal responsibility for issuing gun permits. See N.C. Gen.Stat. § 14–404. The Government, however, did not call Major Holland and, when it called Major Brown as a witness, failed to ask a single question about the gun permit background checks. Moreover, while the Government points to evidence that all names submitted for background checks through 287(g)'s database sounded Hispanic, there was no evidence as to what ACSO did to investigate gun applicants and whether ACSO failed to satisfy itself as to the legal status of only those applicants it referred to ICE—which would have explained the referrals. Finally, and perhaps most importantly, neither party addresses what, if any, adverse effect these additional background checks had on gun permit applicants. The Government produced no witness testifying to effect, such as the delay in the processing or a denial of a gun permit application. The court accepts this, therefore, as some evidence of discriminatory intent, but without more, it is uncertain as to its import and thus weak proof of a Fourteenth Amendment violation.

The third aspect of the 287(g) program the Government charges as discriminatory is ACSO's booking of individuals into the ACDC. The Government's allegation and contention was that ACSO targeted Hispanics for arrest so that they could run Hispanic arrestees through the 287(g) program. The evidence at trial, however, did not bear this contention out. For one, while the Government claims that Sheriff Johnson changed his arrest policy after implementation of the 287(g) program to effectuate this effort, the arrest policy changed well before the 287(g) program began. (Doc. 147 at 33; Doc. 154 at 58–59.) Moreover, Dr. MacDonald's evidence indicates that, following 87.9% of stops, Hispanics were not arrested, undercutting the Government's argument that ACSO targeted Hispanics for arrest to run them through the 287(g) program.⁹⁷ Finally, as to ACSO's after-arrest conduct, there was no evidence as to how many (or few) Hispanics arrested by ACSO were booked into the ACDC and subjected to 287(g) questioning—figures almost certainly available to the Government. ACSO's 287(g) officers in fact processed only about one detainee per week at ACSO under the 287(g) program, "the bulk" of whom actually came from other ICE offices and/or were arrested by non-ACSO officers. (Doc. 152 at 163–64.) Thus, ACSO officers contributed only infrequently to the number of ICE detainees housed in the ACDC, further discrediting the Government's theory. To the extent the Government argues that the Sheriff had a financial motive to arrest Hispanics, therefore, the trial evidence revealed that the real financial motive for using the ACDC for a 287(g) program was the revenue to be received from housing detainees brought in by ICE and the eleven other law enforcement agencies who were using the facility.

97 Because the Government failed to provide similarly situated data or other evidence of improper arrests, there is no comparator or basis for concluding that the Hispanics arrested were subject to unlawful policing.

The court does not find the evidence related to ACSO's 287(g) program to establish a pattern or practice of Fourteenth Amendment violations.

vi. ACSO's Culture, Supervision, and Discipline

The Government also highlights evidence connected to ACSO's culture, supervision, and discipline. That evidence appears to be directed toward showing ACSO's discriminatory intent, and the Government makes no argument that a lack of discipline, deficient supervision, or the use of racially- or ethnically-charged language demonstrates a discriminatory effect.

At least as to discriminatory language, courts have held that such language, while undoubtedly inexcusable, does not constitute a Fourteenth Amendment violation. See DeWalt v. Carter, 224 F.3d 607, 612 (7th Cir.2000) ("The use of racially derogatory language, while unprofessional and deplorable, does not violate the Constitution."); Williams v. Bramer, 180 F.3d 699, 706 (5th Cir.) ("We hold today that an officer's use of a racial epithet, without harassment or some other conduct that deprives the victim of established rights, does not amount to an equal protection violation."), decision clarified on reh'g, 186 F.3d 633 (5th Cir.1999) (per curiam). The evidence in this case thus falls short of demonstrating a constitutional problem, as none of that evidence was tied to any law enforcement action violating a constitutional right.

While this evidence falls short of showing a Fourteenth Amendment violation, the court would nevertheless be remiss if it failed to address the troubling nature of some of the evidence at trial. First, the language, epithets, and slurs used by some ACSO officers, particularly in the ACDC, are abhorrent and, if not ended already, should cease immediately. While most of that language was used outside the presence of the individuals to whom the language referred, that does not excuse the unprofessional and wholly inappropriate nature of that language. The court recognizes that the evidence showed that only a select number of officers used that language, and it was confined largely to the ACDC. Nevertheless, there is simply no place for the use of that type of language by government officials tasked with lawfully and justly enforcing the law.

