WIGGINS, Chief Justice.

Jamie L. Hunter of Dickey & Campbell Law Firm, PLC, Des Moines, and Andrew J. Smith of Mack, Hansen, Gadd, Armstrong & Brown, P.C., Storm Lake, for appellant.

Thomas J. Miller, Attorney General, Louis S. Sloven, Assistant Attorney General, Darren Driscoll, County Attorney, and Brad M. McIntyre, Assistant County Attorney, for appellee.

WIGGINS, Chief Justice.

An applicant sought postconviction relief (PCR), claiming a violation of his constitutional right to an impartial jury drawn from a fair cross section of the community under the United States and Iowa Constitutions. He based his claim on State v. Plain , 898 N.W.2d 801 (Iowa 2017). The district court dismissed his PCR application, and he appealed the order of dismissal.

On appeal, we find Plain is a new ground of law allowing an applicant to bring a PCR action after the three-year statute of limitations in Iowa Code section 822.3 (2018) has run. Nonetheless, we affirm the order of dismissal because we find our holding in Plain does not apply retroactively to cases on collateral review.

I. Background Facts and Proceedings.

A jury convicted Khamfeung Thongvanh of first-degree murder in 1984. He appealed, and the court of appeals affirmed his conviction in 1986. State v. Thongvanh , 398 N.W.2d 182, 184, 189 (Iowa Ct. App. 1986) (en banc). A few years later, he filed a PCR application, raising among other things a fair-cross-section claim. Thongvanh v. State (Thongvanh II ), 494 N.W.2d 679, 680, 683 (Iowa 1993). We affirmed the denial of that application in 1993. Id. at 684.

On June 30, 2017, we decided Plain , which addressed the Duren three-part test for evaluating Sixth Amendment fair-cross-section claims. 898 N.W.2d at 821–28 ; see Duren v. Missouri , 439 U.S. 357, 364, 367–68, 99 S. Ct. 664, 668, 670, 58 L.Ed.2d 579 (1979) (laying out test for evaluating Sixth Amendment fair-cross-section claims). Under Duren , the criminal defendant must first establish a prima facie fair-cross-section violation by showing

(1) that the group alleged to be excluded is a "distinctive" group in the community; (2) that the representation of this group in venires from which juries are selected is not fair and reasonable in relation to the number of such persons in the community; and (3) that this underrepresentation is due to systematic exclusion of the group in the jury-selection process.

Plain , 898 N.W.2d at 822 (quoting Duren , 439 U.S. at 364, 99 S. Ct. at 668 ). Then, if the defendant establishes a prima facie violation, "the burden shifts to the state to justify the disproportionate representation by proving ‘a significant state interest’ is ‘manifestly and primarily advanced’ by the causes of the disproportionate exclusion." Id. (quoting Duren , 439 U.S. at 367–68, 99 S. Ct. at 670 ).

In Plain , we expressly overruled our precedent that had adopted the absolute-disparity method as the exclusive indicator of representativeness under the second prong of Duren . Id. at 826. That is, we held,

Parties challenging jury pools on the ground that they are unrepresentative may base their challenges on multiple analytical models [such as the absolute disparity, comparative disparity, and standard deviation tests]. The district court may use multiple analytical models in its analysis, taking into account the various strengths and weaknesses of each test when determining whether jury pools comport with the Sixth Amendment mandate of representativeness.

Id. at 827.

This past term we modified Plain ’s holding in State v. Lilly , 930 N.W.2d 293, 302 (Iowa 2019). Lilly involved a fair-cross-section claim raised under both the Sixth Amendment and article I, section 10, but the defendant did not advance a distinct analysis under article I, section 10. Id. at 300, 301. Accordingly, we applied the Sixth Amendment framework under article I, section 10 but reserved the right to apply that framework differently. Id. at 301.

