SHWARTZ, Circuit Judge.

Lawrence D. Berger, Reisman Carolla Gran & Zuba, 19 Chestnut Street, Haddonfield, NJ 08033, Mary C. Vargas [ARGUED], Stein & Vargas, 10 G Street, N.E., Suite 600, Washington, DC 20002, Counsel for Plaintiff-Appellee Jessica Ramsay

Robert A. Burgoyne [ARGUED], Caroline M. Mew, Perkins Coie, 700 13th Street, N.W., Suite 600, Washington, DC 20005, Alison R. Caditz, Perkins Coie, 1201 Third Avenue, Suite 4900, Seattle, WA 98101, Counsel for Defendant-Appellant National Board of Medical Examiners

Before: GREENAWAY, JR., SHWARTZ, and RENDELL, Circuit Judges.

OPINION

SHWARTZ, Circuit Judge.

Medical student Jessica Ramsay sought testing accommodations for dyslexia and attention deficit hyperactivity disorder ("ADHD") from the National Board of Medical Examiners ("the Board"). The Board denied her requests, and she sued under the Americans with Disabilities Act ("ADA"). The District Court granted a preliminary injunction, requiring the Board to provide her accommodations. We will affirm.

I

A

The Board administers the United States Medical Licensing Examination ("USMLE"). The USMLE has three components, or "Steps," that medical students must pass before they can apply for a medical license. Step 1 is a computer-based, multiple choice exam that assesses a student's grasp of scientific concepts. Students typically take Step 1 before their final year of medical school. Step 2 has two parts: Clinical Knowledge ("CK"), a computer-based, multiple choice exam that assesses medical knowledge and clinical science, and Clinical Skills ("CS") that assesses students in a clinical setting. Step 2 must be taken before graduation. Step 3 is a computer-based exam that assesses the application of medical and scientific knowledge to the practice of medicine. Step 3 must be taken before applying for a medical license.

Ramsay, while a third-year medical student at Western Michigan University ("WMed"), requested an accommodation, namely extra testing time, for Step 1 and Step 2 CK. The basis of her request was that she had ADHD and dyslexia. She submitted to the Board:

• a diagnosis of ADHD and probable dyslexia by her family physician, Dr. Alan Smiy, made when she was an undergraduate;

• records of accommodations provided by her undergraduate institution and by WMed;

• evaluations from Charles Livingston, a licensed social worker, who administered several assessments that supported a diagnosis of ADHD and a likelihood of dyslexia and showed, in his opinion, that Ramsay had "relatively low attention and concentration and very low processing speed," although "[h]er native intelligence has been some compensation for low abilities in the identified areas";

• her MCAT scores, taken without accommodations, placing her in the 67th and 31st percentiles for verbal reasoning and writing, respectively;

• academic records and other standardized test scores, taken without accommodations, showing a high level of achievement; and

• a personal statement attesting that she struggled from an early age with maintaining concentration, reading, and writing, but that she achieved academic success through mitigating strategies, informal accommodations from teachers, and accommodations from her undergraduate and medical schools.

The Board provided Ramsay's materials to an outside reviewer, Dr. Stephen Zecker, who opined that Ramsay was not "substantially limited in functioning in a manner that warrants accommodations." App. 766. The Board also reviewed Ramsay's documentation and, noting her record of achievement without accommodations, concluded that the documents did not "demonstrate a record of chronic and pervasive problems with inattention, impulsivity, behavioral regulation, or distractibility that has substantially impaired [her] functioning during [her] development or currently." App. 1126. Based on Dr. Zecker's recommendation and the Board's review of Ramsay's materials, the Board denied her request.

Thereafter, Ramsay took Step 1 without accommodations in her third year, but she failed by one point. Because WMed requires students to pass Step 1 by the beginning of their fourth year, she took a leave of absence.

Ramsay renewed her request for extra testing time and submitted an evaluation and test data from neuropsychologist Dr. Alan Lewandowski. Dr. Lewandowski met with Ramsay, conducted assessments, found that she had abnormal functionalities in thinking, processing speed, attention, and sequencing, and concluded that she had ADHD. Ramsay also submitted a letter from her treating psychiatrist, Dr. Bruce Ruekberg, who concurred with Mr. Livingston's and Dr. Lewandowski's assessments, stating that she had abnormal scanning and processing speed that impaired her reading and written expression. The Board denied her request for extra testing time, again concluding that she had not shown she was substantially limited in any functions as compared to most people.