That admonishment goes for the emailed items as well. For a department that likely had tens of thousands of emails or more, the handful of those presented were not overwhelming. However, the jokes and video game sent by ACSO officers are reprehensible, and a sheriff's office is no place for such behavior. The court notes ACSO's recent efforts to curb the abuse of its email system. If not prevented already, however, the sending of racially- and ethnically-insensitive jokes and games must stop. Finally, the evidence at trial highlighted inconsistencies in ACSO's discipline. Some conduct violating ACSO's policies was disciplined, while other conduct was not. A competent, efficient, and professional functioning law enforcement organization requires consistent, regular discipline. The type of language, epithets, slurs, and emails presented at trial should be addressed in a more timely, reliable, and diligent manner.

In the context of this case, the court can only address conduct that rises to the level of a pattern or practice of an Equal Protection violation. See Lewis v. Casey, 518 U.S. 343, 358, 116 S.Ct. 2174, 135 L.Ed.2d 606 (1996) ( "It is the role of courts to provide relief to claimants, in individual or class actions, who have suffered, or will immediately suffer, actual harm; it is not the role of courts, but that of the political branches, to shape the institutions of government in such fashion as to comply with the laws and Constitution."). As the court finds that the Government has failed to demonstrate such a claim, the court's power and duty ends.

2. Fourth Amendment Claim

The Government's second claim under 42 U.S.C. § 14141 alleges a pattern or practice of discriminatory law enforcement on the part of ACSO against Hispanics in violation of the Fourth Amendment. For reasons explained in its summary judgment Order, the court ruled that the Government's complaint alleged a Fourth Amendment claim only insofar as it related to Hispanics and failed to give Sheriff Johnson fair notice of allegations that ACSO used checkpoints generally for other unlawful purposes against all persons in the County. (See Doc. 1 ¶¶ 5, 25, 30, 41–46; Doc. 118 at 31–36 ("[A]llowing the Government to proceed as to claims of generalized unlawful checkpoints untethered to the abiding central claim of this case would unduly prejudice Johnson.").) The court therefore concluded that "the complaint raises a proper Fourth Amendment challenge to the extent it contends that the ACSO, as part of its alleged targeting of Latinos, has conducted checkpoints with a programmatic purpose that violates the Fourth Amendment." (Doc. 118 at 35.)

The Government contends that "ACSO conducts checkpoints for general law enforcement purposes in violation of the Fourth Amendment, and that there is a nexus between these checkpoints and ACSO's targeting of Latinos." (Doc. 158 at 140.) The question before this court, therefore, is whether—through its checkpoint operations—ACSO has engaged in a pattern or practice of Fourth Amendment violations involving Hispanics. To answer that question, this court looks to the Supreme Court's decision in City of Indianapolis v. Edmond, 531 U.S. 32, 121 S.Ct. 447, 148 L.Ed.2d 333 (2000), and its progeny for guidance.

The Fourth Amendment protects "against unreasonable searches and seizures." U.S. Const. amend. IV. This protection "generally bars officials from undertaking a search or seizure absent individualized suspicion." Chandler v. Miller, 520 U.S. 305, 308, 117 S.Ct. 1295, 137 L.Ed.2d 513 (1997). While "some quantum of individualized suspicion is usually a prerequisite to a constitutional search or seizure," the Fourth Amendment does not impose an "irreducible requirement of such suspicion." United States v. Martinez–Fuerte, 428 U.S. 543, 560–61, 96 S.Ct. 3074, 49 L.Ed.2d 1116 (1976). Rather, in "certain limited circumstances," the absence of individualized suspicion may not doom a search or seizure. Nat'l Treasury Emps. Union v. Von Raab, 489 U.S. 656, 668, 109 S.Ct. 1384, 103 L.Ed.2d 685 (1989) ; see also Martinez–Fuerte, 428 U.S. at 560–61, 96 S.Ct. 3074. Law enforcement checkpoints sometimes fall within those limited circumstances permitting a lack of individualized suspicion. See Mich. Dep't of State Police v. Sitz, 496 U.S. 444, 455, 110 S.Ct. 2481, 110 L.Ed.2d 412 (1990) (holding a sobriety checkpoint valid under the Fourth Amendment); Martinez–Fuerte, 428 U.S. at 560–64, 96 S.Ct. 3074 (upholding a "brief" border patrol checkpoint). Not every law enforcement checkpoint, however, necessarily withstands Fourth Amendment scrutiny. Under the Supreme Court's decision in Edmond, courts should inquire as to a checkpoint's "primary" programmatic purpose to determine its constitutional validity. 531 U.S. at 45–46, 121 S.Ct. 447 ("[P]rogrammatic purposes may be relevant to the validity of Fourth Amendment intrusions undertaken pursuant to a general scheme without individualized suspicion."); see also Hunsberger v. Wood, 570 F.3d 546, 554 (4th Cir.2009) ("[W]hen analyzing a search made as the result of a routine police procedure, ... the court should examine the programmatic purpose of the policy.").