We held that neither the absolute disparity method nor the comparative-disparity method is appropriate to use when considering the underrepresentation prong of a fair-cross-section claim. Id. at 302. However, we acknowledged that the standard deviation method is appropriate. Id. In State v. Veal and State v. Williams , two companion cases to Lilly , we applied Lilly ’s holding with modifications to Sixth Amendment fair-cross-section claims and further discussed the application of Plain to such claims. Veal , 930 N.W.2d 319, 328–30, 328 n.5 (Iowa 2019) ; Williams , 929 N.W.2d 621, 629–30, 629 n.1 (Iowa 2019).

On January 26, 2018, Thongvanh filed the instant PCR application. Relying on our holding in Plain , he alleged he was denied his rights to due process, equal protection, and a fair and impartial trial under the United States and Iowa Constitutions. The State moved to dismiss Thongvanh’s application, contending no new ground of law or fact obviated Iowa Code section 822.3 ’s three-year statute of limitations and section 822.8 barred Thongvanh’s fair-cross-section claim. The State did not argue or contend that Thongvanh’s application should be dismissed because Plain is not retroactive.

Thongvanh resisted, arguing section 822.3 ’s limitations period did not apply because Plain constitutes a new ground of law that could not have been raised within the applicable time period. Like the State, he did not discuss Plain ’s retroactivity.

During the hearing on the State’s motion to dismiss, the court inquired whether Plain can apply retroactively to a fair-cross-section claim made on collateral review and then ordered the attorneys to brief that issue in more detail. In his posthearing brief, Thongvanh argued Plain created a watershed rule of criminal procedure that implicates the fundamental fairness of a trial and, thus, could apply retroactively to cases on collateral review.

The district court disagreed with Thongvanh and granted the State’s motion to dismiss. It first concluded that Plain is not retroactive because, "[d]espite the imperative of fair jury representation in criminal matters, by merely permitting challenges based on different statistical models, Plain does not make a ‘watershed rule of criminal procedure.’ "

The court also concluded that neither equal protection nor due process require retroactive application of Plain to cases on collateral review. It acknowledged Thongvanh’s contention that the Iowa Constitution provides greater guarantees of equal protection than the Federal Constitution but noted Thongvanh did not explain why the Iowa provision should provide protection that is any different than that of the federal provision.

Lastly, the court noted that Thongvanh had been unable to establish the third prong of the Duren fair-cross-section test—systematic exclusion—in his 1993 PCR case. It found he had not presented any new ground of law or fact that would allow reconsideration of our conclusion in Thongvanh’s 1993 appeal. Particularly, that he had not established the underrepresentation of Asians on his jury panel was due to systematic exclusion of Asians from jury duty.

Thongvanh appealed, and we transferred the case to the court of appeals. The court of appeals agreed with the district court that Thongvanh’s claim based on Plain ’s holding on the second prong of the Duren test was a new ground of law not previously available to Thongvanh and, therefore, was not time-barred by section 822.3. It also agreed that Plain is not retroactive because it did not create a watershed rule of criminal procedure. Finally, it declined Thongvanh’s invitation to interpret the Iowa Constitution to give broader retroactive application to new rules of criminal procedure than under federal caselaw, reasoning our precedent precluded it from doing so and it was not at liberty to ignore or modify that precedent.

We granted Thongvanh’s petition for further review.

II. Scope and Standards of Review.

"Our review of the court’s ruling on the State’s statute-of-limitations defense is for correction of errors of law." Phuoc Thanh Nguyen v. State , 829 N.W.2d 183, 186 (Iowa 2013) (quoting Harrington v. State , 659 N.W.2d 509, 519 (Iowa 2003) ). We will affirm if substantial evidence supports the district court’s findings of fact and the court correctly applied the law. Id.

Similarly, we review the district court’s ruling on the motion to dismiss for correction of errors of law. Shumate v. Drake Univ. , 846 N.W.2d 503, 507 (Iowa 2014). "For purposes of reviewing a ruling on a motion to dismiss, we accept as true the petition’s well-pleaded factual allegations, but not its legal conclusions." Id. When the petition’s allegations are taken as true yet fail to state a claim upon which relief may be granted, we will affirm the district court’s granting of the motion to dismiss. Id.