Ramsay sought reconsideration of the Board's denial. As additional support, she provided an evaluation by Dr. Robert D. Smith, a psychologist and neuropsychologist. Dr. Smith met with Ramsay, reviewed her records, and performed similar assessments. He reported that the assessments revealed that she had abnormally low abilities in processing information, writing, and reading, indicating dyslexia and ADHD. Among other things, his testing revealed that Ramsay, as compared to others in her age group, was in the fourth percentile in reading comprehension and fluency, second percentile in word reading speed, and first percentile in oral reading fluency.

The Board provided Ramsay's file to outside expert Dr. Benjamin Lovett, who concluded that Ramsay did not show poor academic skills or impairments compared to the general population and thus lacked a condition that would warrant accommodations. Based on Dr. Lovett's recommendation and further review, the Board denied Ramsay's request for reconsideration.

B

Ramsay sued the Board in May 2019, alleging that it had violated the ADA. The next month, WMed informed Ramsay that it could extend her leave only until March 2020, "with the expectation that [she] will sit for the USMLE Step 1 exam in a manner that allows [her] to return to the WMed curriculum by that date." App. 1520. WMed informed Ramsay that if she did not pass Step 1 and return by March 2020, she would be dismissed or could voluntarily withdraw, but readmission would not be guaranteed. Ramsay accepted WMed's conditional extension of leave.

Because Ramsay had to pass Step 1 to avoid dismissal, she sought a preliminary injunction to require the Board to grant her accommodations. The District Court held a three-day evidentiary hearing featuring testimony from, among others, Ramsay, Dr. Smith, Dr. Zecker, and Dr. Lovett.

For the reasons explained in its careful and thorough opinion, the District Court granted Ramsay a preliminary injunction and required the Board to provide Ramsay with double the testing time on Step 1, Step 2 CK, any written or reading portions of Step 2 CS, and Step 3. Ramsay v. Nat'l Bd. of Med. Exam'rs, No. 19-CV-2002, 2019 WL 7372508 (E.D. Pa. Dec. 31, 2019). The Court found that all the experts were qualified, but that the testimony and reports of the experts who met with Ramsay were more persuasive. Id. at *17. Those experts stated that their assessments and evaluations all showed that Ramsay had low reading, writing, and processing abilities. Id. at *15-16. The Court also found that the Board's experts’ analyses contradicted applicable regulations by focusing too much on Ramsay's academic achievements, substituting their own opinions for those of experts who met with Ramsay, and placing too demanding a burden on Ramsay. Id. at *17-18. Based on this evidence and the governing law, the Court found that Ramsay had a disability under the ADA. Id. at *18.

The Court also found that: (1) Ramsay established irreparable harm because she would likely be forced to withdraw from WMed if she could not take Step 1 with accommodations and pass, (2) the balance of equities tipped in her favor because granting her accommodations would not undermine the Board's interests in fair and accurate testing, and (3) it was in the public interest for the ADA to be followed and to increase the number of physicians. Id. at *18-19. The Board appeals.

II

In issuing a preliminary injunction, a district court considers four factors:

(1) the likelihood that the plaintiff will prevail on the merits at final hearing; (2) the extent to which the plaintiff is being irreparably harmed by the conduct complained of; (3) the extent to which the defendant will suffer irreparable harm if the preliminary injunction is issued; and (4) [that] the public interest [weighs in favor of granting the injunction].

Greater Phila. Chamber of Commerce v. City of Philadelphia, 949 F.3d 116, 133 (3d Cir. 2020) (alterations in original) (quoting Am. Tel. & Tel. Co. v. Winback & Conserve Program, Inc., 42 F.3d 1421, 1427 (3d Cir. 1994) ).