When directing courts to ascertain the primary purpose behind checkpoints, the Supreme Court "recognize[d] the challenges inherent in a purpose inquiry." Edmond, 531 U.S. at 46–47, 121 S.Ct. 447. Because the relevant law enforcement agency in Edmond stipulated to the primary purpose behind its checkpoint program, however, the Court provided only limited direction on how lower courts should undertake the primary-purpose analysis for law enforcement checkpoints. See id. (noting confidence in courts' ability to make the primary-purpose determination and stating "courts routinely engage in this enterprise in many areas of constitutional jurisprudence as a means of sifting abusive governmental conduct from that which is lawful"). The Court observed that the primary-purpose inquiry requires examination of "the available evidence," but that such an examination was "not an invitation to probe the minds of individual officers acting at the scene." Id. at 46, 48, 121 S.Ct. 447 ; see also United States v. Gonsalves, 435 F.3d 64, 69 (1st Cir.2006) (upholding administrative search under Edmond 's primary-purpose test, regardless of government official's subjective intent). Ultimately, the Court held that, because the stipulated primary purpose of the relevant checkpoints was "the general interest in crime control" (i.e., drug interdiction), "the checkpoints violate[d] the Fourth Amendment." Edmond, 531 U.S. at 48, 121 S.Ct. 447.

Cases following Edmond have since further elucidated the contours of the primary-purpose test. Specifically, the Supreme Court has held that the primary-purpose inquiry focuses on the "immediate objective" of a program, not the program's ultimate goal. Ferguson v. City of Charleston, 532 U.S. 67, 83–84, 121 S.Ct. 1281, 149 L.Ed.2d 205 (2001). In Ferguson, the Supreme Court applied Edmond to determine the primary purpose of a state hospital's program of drug testing pregnant women. Id. at 70–71, 121 S.Ct. 1281. "In looking to the programmatic purpose," the Court explained, "we consider all the available evidence in order to determine the relevant primary purpose." Id. at 81, 121 S.Ct. 1281 (citing Edmond, 531 U.S. at 45–47, 121 S.Ct. 447 ). The Ferguson Court thus examined "the document codifying the policy" as well as "the development and application of the policy" to determine its primary purpose. Id. at 81–86, 121 S.Ct. 1281. Importantly, the Court distinguished between the "ultimate" and "immediate" purposes of a program. Id. at 83–84, 121 S.Ct. 1281. For the Edmond inquiry, a program's "immediate objective," not its ultimate goal, is the pertinent measure of a program's primary purpose. Id. ("While the ultimate goal of the program may well have been to get the women in question into substance abuse treatment and off of drugs, the immediate objective of the searches was to generate evidence for law enforcement purposes in order to reach that goal."); see also United States v. Davis, 270 F.3d 977, 982 (D.C.Cir.2001) ("One must be careful not to fall into the trap of thinking that any 'but for' cause of a roadblock represents its primary purpose within Edmond's meaning."). The Ferguson Court thus concluded that, because the hospital policy's "immediate objective" was unconstitutional, the policy violated the Fourth Amendment. 532 U.S. at 84–86, 121 S.Ct. 1281.

Here, the Government cites no case supporting its contention that a checkpoint with the programmatic purpose of targeting members of a particular racial or ethnic group may constitute a Fourth Amendment violation as opposed to a Fourteenth Amendment violation. Nevertheless, assuming such a programmatic purpose could constitute a Fourth Amendment violation, the court concludes—based on its Findings of Fact—that the purpose of ACSO's checkpoint operations, primary or otherwise, was not to target Hispanics.

The testimony of multiple officers inside and outside ACSO demonstrates that ACSO routinely conducted its checkpoints with lawful primary purposes, mainly for checking motor vehicle violations. For one, ACSO's checkpoint policy describes the three types of permissible checkpoints. (Gov't Trial Ex. 113 at 2–4 (listing standard (motor vehicle), informational, and special operations as available checkpoint options)). The "standard" checkpoint is set up for the purpose of determining "compliance with motor vehicle laws." (Id. at 2.) As the policy states, "Examples for which this [standard] ... checking station may be used include verification of drivers' licenses, vehicle registration checks, insurance checks, seat belt compliance checks, and driving while impaired checks." (Id. ) "Informational" checkpoints function to seek information regarding a crime, and "special operations" checkpoints focus on the immediate apprehension of a suspect. (Id. at 3.)