III. Issues.

We address two issues in this appeal. First, whether Thongvanh’s Plain claim is time-barred by section 822.3. Second, whether Plain can apply retroactively to a case on collateral review.

IV. Whether Thongvanh’s Plain Claim Is Time-Barred by Iowa Code Section 822.3.

The State appears to challenge the district court’s conclusion that, because Thongvanh filed his PCR application raising his Plain claim within three years of Plain , his Plain claim is not time-barred by section 822.3. Section 822.3 requires PCR applications "be filed within three years from the date the conviction or decision is final or, in the event of an appeal, from the date the writ of procedendo is issued." Iowa Code § 822.3. There is an exception to that three-year statute of limitations for "a ground of fact or law that could not have been raised within the applicable time period." Id.

Thongvanh contends that a challenge to a jury pool based on Plain is a ground of law that could not have been raised before Plain was decided. We agree.

When Thongvanh raised his fair-cross-section claim in his original PCR application and we rejected that claim on appeal in 1993, the absolute-disparity method was the exclusive test used in Iowa when evaluating the second prong of the Duren test. See State v. Jones , 490 N.W.2d 787, 793 (Iowa 1992) (expressly rejecting reliance on the comparative-disparity method and holding the absolute-disparity method "is the appropriate method to be used"), overruled in part by Plain , 898 N.W.2d at 826 ; see also Plain , 898 N.W.2d at 825 (indicating Jones resulted in the exclusive use of the absolute-disparity method); Thongvanh II , 494 N.W.2d at 683–84 (finding criminal defense counsel was not ineffective for failing to raise a fair-cross-section claim because the results of the absolute-disparity method prevented the PCR applicant from making a prima facie case of underrepresentation); State v. Huffaker , 493 N.W.2d 832, 833–34 (Iowa 1992) (rejecting the defendant’s fair-cross-section claim after considering only the absolute disparity method).

Plain effected an unmistakable change in the law when it expressly overruled Jones . Cf. Phuoc Thanh Nguyen , 829 N.W.2d at 188 (discussing the felony-murder rule as law at the time of the defendant’s conviction and our decision in State v. Heemstra , 721 N.W.2d 549 (Iowa 2006) was not "simply a ‘clarification of the law’ or ‘an application of preexisting law’ " but an express overruling of prior law (quoting Perez v. State , 816 N.W.2d 354, 360–61 (Iowa 2012) )). Specifically, we stated in Plain ,

Our decision to adopt absolute disparity as the exclusive test and to reject comparative disparity in Jones rested upon an error of law and on cases from other jurisdictions that have since been overruled or criticized. After surveying the various tests, and bearing in mind the practical problems associated with the use of the absolute disparity test in Iowa, we conclude it is no longer appropriate to rely exclusively upon the absolute disparity test as an indicator of representativeness. We therefore overrule Jones , 490 N.W.2d at 792–93.

898 N.W.2d at 826.

Plain ’s overruling of Jones is distinguishable from a situation such as the one in Perez , 816 N.W.2d 354. In Perez , we concluded the Supreme Court’s holding in Padilla v. Kentucky , 559 U.S. 356, 374, 130 S. Ct. 1473, 1486, 176 L.Ed.2d 284 (2010) —that a criminal defendant has a Sixth Amendment right to be advised by counsel of the risk of deportation before pleading guilty—was not a new ground of law under section 822.3. Perez , 816 N.W.2d at 360. We found Perez could have raised such an argument before Padilla was decided, even though at the time of Perez’s guilty plea our caselaw imposed a different rule. Id. at 360–61. We explained that a case challenging our precedent was pending at the time Perez’s conviction became final and that while we did not overrule our precedent in that pending case, we acknowledged there was some merit to the argument to do so. Id. at 360. Additionally, we noted that at any time, the Supreme Court could have overturned any of our precedents, which is what eventually happened in Padilla . Id. Finally, we noted that shortly after we declined to overrule our precedent, we amended Iowa Rule of Criminal Procedure 2.8 to require defendants be informed that pleading guilty may affect their immigration status under federal law. Id.