A

We first address Ramsay's likelihood of success on the merits of her ADA claim. "On this factor, a sufficient degree of success for a strong showing exists if there is a reasonable chance or probability, of winning" on her ADA claim. Ass'n of N.J. Rifle & Pistol Clubs, Inc. v. Att'y Gen. N.J., 910 F.3d 106, 115 (3d Cir. 2018) (internal quotation marks and citation omitted). The ADA provides in relevant part:

Any person that offers examinations ... related to applications, licensing, certification, or credentialing for ... professional ... purposes shall offer such examinations ... in a place and manner accessible to persons with disabilities or offer alternative accessible arrangements for such individuals.

42 U.S.C. § 12189. The issue here is whether Ramsay has a "disability" that entitles her to an accommodation. Ramsay, 2019 WL 7372508, at *8.

The ADA defines "disability" in relevant part as "a physical or mental impairment that substantially limits one or more major life activities of such individual." 42 U.S.C. § 12102(1)(A). We construe the term "disability" broadly. Id. § 12102(4)(A). As to the term "impairment," the applicable Department of Justice ("DOJ") regulations provide that the term "physical or mental impairment" includes ADHD and "dyslexia and other specific learning disabilities." 28 C.F.R. § 36.105(b)(2). As to "life activities," the ADA provides that "major life activities include ... reading, concentrating, thinking, communicating, and working." 42 U.S.C. § 12102(2)(A). Finally, the regulations explain that "[a]n impairment is a disability ... if it substantially limits the ability of an individual to perform a major life activity as compared to most people in the general population." 28 C.F.R. § 36.105(d)(1)(v). Accordingly, " ‘[n]ot every impairment will constitute a disability ...,’ but [an impairment] will meet the definition [of disability] if ‘it substantially limits the ability of an individual to perform a major life activity as compared to most people in the general population.’ " J.D. by Doherty v. Colonial Williamsburg Found., 925 F.3d 663, 670 (4th Cir. 2019) (quoting 28 C.F.R. § 36.105(d)(1)(v) ).

1

The Board argues that the District Court did not determine that Ramsay is substantially limited in comparison to most people in the general population. We first address the concept of "most people in the general population" in the learning disability context. In general,

[t]he comparison to most people in the general population ... mean[s] a comparison to other people in the general population, not a comparison to those similarly situated. For example, the ability of an individual with an amputated limb to perform a major life activity is compared to other people in the general

population, not to other amputees. This does not mean that disability cannot be shown where an impairment, such as a learning disability, is clinically diagnosed based in part on a disparity between an individual's aptitude and that individual's actual versus expected achievement, taking into account the person's chronological age, measured intelligence, and age-appropriate education. Individuals diagnosed with dyslexia or other learning disabilities will typically be substantially limited in performing activities such as learning, reading, and thinking when compared to most people in the general population ....

Regulations to Implement the Equal Employment Provisions of the Americans with Disabilities Act, as Amended, 76 Fed. Reg. 16,978, 17,009 (Mar. 25, 2011) (explanation by the Equal Employment Opportunity Commission ("EEOC")); see Amendment of Americans with Disabilities Act Title II and Title III Regulations to Implement ADA Amendments Act of 2008, 81 Fed. Reg. 53,204, 53,230 (Aug. 11, 2016) (DOJ "concur[ring] with" EEOC's "view"). Thus, a clinical diagnosis of a learning disability is typically based upon a comparison between the individual and others in the general population who are of similar age and have received age-appropriate education.

Here, the District Court relied on such diagnostic information to conclude that Ramsay had ADHD and dyslexia that caused her to read and write with more difficulty than most people. For example, Dr. Smith's and Dr. Lewandowski's diagnostic assessments showed that Ramsay had abnormal functionalities in thinking, processing speed, attention, and sequencing. Indeed, some of the reading tests Dr. Smith administered placed Ramsay in less than the fifth percentile as compared to individuals her age. This is exactly the type of data DOJ contemplates as showing a learning disability that substantially limits an individual as compared to others in the general population. Equal Employment Provisions, 76 Fed. Reg. at 17,009 ; Title II and Title III Regulations, 81 Fed. Reg. at 53,230. Further, Ramsay explained in her personal statement that she had struggled with reading and writing tasks in comparison to her classmates since elementary school. Thus, the Court's finding that Ramsay's ADHD and dyslexia constituted a disability was based on evidence that these conditions substantially limit her reading and writing in comparison to most people. See Pryer v. C.O. 3 Slavic, 251 F.3d 448, 453 n.4 (3d Cir. 2001) (inferring the district court's reasoning where it was "otherwise apparent from the record").