The evidence showed that ACSO occasionally conducted checkpoints seeking information; Sheriff Johnson described such checkpoints as attempts "to get information on" a crime. (Doc. 154 at 149–50.) Chief Deputy Britt also recalled an informational checkpoint occurring after a series of break-ins. (Id. at 24–25.) Other ACSO officers also confirmed they have on occasion conducted informational checkpoints. (See, e.g., Doc. 149 at 108–09; Doc. 150 at 84 (noting that informational checkpoints occurred but were "less common"); id. at 89–90.)

Little testimony was provided on special operations checkpoints, which focused on specific events, such as capturing someone "escaped from jail" or when "somebody was on the run in the community." (Doc. 154 at 105; see also id. at 150.) In fact, no ACSO officer stated that they had conducted such checkpoints.

The vast majority of ACSO's checkpoints operated as standard checkpoints with the primary purpose of checking motor vehicle violations. (Doc. 150 at 84.) Most officers testified to their performance of standard checkpoints. Lieutenant Hoover testified that that ACSO's checkpoints are "generally ... for your ... driver's license or registration or insurance." (Id. at 9.) He further stated, "The purpose of a checkpoint, [is] to check [the] validity of driver's license, registration, insurance." (Id. at 27–28; see also id. at 10 ("The checkpoint would be set up for driver's license, registrations, or insurance.").) Officer Anthony similarly testified that ACSO primarily conducted motor vehicle compliance checkpoints. (Doc. 148 at 22– 23; see also Doc. 147 at 204–05 (Officer Anthony stating that, in addition to checking for valid driver's licenses, ACSO would also conduct drunk driving checkpoints).) Officer Culler—a state highway patrol officer—recalled performing only "traffic check, drivers check" checkpoints in conjunction with ACSO. (Doc. 152 at 171–72.) Deputy Conklin reiterated that the "purpose of setting up" standard checkpoints is "motor vehicle violations." (Doc. 150 at 82; see also id. at 90 ("We check driver's license and registrations.... We don't set them up specifically for drugs. If we stop a vehicle and either smell or see something, that would indicate to us we move forward.").) Finally, Sergeant Crain testified that the purpose of ACSO's checkpoints was to "ensur[e] compliance with North Carolina motor vehicle law primarily." (Doc. 151 at 112–13.)

This overwhelming account of ACSO's checkpoint practices demonstrates that the primary purposes of ACSO's checkpoints were routinely lawful, with the vast majority of checkpoints ensuring compliance with North Carolina's motor vehicle laws, specifically checking for licenses, vehicle registrations, and proof of insurance.

Under existing Fourth Circuit case law, such checkpoints are constitutionally valid. See United States v. Brugal, 209 F.3d 353, 357 (4th Cir.2000) ; see also Edmond, 531 U.S. at 37–38, 121 S.Ct. 447 (recognizing that Supreme Court precedent at least "suggested" that checkpoints "with the purpose of verifying drivers' licenses and vehicle registrations would be permissible"); Delaware v. Prouse, 440 U.S. 648, 663, 99 S.Ct. 1391, 59 L.Ed.2d 660 (1979) (indicating that "[q]uestioning of all oncoming traffic at roadblock-type stops" to check for valid driver's licenses would be legitimate under the Fourth Amendment); United States v. Bernacet, 724 F.3d 269, 273 (2d Cir.) (finding inspection for driver's license, vehicle registration, and insurance information at license checkpoint within bounds of the Fourth Amendment), cert. denied, ––– U.S. ––––, 134 S.Ct. 806, 187 L.Ed.2d 610 (2013) ; United States v. Galindo–Gonzales, 142 F.3d 1217, 1221 (10th Cir.1998) (same); United States v. McFayden, 865 F.2d 1306, 1310–12 (D.C.Cir.1989) (same), abrogated in part by Davis, 270 F.3d at 981.

Similarly, checkpoints operated with the primary purpose of obtaining information about a crime are constitutionally valid. See Illinois v. Lidster, 540 U.S. 419, 423, 124 S.Ct. 885, 157 L.Ed.2d 843 (2004) ("The stop's primary law enforcement purpose was not to determine whether a vehicle's occupants were committing a crime, but to ask vehicle occupants, as members of the public, for their help in providing information about a crime in all likelihood committed by others."); United States v. Faulkner, 450 F.3d 466, 472–74 (9th Cir.2006) (upholding operation of an informational checkpoint). Similarly, and although no testimony was offered that ACSO conducted a special operations checkpoint, those checkpoints appear to be constitutional as well. See Edmond, 531 U.S. at 44, 121 S.Ct. 447 ("[T]he Fourth Amendment would almost certainly permit an appropriately tailored roadblock set up ... to catch a dangerous criminal who is likely to flee by way of a particular route."); United States v. Whitehead, 567 Fed.Appx. 758, 767 (11th Cir.), cert. denied, ––– U.S. ––––, 135 S.Ct. 308, 190 L.Ed.2d 223 (2014).