Unlike in Perez , we had not considered or espoused any disagreement with Jones ’s holding before Plain . Rather, we had reaffirmed sole use of the absolute-disparity method. See Thongvanh II , 494 N.W.2d at 683–84 ; Huffaker , 493 N.W.2d at 833–34 ; see also State v. Lambert , 501 N.W.2d 64, 68 (Iowa Ct. App. 1993) (utilizing the absolute-disparity method without questioning our exclusive reliance on it).

Because we had clearly held to the contrary in Jones , between 1993—when procedendo issued following our rejection of Thongvanh’s fair-cross-section claim in his original PCR application—and 1996—when the three-year statute of limitations period ran—Thongvanh could not have successfully argued that the jury pool in his criminal trial was not drawn from a fair cross section of the community as demonstrated by using any method other than absolute disparity as the indicator of representativeness. See Phuoc Thanh Nguyen , 829 N.W.2d at 188 (stating that the exception to section 822.3 ’s limitations period "must envision a category of legal claims that were viewed as fruitless at the time but became meritorious later on"). Thus, Thongvanh’s Plain claims are not barred by section 822.3 ’s three-year statute of limitations.

V. Whether Plain Can Apply Retroactively to a Case on Collateral Review.

Thongvanh contends our holding in Plain applies retroactively to cases on collateral review under the retroactivity analysis provided by the United States Supreme Court in Teague v. Lane , 489 U.S. 288, 109 S. Ct. 1060, 103 L.Ed.2d 334 (1989). Alternatively, he asks us to adopt a broader approach to retroactivity under the Iowa Constitution’s due process and equal protection clauses.

A. Retroactivity Under Teague . Teague and its progeny lay out the framework for whether a new rule announced in a Supreme Court opinion applies retroactively to cases on federal habeas review—what we will refer to as the " Teague framework." E.g. , Danforth v. Minnesota , 552 U.S. 264, 266, 281–82, 128 S. Ct. 1029, 1032, 1042, 169 L.Ed.2d 859 (2008) ; Whorton v. Bockting , 549 U.S. 406, 416, 127 S. Ct. 1173, 1180, 167 L.Ed.2d 1 (2007) ; Schriro v. Summerlin , 542 U.S. 348, 351–52, 124 S. Ct. 2519, 2522–23, 159 L.Ed.2d 442 (2004) ; Nguyen v. State (Nguyen II ), 878 N.W.2d 744, 753 & n.3 (Iowa 2016). Under this framework, a "new rule" includes one that "was not dictated by precedent existing at the time the defendant’s conviction became final." Teague , 489 U.S. at 301, 109 S. Ct. at 1070 (emphasis omitted).

New substantive rules will generally apply retroactively. E.g. , Schriro , 542 U.S. at 351, 124 S. Ct. at 2522. These are rules "that narrow the scope of a criminal statute by interpreting its terms as well as constitutional determinations that place particular conduct or persons covered by the statute beyond the State’s power to punish." Id. at 351–52, 124 S. Ct. at 2522 (citation omitted); accord Montgomery v. Louisiana , 577 U.S. ––––, ––––, 136 S. Ct. 718, 728, 193 L.Ed.2d 599 (2016) ("Substantive rules include ‘rules forbidding criminal punishment of certain primary conduct,’ as well as ‘rules prohibiting a certain category of punishment for a class of defendants because of their status or offense.’ " (quoting Penry v. Lynaugh , 492 U.S. 302, 330, 109 S. Ct. 2934, 2953, 106 L.Ed.2d 256 (1989), abrogated on other grounds by Atkins v. Virginia , 536 U.S. 304, 321, 122 S. Ct. 2242, 2252, 153 L.Ed.2d 335 (2002) )). They apply retroactively because they "necessarily carry a significant risk that a defendant stands convicted of ‘an act that the law does not make criminal’ or faces a punishment that the law cannot impose." Schriro , 542 U.S. at 352, 124 S. Ct. at 2522–23 (quoting Bousley v. United States , 523 U.S. 614, 620, 118 S. Ct. 1604, 1610, 140 L.Ed.2d 828 (1998) ).