Moreover, the regulations provide that the "substantially limits" inquiry "should not demand extensive analysis," 28 C.F.R. § 36.105(d)(1)(ii), and that "[t]he comparison of an individual's performance of a major life activity to the performance of the same major life activity by most people in the general population usually will not require scientific, medical, or statistical evidence," id. § 36.105(d)(1)(vii). Accordingly, the District Court's reliance on evidence that Ramsay's reading, processing, and writing skills were abnormally low by multiple measures provided a sufficient comparison of her abilities to those of the general population to support the finding of disability.

2

Next, the Board argues that the District Court erred by giving "considerable weight" to Ramsay's past accommodations when determining that she has a disability. Appellant's Br. at 45 (quoting 28 C.F.R. § 36.309(b)(1)(v) ). According to the Board, a court should consider past accommodations only after finding the individual is disabled. This argument fails.

The regulation defining disability, § 36.105, does not bar consideration of past accommodations. Indeed, § 36.309(b)(1)(v) provides that "[w]hen considering requests for ... accommodations ... the [testing] entity gives considerable weight to documentation of past ... accommodations." Moreover, as the preamble to the applicable regulations states, "a recent history of past accommodations is critical to an understanding of the applicant's disability and the appropriateness of testing accommodations." Nondiscrimination on the Basis of Disability by Public Accommodations and in Commercial Facilities, 75 Fed. Reg. 56,236, 56,298 (Sept. 15, 2010) (to be codified at 28 C.F.R. pt. 36). Thus, the District Court did not err in considering Ramsay's past accommodations.

3

The Board also argues that the District Court wrongly believed that the statute and regulations compelled it to defer to experts who met with and tested Ramsay. While the Court viewed Ramsay's experts more favorably and found the Board's experts unpersuasive, there is no indication that the Court believed that it was compelled to defer to Ramsay's experts. Rather, the Court discounted the Board's experts because they (1) never met with Ramsay, (2) engaged in too demanding an analysis of whether Ramsay had a disability, and (3) focused too much on Ramsay's academic achievements. Ramsay, 2019 WL 7372508, at *17-18. The Court's reasoning was within its discretion and supported by the regulations.

First, it is within the trial judge's discretion to credit a physician with firsthand observations of a patient over one who only reviewed the patient's records. See United States v. Olhovsky, 562 F.3d 530, 548-49 (3d Cir. 2009). Such a professional has the benefit of seeing how the patient actually acts and speaks and provides a perspective not limited to the cold record. This principle is not unlike the deference an appellate court gives to a trial court who physically sees a witness. Cooper v. Harris, ––– U.S. ––––, 137 S. Ct. 1455, 1474, 197 L.Ed.2d 837 (2017). This is why we rarely second-guess a district court's weighing of evidence, see, e.g., United States v. Turner, 718 F.3d 226, 231 (3d Cir. 2013), and why it makes sense for the District Court to credit the professionals who personally met with Ramsay.

Second, the regulations mandate that "[t]he determination of whether an impairment substantially limits a major life activity requires an individualized assessment." 28 C.F.R. § 36.105(d)(1)(vi). Such assessments benefit from the reports of professionals who know or have personally examined the individual. Because such examinations allow the professional to evaluate the individual's behavior, effort, and candor, DOJ understandably has stated that "[r]eports from experts who have personal familiarity with the candidate should take precedence over those from ... reviewers for testing agencies, who have never personally met the candidate or conducted the requisite assessments for diagnosis and treatment." Nondiscrimination on the Basis of Disability, 75 Fed. Reg. at 56,297. As a result, DOJ has directed that testing entities "shall generally accept" "documentation provided by a qualified professional who has made an individualized assessment of an applicant that supports the need for the modification, accommodation, or aid requested ... and provide the accommodation." Id. Thus, the Court's decision to weigh Ramsay's experts more favorably than those of the Board was consistent with DOJ regulations.