The Government's evidence to support its contention that checkpoints were operated with the primary purpose of targeting Hispanics was very limited, and the little that exists is speculative at best. As evidence of a purpose to target or discriminate against Hispanics in ACSO's checkpoint operations, the Government cites Dr. Banks' study on checkpoint arrests and Dr. MacDonald's hit rate study, statements by ACSO officers about the siting of checkpoints, and two specific checkpoints.⁹⁸ (See Doc. 158 at 140–42.)

98 Although not cited by the Government in its briefing as evidence of checkpoint targeting of Hispanics, there are two other alleged incidents involving ACSO checkpoints. First, according to some witnesses, Sheriff Johnson directed ACSO officers to "arrest Hispanics" at a checkpoint near a predominantly Hispanic mobile home park. (Doc. 158 at 11, 121.) Second, a witness testified that an ACSO officer misinterpreted an order by Sheriff Johnson at a checkpoint but, upon clarification, the Sheriff told officers to enforce the law in an equal manner. (Doc. 147 at 156–59.) As observed previously, those two incidents evince no purpose to target Hispanics.

First, the checkpoint studies of Drs. Banks and MacDonald are unpersuasive to show a purpose to target Hispanics at ACSO checkpoints. For the reasons already noted, the Government's use of Dr. Banks' finding that 36.8% of ACSO's stops at checkpoints were of Hispanics is unreliable. Dr. MacDonald's study—finding that searches of stopped Hispanics at checkpoints produced less drugs and overall contraband—failed to show that ACSO applied different search standards based on race/ethnicity, in large part because it suffered from omitted variable bias. Those studies show no purpose to target Hispanics at ACSO checkpoints.

Second, the Government cites statements that ACSO officers conducted checkpoints near predominately Hispanic areas. ACSO undoubtedly set up some checkpoints near predominately Hispanic areas. (Doc. 147 at 213; Doc. 149 at 217.) The placement of a few of ACSO's 435 checkpoints over the years fails to demonstrate that those checkpoints were set up for the purpose, primary or otherwise, of targeting Hispanics. Moreover, Dr. Banks' permutation study indicated that little "evidence [existed] that checkpoints were being sited closer to Hispanic communities than would have occurred if they were done just at chance." (Doc. 153 at 94–95, 99.) The trial evidence demonstrated that ACSO administered checkpoints all over the County.

The first incident the Government cites similarly fails to evidence a purpose, primary or otherwise, of targeting Hispanics. The Government cites Corporal Nicholson's operation of a checkpoint near Rocky Top mobile home park—a park in which residents are predominantly Hispanic. As noted, however, the evidence at trial demonstrated that Corporal Nicholson conducted the checkpoint as a measure to show law enforcement presence. He did so in response to Sheriff Johnson's unspecified direction that came following the park owner's complaint about a specific Mexican gang then under ACSO investigation for criminal activity and thought to be operating in the owner's park.

Second, the Government cites the checkpoint interaction between Deputy Keller and Crotts. That interaction involved a checkpoint at which Deputy Keller (who knew Crotts) told Crotts' husband that she did not need to see his license because "they were there to get them some," gesturing in the direction of a mobile home park predominantly populated by Hispanics. As noted, the court finds Deputy Keller's statement and conduct vague and Crotts' assumption that it referred to Hispanics speculative. (See Doc. 149 at 120, 125.) Crotts admitted as much on cross-examination. (Id. at 125 ("I don't know what [Deputy Keller] meant.").) Moreover, even assuming that Deputy Keller's statement referred to Hispanics generally (as opposed to perhaps certain suspects, if an informational checkpoint, or other motor vehicle law violators, if a standard checkpoint), Deputy Keller's statement appears to be outside of Edmond's programmatic purpose inquiry, which "is not an invitation to probe the minds of individual officers acting at the scene." Edmond, 531 U.S. at 48, 121 S.Ct. 447.

In sum, based on the totality of the circumstances considering all the evidence adduced at trial, the Government lacks proof that the purposes of ACSO's checkpoint operations, primary or otherwise, were to target Hispanics, and it certainly fails to show a pattern or practice of the same. See Int'l Bhd. of Teamsters, 431 U.S. at 336 & n. 16, 97 S.Ct. 1843 (holding that a plaintiff must "prove more than the mere occurrence of isolated ... or sporadic discriminatory acts" but rather that the classifications were "standard operating procedure [-] the regular rather than the unusual practice"); cf. Monell, 436 U.S. at 690–91, 98 S.Ct. 2018 (requiring proof of "persistent and widespread discriminatory practices of state officials" to demonstrate a governmental "custom" of constitutional rights deprivations under § 1983 ).