Conversely, under the Teague framework, new rules of procedure generally do not apply retroactively. E.g. , Montgomery , 577 U.S. at –––, 136 S. Ct. at 728. The Court has explained that this is because their connection to innocence is more attenuated:

New rules of procedure ... do not produce a class of persons convicted of conduct the law does not make criminal, but merely raise the possibility that someone convicted with use of the invalidated procedure might have been acquitted otherwise.

Schriro , 542 U.S. at 352, 124 S. Ct. at 2523. Accordingly, it gives retroactive effect to only new " ‘watershed rules of criminal procedure’ implicating the fundamental fairness and accuracy of the criminal proceeding." Montgomery , 577 U.S. at ––––, 136 S. Ct. at 728 (quoting Schriro , 542 U.S. at 352, 124 S. Ct. at 2523 ). Watershed rules implicating fundamental fairness are those

best illustrated by recalling the classic grounds for the issuance of a writ of habeas corpus—that the proceeding was dominated by mob violence; that the prosecutor knowingly made use of perjured testimony; or that the conviction was based on a confession extorted from the defendant by brutal methods.

Teague , 489 U.S. at 313–14, 109 S. Ct. at 1077 (quoting Rose v. Lundy , 455 U.S. 509, 544, 102 S. Ct. 1198, 1216–17, 71 L.Ed.2d 379 (1982) (Stevens, J., dissenting)). See generally Danforth , 552 U.S. at 271–73, 128 S. Ct. at 1036 (summarizing the development of under what grounds federal habeas relief was available).

The Teague retroactivity framework resulted from the Court interpreting the scope of the relief available under the federal habeas statute. Danforth , 552 U.S. at 278–79, 128 S. Ct. at 1039–40. In other words, the Teague test indicates whether the federal habeas statute requires that a new rule be given retroactive application to cases that were already final when the new rule was announced. But that framework "does not in any way limit the authority of a state court, when reviewing its own state criminal convictions, to provide a remedy for a violation that is deemed ‘nonretroactive’ under Teague ." Id. at 282, 128 S. Ct. at 1042.

We have applied the Teague framework when determining whether we will retroactively apply Supreme Court holdings. See Nguyen II , 878 N.W.2d at 753 ; see also, e.g. , Perez , 816 N.W.2d at 358–60 (using Teague framework when considering whether the Court’s holding in Padilla , 559 U.S. 356, 130 S. Ct. 1473, 176 L.Ed.2d 284, applies retroactively); Bonilla v. State , 791 N.W.2d 697, 700–01 (Iowa 2010) (concluding the Court’s holding in Graham v. Florida , 560 U.S. 48, 130 S. Ct. 2011, 176 L.Ed.2d 825 (2010), applies retroactively under the Teague framework); Morgan v. State , 469 N.W.2d 419, 422–25 (Iowa 1991) (applying Teague framework when considering whether to retroactively apply the Court’s holding in Coy v. Iowa , 487 U.S. 1012, 108 S. Ct. 2798, 101 L.Ed.2d 857 (1988) ); Brewer v. State , 444 N.W.2d 77, 81–82 (Iowa 1989) (relying on Teague framework to determine whether the Court’s holding in Duren , 439 U.S. 357, 99 S. Ct. 664, 58 L.Ed.2d 579, should be given retroactive effect to a case on collateral review). However, we have not yet adopted that framework for evaluating the retroactive effect of our own holdings. See Nguyen II , 878 N.W.2d at 753–54.