Third, "the threshold issue of whether an impairment substantially limits a major life activity should not demand extensive analysis." 28 C.F.R. § 36.105(d)(1)(ii). The Court could reasonably have concluded that the Board's experts were too demanding in what they required to prove a disability, for example, by demanding evidence of a lifetime of academic struggles, and "substituting their own opinions" for those of Ramsay's healthcare providers. Ramsay, 2019 WL 7372508, at *17. In fact, the Board's reliance on Ramsay's academic achievement was contrary to the regulations that explain that "someone with a learning disability may achieve a high level of academic success, but may nevertheless be substantially limited in one or more major life activities, including, but not limited to, reading, writing, speaking, or learning because of the additional time or effort he or she must spend to read, write, speak, or learn compared to most people." 28 C.F.R. § 36.105(d)(3)(iii). Because Ramsay's high academic performance does not foreclose her from having a disability, the Court reasonably discounted the Board's experts’ opinions, which focused mostly on Ramsay's academic accomplishments and ignored evidence of her limitations. Ramsay, 2019 WL 7372508, at *18.

In sum, nothing in the District Court's discussion indicates that it held that the statute and regulations "compel" deference to Ramsay's experts. Rather, the Court found that Ramsay's experts provided facts more probative to the relevant inquiries under the ADA, and its decision to view these witnesses more favorably is consistent with the regulations. Thus, we will not disturb how the Court chose to weigh evidence.

4

The additional errors the Board identifies in the Court's factual findings do not amount to clear error. "A finding of fact is clearly erroneous when it is completely devoid of minimum evidentiary support displaying some hue of credibility or bears no rational relationship to the supportive evidentiary data." VICI Racing, LLC v. T-Mobile USA, Inc., 763 F.3d 273, 283 (3d Cir. 2014) (internal quotation marks and citation omitted). We examine the entire record to determine whether there is evidentiary support for a finding, not just the evidence a district court cites. See N.J. Rifle, 910 F.3d at 120 n.24.

First, the Board argues that the District Court erred in finding that the Board's consultants found that Dr. Smith's assessments were valid and credible. Contrary to the Board's assertion, the record supports the Court's finding. Both of the Board's consultants testified that they had no reason to doubt that the assessments were properly administered, that the results were accurate, and that the data could be useful, although they disagreed with Dr. Smith's interpretation of the results. The credibility of evidence is different from the inferences a factfinder can draw from that evidence, so the Court's finding that all experts agreed the assessments were credible was supported by the consultants’ testimony, even if the Board's consultants reached different conclusions from the test results themselves.

Second, the Board argues that the District Court erred in finding that Ramsay could not finish reading and had to guess on about a third of the questions on Step 1 because the time Ramsay spent on each question shows that "she had time to read every question." Appellant's Br. at 61 (emphasis omitted) (citing Ramsay, 2019 WL 7372508, at *3 ). The record does not contradict the Court's finding. The Board's evidence does not indicate how much time Ramsay spent reading each question. Rather, it shows only that she spent, on average, seventeen seconds more on the questions she got incorrect. Further, Ramsay testified that she took a pass through the questions before answering them, answered the ones she felt she could, and repeated that strategy until she was left with a few questions she could not answer even after multiple reads. Her strategy provides a reasonable explanation for why the time spent on correct versus incorrect answers was similar. The Court was free to credit Ramsay's testimony over the inferences that the Board argued should be drawn from its measurements. See Cloverland-Green Spring Dairies, Inc. v. Pa. Milk Mktg. Bd., 462 F.3d 249, 271 (3d Cir. 2006) ("Where there are two permissible views of the evidence, the factfinder's choice between them cannot be clearly erroneous." (quoting Scully v. US WATS, Inc., 238 F.3d 497, 506 (3d Cir. 2001) )).

Finally, the Board argues that the District Court erred in finding that Ramsay had received informal accommodations in her early school years. Ramsay testified about, and her mother relayed to Dr. Smith information concerning, these informal accommodations. While the Board asserts that there is no written record of these informal accommodations, Ramsay's corroborated testimony provided "minimum evidentiary support" for the Court's finding, so there was no clear error. VICI Racing, 763 F.3d at 283 (citation omitted).