The Government, however, further argues that ACSO conducts checkpoints for reasons other than the purposes provided. (See Doc. 158 at 140–46; see also Doc. 1 ¶ 30.) In light of the testimony by ACSO's officers, the Government's claim essentially boils down to a contention that ACSO conducted pretextual checkpoints for the purpose of targeting Hispanics.

Although the Supreme Court has never explicitly so held, it has certainly implied that pretextual checkpoints would contravene the Fourth Amendment. In Texas v. Brown, 460 U.S. 730, 103 S.Ct. 1535, 75 L.Ed.2d 502 (1983), the Court suggested that checkpoints with a pretextual lawful purpose that are in fact for an unlawful purpose are constitutionally prohibited. Id. at 743–44, 103 S.Ct. 1535 ; see also Whren, 517 U.S. at 811–12, 116 S.Ct. 1769 ("[T]he exemption from the need for probable cause (and warrant), which is accorded to searches made for the purpose of inventory or administrative regulation, is not accorded to searches that are not made for those purposes."); New York v. Burger, 482 U.S. 691, 716 n. 27, 107 S.Ct. 2636, 96 L.Ed.2d 601 (1987) ; Brugal, 209 F.3d at 357 (citing Brown ). Prior to Edmond, which set out the primary-purpose test, several courts had held unconstitutional checkpoints that operated for pretextual lawful purposes but with actual unlawful purposes. See United States v. Huguenin, 154 F.3d 547, 558–59 (6th Cir.1998) ; United States v. Morales–Zamora, 974 F.2d 149, 152–53 (10th Cir.1992) ; State v. DeBooy, 996 P.2d 546, 551 (Utah 2000). Since Edmond, lower federal courts have continued to intimate that checkpoints set up for pretextual purposes are unconstitutional under the Fourth Amendment. See United States v. Knight, 306 F.3d 534, 537 (8th Cir.2002) ("We observe, finally, that the Supreme Court has warned against administrative stops becoming pretexts for 'crime control.' " (quoting Edmond, 531 U.S. at 40, 121 S.Ct. 447 )); United States v. Gabriel, 405 F.Supp.2d 50, 61 (D.Me.2005) ("There is no evidence the Border Patrol was using terrorism as a pretext to operate a checkpoint otherwise forbidden by Edmond. "); see also State v. Hicks, 55 S.W.3d 515, 537–38 (Tenn.2001). Nevertheless, because—as noted earlier in this section—the evidence at trial does not support a finding of pretextual purposes targeting or discriminating against Hispanics as part of ACSO's checkpoint operations, this court need not decide whether pretextual checkpoints ultimately contravene the Fourth Amendment.

Left unclear by the Supreme Court is whether a checkpoint with an illegitimate "secondary purpose" violates the Fourth Amendment. Edmond left that question unanswered but in a footnote observed, "Because petitioners concede that the primary purpose of the Indianapolis checkpoints is narcotics detection, we need not decide whether the State may establish a checkpoint program with the primary purpose of checking licenses or driver sobriety and a secondary purpose of interdicting narcotics." Edmond, 531 U.S. at 47 n. 2, 121 S.Ct. 447 ; see also Anobile v. Pelligrino, 303 F.3d 107, 122 (2d Cir.2002) ("The Supreme Court has not directly addressed the constitutionality of searches subject to a 'mixed purpose,' where the primary purpose is legitimate and the secondary purpose is not.").

Only a handful of lower courts have analyzed the issue of checkpoints with secondary purposes post-Edmond. Two circuit courts addressing the issue—the Fifth and D.C. Circuits—have concluded that Edmond tolerates checkpoints having a secondary, but invalid, purpose. See United States v. Moreno–Vargas, 315 F.3d 489, 491 (5th Cir.2002) ("We accordingly hold that Moreno's immigration stop at the Sarita checkpoint was valid because the checkpoint has as its primary programmatic purpose the enforcement of the immigration laws, regardless of whether or not it could also be said to have a secondary programmatic purpose of drug interdiction. " (emphasis added)); Davis, 270 F.3d at 979–83 ("[Edmond ] more than suggests that if the 'primary purpose' had been for a purpose the Court had already endorsed—such as detecting drunk drivers, or checking licenses—the roadblock would be constitutional."); see also United States v. Gasca–Castillo, No. CR. 06–CR–0060–L, 2007 WL 173888, at *6 (S.D.Cal. Jan. 8, 2007) ; Dale v. State, 785 So.2d 1102, 1105 (Miss.Ct.App.2001). But see 5 Wayne R. LaFave, Search & Seizure § 10.8(a) (5th ed. 2012) ("Surely an illegal multipurpose checkpoint cannot be made legal by the simple device of assigning 'the primary purpose' to one objective instead of the other.").