There can be little dispute that Plain announced a new rule under the Teague framework as Plain ’s holding on the second Duren prong "was not dictated by precedent existing at the time [Thongvanh]’s [original post]conviction[-relief judgment] became final." Teague , 489 U.S. at 301, 109 S. Ct. at 1070 (emphasis omitted). Plain ’s holding expressly departed from the rule dictated by precedent at the time the judgment in Thongvanh’s original PCR action became final. Additionally, it would be contradictory to conclude Plain announced a new ground of law under section 822.3 while also concluding it did not announce a new rule—but was merely an application of preexisting law—for Teague retroactivity purposes. See Perez , 816 N.W.2d at 360.

On appeal, the court of appeals found that our holding in Brewer foreclosed any relief for Thongvanh. In Brewer , we applied the Teague framework and concluded that the new rule of criminal procedure from the Court’s Duren case, i.e., a modification of the standard applied in fair-cross-section challenges, would not be given retroactive effect to a case on PCR. 444 N.W.2d at 80, 81–82. We found Duren did not create a watershed rule of criminal procedure and relied on the following reasoning from Teague :

Because the absence of a fair cross section on the jury venire does not undermine the fundamental fairness that must underlie a conviction or seriously diminish the likelihood of obtaining an accurate conviction, we conclude that a rule requiring that petit juries be composed of a fair cross section of the community would not be a "bedrock procedural element" that would be retroactively applied under the second exception we have articulated.

Id. at 81–82 (quoting Teague , 489 U.S. at 315, 109 S. Ct. at 1078 ). The court of appeals, here, reasoned that because Duren ’s new fair-cross-section rule was not retroactive in Brewer , Plain ’s new fair-cross-section rule could not be retroactive in Thongvanh’s collateral review case.

Under a straight application of the Teague framework, as we utilized in Brewer , Plain ’s holding would not qualify as a watershed rule. Nothing in subsequent Supreme Court caselaw has directly rejected or abrogated Teague ’s explanation of why a violation of the right to a jury venire drawn from a fair cross section (or a modification to the standard used to determine if there was such a violation) does not undermine the fundamental fairness of the trial or seriously diminish the likelihood of obtaining an accurate conviction. See Teague , 489 U.S. at 315, 109 S. Ct. at 1078 ).

The same analysis applies even if it were argued that Plain error amounts to structural error for which no prejudice need be shown. Structural errors are those "affecting the framework within which the trial proceeds, rather than simply an error in the trial process itself." Neder v. United States , 527 U.S. 1, 8, 119 S. Ct. 1827, 1833, 144 L.Ed.2d 35 (1999) (quoting Arizona v. Fulminante , 499 U.S. 279, 310, 111 S. Ct. 1246, 1265, 113 L.Ed.2d 302 (1991) ). "Such errors ‘infect the entire trial process,’ and ‘necessarily render a trial fundamentally unfair.’ " Id. (first quoting Brecht v. Abrahamson , 507 U.S. 619, 630, 113 S. Ct. 1710, 1717, 123 L.Ed.2d 353 (1993) ; and then quoting Rose v. Clark , 478 U.S. 570, 577, 106 S. Ct. 3101, 3106, 92 L.Ed.2d 460 (1986) ). In contrast, a trial error is "error which occurred during the presentation of the case to the jury, and which may therefore be quantitatively assessed in the context of other evidence presented in order to determine whether its admission was harmless beyond a reasonable doubt." Fulminante , 499 U.S. at 307–08, 111 S. Ct. at 1264. When there is structural error, "the criminal adversary process itself [becomes] ‘presumptively unreliable.’ " Lado v. State , 804 N.W.2d 248, 252 (Iowa 2011) (quoting United States v. Cronic , 466 U.S. 648, 659, 104 S. Ct. 2039, 2047, 80 L.Ed.2d 657 (1984) ). See generally United States v. Gonzalez-Lopez , 548 U.S. 140, 148–49, 126 S. Ct. 2557, 2563–64, 165 L.Ed.2d 409 (2006) (dividing constitutional errors into trial errors, which are subject to harmless-error review, and structural defects, which defy analysis under the harmless error standard).