B

We next determine whether Ramsay proved irreparable harm. "[T]o show irreparable harm a plaintiff must demonstrate potential harm which cannot be redressed by a legal or an equitable remedy following a trial." Acierno v. New Castle County, 40 F.3d 645, 653 (3d Cir. 1994) (internal quotation marks and citation omitted). The harm must be "likely" to occur "in the absence of an injunction." Ferring Pharms., Inc. v. Watson Pharms., Inc., 765 F.3d 205, 217 n.11 (3d Cir. 2014) (emphasis omitted) (quoting Winter v. Nat. Res. Def. Council, Inc., 555 U.S. 7, 22, 129 S.Ct. 365, 172 L.Ed.2d 249 (2008) ).

The District Court had a basis to conclude that Ramsay would be irreparably harmed absent an injunction. The Court could reasonably conclude that given Ramsay's disability and that she had previously failed Step 1, she likely would fail again and be forced to leave medical school. Ramsay, 2019 WL 7372508, at *18. Her termination from medical school and its consequences could not later "be redressed by a legal or an equitable remedy." Acierno, 40 F.3d at 653 (citation omitted). No damages remedy is available under the ADA. 42 U.S.C. § 12188(a)(1) (providing that the only remedies available in an ADA action are those in § 2000a-3(a)); id. § 2000a-3(a) (providing for injunctive relief). Furthermore, because WMed is not a party to this case, the Court could not require it to reinstate her, and the Board presents no theory for how the Board could redress the termination of Ramsay's medical education. Moreover, an examiner's refusal to provide accommodations can cause the exam-taker irreparable harm because doing so jeopardizes her "opportunity to pursue her chosen profession." Enyart v. Nat'l Conf. of Bar Exam'rs, 630 F.3d 1153, 1166 (9th Cir. 2011) ; accord Doe v. Pa. State Univ., 276 F. Supp. 3d 300, 313-14 (M.D. Pa. 2017) (holding that gap in medical school education and likelihood that the student could not gain acceptance to another school constituted irreparable harm). Accordingly, the District Court correctly concluded that Ramsay established she would be irreparably harmed absent an injunction.

C

We next consider how the District Court "balanc[ed] the parties’ relative harms; that is, the potential injury to the plaintiff[ ] without this injunction versus the potential injury to the defendant with it in place." Issa v. Sch. Dist. of Lancaster, 847 F.3d 121, 143 (3d Cir. 2017). In balancing the harms, the Court noted the Board's "concern for the fulfillment of its mission to provide [qualified] physicians," Ramsay, 2019 WL 7372508, at *19, and that accommodations "can affect the comparability of the resulting scores and scores achieved under standard testing conditions," id. at *4 (quoting App. 931). Nonetheless, the Court appropriately reasoned that granting a preliminary injunction would not undermine the Board's mission because the injunction would give Ramsay only "the opportunity to move forward" in her medical career "should she succeed in passing her examinations with appropriate accommodations." Id. at *19 (emphasis omitted). Moreover, the Board's concerns regarding impacts from undeserved accommodations do not apply here because Ramsay has shown a reasonable likelihood that she deserves accommodations. Cf. Issa, 847 F.3d at 143 (holding that a defendant could not assert an interest in continuing to violate a civil rights statute).

D

Finally, we consider the District Court's finding that "the public interest favors this preliminary injunction." Id. The Court concluded that an injunction furthers the public interest in ADA compliance and serves to increase the number of qualified physicians. Ramsay, 2019 WL 7372508, at *19. We agree. "In enacting the ADA, Congress demonstrated its view that the public has an interest in ensuring the eradication of discrimination on the basis of disabilities." Enyart, 630 F.3d at 1167 ; see Issa, 847 F.3d at 143 (concluding that it was in the public interest for covered entities to comply with a civil rights statute). Further, the injunction allows Ramsay to continue her medical education and therefore serves the public interest in training more physicians. "Although it is true that the public also has an interest in ensuring the integrity of licensing exams," Enyart, 630 F.3d at 1167, Ramsay has shown a reasonable likelihood that the ADA affords her accommodations, and there is no evidence that providing her the requested accommodations will jeopardize the test's integrity. Thus, the public interest weighs in favor of an injunction.

III

For the foregoing reasons, we will affirm the District Court's preliminary injunction.