Furthermore, two circuit courts have upheld other administrative programs with secondary purposes when applying Edmond. The Second Circuit upheld a police department's policy requiring a breathalyzer test for an officer causing injury or death by firing his or her gun, noting "the mere fact that crime control is one purpose—but not the primary purpose—of a program of searches does not bar the application of the special needs doctrine." See Lynch v. New York City, 589 F.3d 94, 102 (2d Cir.2009). The Second Circuit further observed, "[E]ven if crime control is one purpose of a program of searches, the program may nevertheless be reasonable under the special needs doctrine so long as crime control is not the program's primary purpose." Lynch, 589 F.3d at 102 ; see also Davis, 270 F.3d at 979–80. When analyzing the validity of a stop at a national park's information station, the Ninth Circuit held, "While one of the information station's purposes may have been to advance a general interest in crime control, it was not the primary purpose. Indeed, 'the phrase general interest in crime control does not refer to every law enforcement objective.' " Faulkner, 450 F.3d at 471 (quoting Lidster, 540 U.S. at 424, 124 S.Ct. 885 ) (internal quotation marks omitted).

This reading of Edmond aligns with prior Supreme Court cases like Sitz and Martinez–Fuerte. See Lynch, 589 F.3d at 102 (stating that this interpretation of Edmond "derives naturally from prior case law"). In Sitz, the Court upheld the use of sobriety checkpoints with the primary purpose of "reducing the immediate hazard posed by the presence of drunk drivers on the highways." Edmond, 531 U.S. at 39, 121 S.Ct. 447. Analyzing Sitz, the Second Circuit observed "that aiding criminal prosecutions of drunk drivers was another purpose of the checkpoints. It was simply not the checkpoints' primary purpose." Lynch, 589 F.3d at 102. Similarly, the Second Circuit reasoned that, in Martinez–Fuerte, while the Court held that the primary purpose of border checkpoints was "policing the Nation's borders," that purpose was likely not the sole purpose of the checkpoints. Edmond, 531 U.S. at 39, 121 S.Ct. 447. "[A]iding criminal prosecutions of smugglers of illegal immigrants was ... one purpose of the boarder [sic] patrol checkpoints. It was simply not the checkpoints' primary purpose." Lynch, 589 F.3d at 102.

The one wrinkle in this analysis is the Court's footnote in Edmond reserving the question of whether a State may establish a checkpoint program with a valid primary purpose of checking licenses but an invalid secondary purpose. See Edmond, 531 U.S. at 47 n. 2, 121 S.Ct. 447. That footnote, however, "seems divorced from the rest of the opinion." Davis, 270 F.3d at 979. The D.C. Circuit scrutinized Edmond's footnote in some depth, reasoning that the footnote did not ultimately indicate that checkpoints could not have invalid secondary purposes. Id. at 979–80. The Davis court explained in detail:

The record in Edmond suggested that enforcement of the drug laws was not simply Indianapolis's primary reason for establishing the checkpoint program, but its only reason. A sign near each of the checkpoints announced: " 'NARCOTICS CHECKPOINT ____ MILE AHEAD, NARCOTICS K–9 IN USE, BE PREPARED TO STOP.' " If the city's only purpose was narcotics enforcement, it is hard to explain why the Court framed the inquiry in terms of its "primary" purpose, unless the Court believed that it would be constitutional for a State to "establish a checkpoint program with the primary purpose of checking licenses or driver sobriety and a secondary purpose of interdicting narcotics."

Id. (citation omitted) (quoting Edmond, 531 U.S. at 47 n. 2, 121 S.Ct. 447 ).

Based on this analysis, it appears that a checkpoint with an invalid secondary purpose is likely not rendered unconstitutional per se. Here, the Government's evidence fails to establish even a secondary purpose of targeting Hispanics by a preponderance of the evidence. In sum, the Government's § 14141 claim falls short of showing a pattern or practice of targeting or discriminating against Hispanics as part of ACSO's checkpoint operations.