While neither we nor the Supreme Court has expressly weighed in on whether a violation of the Court’s holding in Taylor —that defendants have a Sixth Amendment right to a jury drawn from a fair cross section of the community—or in Duren —establishing the test for a fair-cross-section claim—qualifies as a structural error, one circuit court has. In United States v. Rodriguez-Lara , the Ninth Circuit concluded, "The selection of a grand or petit jury in violation of either the equal protection or the fair cross-section guarantee is structural error." 421 F.3d 932, 940 (9th Cir. 2005), overruled on other grounds by United States v. Hernandez-Estrada , 749 F.3d 1154, 1164 (9th Cir. 2014). The Ninth Circuit’s conclusion accords with the well-established rule that unlawful discrimination in the jury-selection process is structural error. See, e.g. , Batson v. Kentucky , 476 U.S. 79, 87, 100, 106 S. Ct. 1712, 1718, 1725, 90 L.Ed.2d 69 (1986) (noting harms of racial discrimination in jury selection and holding that a violation of its holding required reversal of the conviction); Vasquez v. Hillery , 474 U.S. 254, 263, 106 S. Ct. 617, 623, 88 L.Ed.2d 598 (1986) ("[W]hen a petit jury has been selected upon improper criteria or has been exposed to prejudicial publicity, we have required reversal of the conviction because the effect of the violation cannot be ascertained."); Duren , 439 U.S. at 370, 99 S. Ct. at 671–72 (establishing fair-cross-section test and reversing without engaging in a harmless-error or prejudice analysis); Taylor v. Louisiana , 419 U.S. 522, 537–38, 95 S. Ct. 692, 701–02, 42 L.Ed.2d 690 (1975) (holding categorically excluding women from juries violates the Sixth Amendment’s fair-cross-section requirement and reversing without engaging in a harmless-error or prejudice analysis); Strauder v. West Virginia , 100 U.S. 303, 305, 312, 25 L.Ed. 664 (1879) (invalidating a state statute that provided only white men could serve as jurors and holding it was error to proceed to trial with such a jury), abrogated on other grounds by Taylor , 419 U.S. at 536 n.19, 537, 95 S. Ct. at 701 & n.19.

However, even if a violation of the fair-cross-section right is structural error, this does not mean that it affects the fundamental fairness of the criminal trial or is "central to an accurate determination of innocence or guilt." Teague , 489 U.S. at 313, 109 S. Ct. at 1077 ; see Weaver v. Massachusetts , 582 U.S. ––––, ––––, 137 S. Ct. 1899, 1908, 198 L.Ed.2d 420 (2017) ("An error can count as structural even if the error does not lead to fundamental unfairness in every case."). As the Court explained in Weaver , there are three general reasons for why a particular error is determined to be structural: (1) "if the right at issue is not designed to protect the defendant from erroneous conviction but instead protects some other interest," such as the defendant’s right to self-representation; (2) "if the effects of the error are simply too hard to measure," such as when a defendant is denied the right to counsel of choice; and (3) "if the error always results in fundamental unfairness," for example, when an indigent is denied the right to an attorney. 582 U.S. at ––––, 137 S. Ct. at 1908.

Here, even if a violation of the right to a fair cross section is a structural error, it would not be of the kind that would protect the defendant from a fundamentally unfair trial. Likewise, a new rule changing how a defendant can challenge a violation of the fair-cross-section right would not implicate the fundamental fairness of the trial and would, therefore, not qualify as a watershed rule.

Under the Teague framework, we conclude that Plain ’s holding on the second prong of the Duren test constitutes a new rule under the Teague framework. However, because it is not a watershed rule of criminal procedure, it does not apply retroactively to cases on collateral review. The Teague framework does not permit Thongvanh to make a Plain claim on PCR.