F. Statute of Limitations

Sheriff Johnson argues that the four-year statute of limitations of 28 U.S.C. § 1658(a)⁹⁹ applies to the Government's § 14141 claims and that any proven discriminatory acts are discrete incidents rather than a continuing violation. (Doc. 157 at 127–29.) He therefore contends that the Government cannot predicate § 14141 liability on any acts that occurred prior to December 20, 2008, and that only three facts brought out at trial fall within § 1658(a)'s four-year limitation. (Id. at 128.) The Government's post-trial brief presents no argument on this issue, but the Government previously submitted briefing countering this argument on a motion for summary judgment. (Doc. 106.) The court previously addressed the statute of limitations argument but deferred ruling at the time. See Johnson, 28 F.Supp.3d at 504–06 (observing that statute of limitations applies to claims and not evidence and, even assuming that § 1658(a)'s four-year limitations period applies, Sheriff Johnson failed to explain how the Government's § 14141"pattern or practice" claims accrued before December 20, 2008).¹⁰⁰ Similarly, in light of the court's determinations on the merits, it is a moot point and the court need not do so now.

99 Section 1658 is entitled "Time limitations on the commencement of civil actions arising under Acts of Congress" and provides:

Except as otherwise provided by law, a civil action arising under an Act of Congress enacted after the date of the enactment of this section may not be commenced later than 4 years after the cause of action accrues.

28 U.S.C. § 1658(a). The statute was enacted December 1, 1990, some four years before § 14141's enactment on September 13, 1994. Therefore, by the plain language of § 1658, it applies to claims under § 14141, unless "otherwise provided by law."

100 See Brinkley–Obu v. Hughes Training, Inc., 36 F.3d 336, 346 (4th Cir.1994) (stating, regarding a Title VII claim, that "[s]tatutes of limitations do not operate as an evidentiary bar controlling the evidence admissible at the trial of a timely-filed cause of action"); see also 28 U.S.C. § 1658(a) ("[A] civil action ... commenced later than 4 years after the cause of action accrues" is time barred. (emphasis added)). A "pattern or practice" claim, by definition, requires multiple acts for a claim to accrue. See United States v. City of Parma, Ohio, 661 F.2d 562, 573 (6th Cir.1981) (holding that a pattern or practice claim "necessarily involves a number of discriminatory acts, not a particular one from which the time for bringing suit may be measured").

III. CONCLUSION

What has been presented to the court are the Government's claims that Sheriff Johnson and ACSO violated federal law under § 14141 by engaging in a "pattern or practice" of discriminatory law enforcement against Hispanics in contravention of the Fourth and Fourteenth Amendments. After careful and thorough consideration, as detailed more extensively above, the court concludes that the Government has not met its burden of demonstrating that they have done so. With no evidence that any individual was unconstitutionally deprived of his or her rights under the Fourth or Fourteenth Amendments, the Government's case rested largely on vague, isolated statements attributed to Sheriff Johnson and on statistical analyses. Yet, not a single person testified that any ACSO employee carried out any alleged improper directive or otherwise violated any individual's constitutional rights—indeed, all witnesses, including those called by the Government, denied that they ever did or knew any ACSO officer who did. The Government's statistical analyses similarly failed to constitute reliable and persuasive proof of the claims under applicable legal standards, having failed to sufficiently compare ACSO's treatment of Hispanics to others who were similarly situated. In the context of the significant law enforcement challenges facing ACSO, the Government's evidence falls short.

While ACSO's law enforcement practices do not constitute an unlawful "pattern or practice" of constitutional deprivations in violation of federal law, the court's decision cannot be read to approve or condone all the conduct presented as evidence at trial. Indeed, some of it—for example, the use of ethnic slurs by a few officers largely in the County jail—demonstrated offensive and reprehensible activity that should not be tolerated in any civil society, much less in a law enforcement environment. Other evidence demonstrated potential internal weaknesses in ACSO, such as lack of a system to monitor selection of checkpoint locations, weakness in internal reporting and condemnation of conduct that violates ACSO's internal policy manual, and a lack of substantive review processes for stops and post-stop outcomes. The absence of a finding of a violation of federal law should not be construed as approval of the status quo, and such matters deserve immediate attention.

For the reasons set forth herein, therefore, IT IS ORDERED that the claims of the United States be DENIED, that Judgment be entered for Defendant Sheriff Johnson, and that the complaint (Doc. 1) be DISMISSED WITH PREJUDICE.

IT IS FURTHER ORDERED that any request for costs be filed within thirty (30) days, pursuant to the requirements of Federal Rule of Civil Procedure 54(d), Local Rule 54, and the Equal Access to Justice Act ("EAJA"), 28 U.S.C. § 2412.

A Judgment in conformance with this Order will be entered simultaneously.