B. Retroactivity Under the Iowa Constitution’s Due Process and Equal Protection Guarantees. Thongvanh asks us to adopt our own framework for retroactivity under the Iowa Constitution’s due process and equal protection guarantees that provide for broader retroactivity than the Court’s Teague framework. See Iowa Const. art. I, §§ 6, 9. Thongvanh proposes that we use the retroactivity rule from Griffith v. Kentucky , 479 U.S. 314, 107 S. Ct. 708, 93 L.Ed.2d 649 (1987), as the baseline for our constitutional retroactivity framework.

In Griffith , the Court held that

a new rule for the conduct of criminal prosecutions is to be applied retroactively to all cases, state or federal, pending on direct review or not yet final, with no exception for cases in which the new rule constitutes a ‘clear break’ with the past.

479 U.S. at 328, 107 S. Ct. at 716. At issue there was the retroactivity of the holding in Batson , 476 U.S. 79, 106 S. Ct. 1712, 90 L.Ed.2d 69, to cases "pending on direct state or federal review or not yet final when Batson was decided." Griffith , 479 U.S. at 316, 107 S. Ct. at 709. However, Thongvanh’s criminal conviction and sentence became final long ago, and thus he would not benefit from the Griffith holding.

In essence, Thongvanh is asking us to apply the same retroactivity rule for cases that are not final at the time the new rule is announced to cases that have become final at the time the new rule is announced. In support of this request, he appears to argue that people whose convictions have become final at the time the new rule is announced are similarly situated to people whose convictions have not become final.

He notes that Plain and empirical evidence recognize that underrepresentation of minorities in jury pools can affect trial outcomes. See Plain , 898 N.W.2d at 826. He appears to argue that, therefore, any person denied the application of Plain is similarly situated—regardless of whether that person’s conviction had become final at the time Plain was decided.

We are not persuaded by his argument. The Griffith Court expressly acknowledged that in United States v. Johnson , 457 U.S. 537, 102 S. Ct. 2579, 73 L.Ed.2d 202 (1982), it "concluded that the retroactivity analysis for convictions that have become final must be different from the analysis for convictions that are not final at the time the new decision is issued." Griffith , 479 U.S. at 321–22, 107 S. Ct. at 712. And it noted that the Johnson Court largely adopted the rationale for distinguishing between cases that had become final and those that had not as explained in Justice Harlan’s separate opinions in Desist v. United States , 394 U.S. 244, 256, 257–58, 89 S. Ct. 1030, 1038, 22 L.Ed.2d 248 (1969) (Harlan, J., dissenting), and Mackey v. United States , 401 U.S. 667, 679–81, 91 S. Ct. 1171, 1173–74, 28 L.Ed.2d 404 (1971) (Harlan, J., concurring in the judgment). Griffith , 479 U.S. at 322, 107 S. Ct. at 712–13. Thongvanh makes no attack on Justice Harlan’s rationale or the Johnson Court’s decision to largely adopt that rationale. Thus, Thongvanh does not explain why persons whose cases had become final are similarly situated to persons whose cases had not become final.

We recognize that the composition of jury pools can have real-world effects. That is why we changed the law in Plain . In fact, since 1984, when Thongvanh was tried and convicted, Iowa’s criminal justice system has evolved in many ways—hopefully for the better. We believe if Thongvanh were tried today, thirty-six years later, he would receive better procedural protections on the whole. But against this consideration, we have to weigh the need for finality of judgments when the issue does not bear directly on guilt or innocence and the impracticality of reconstructing events that occurred between three and four decades ago. We are not persuaded that either the Teague framework or the Iowa Constitution’s due process and equal protection guarantees require Plain to apply retroactively to convictions that were already final at the time we decided Plain . The district court did not err in concluding Thongvanh could not rely on Plain as the basis for his current PCR application.

VI. Disposition.

We conclude Plain ’s holding on the second prong of the Duren test constitutes a new ground of law under section 822.3. However, because we find that the new law of criminal procedure announced in Plain does not apply retroactively to cases on collateral review, we affirm the dismissal of Thongvanh’s PCR application.

AFFIRMED.

All justices concur except McDonald, J., who takes no part.