Defendant appeals as of right the trial court’s decision on remand, which found that defendant’s Sixth Amendment right to an impartial jury drawn from a fair cross section of the community was not violated because African-Americans were not underrepresented in the venire from which defendant’s jury was selected and Kent County’s jury-selection process, at the time of defendant’s trial, did not system*264atically exclude African-Americans. For the reasons set forth in this opinion, we reverse and remand for a new trial.
I. FACTS AND PROCEDURAL HISTORY
Defendant was convicted of first-degree criminal sexual conduct, MCL 750.520b(1)(e), armed robbery, MCL 750.529, and possession of marijuana, MCL 333.7403(2)(d), by a jury in the Kent Circuit Court in February 2002. He appealed, arguing, in part, that he was deprived of his Sixth Amendment right to an impartial jury drawn from a fair cross section of the community because there was only one African-American in the jury venire of 42 people. In an unpublished opinion, we affirmed in part and remanded “for the sole purpose of conducting an evidentiary hearing regarding defendant’s challenge to the jury venire.” People v Bryant, unpublished opinion per curiam of the Court of Appeals, issued March 16, 2004 (Docket No. 241442), p 7.
On remand, the trial court held several evidentiary hearings and issued a written opinion. The trial court rejected defendant’s reliance on statistical estimates, reasoning that they were not sufficient to prove under-representation. The trial court made four holdings: that “defendant has failed to sustain his burden of proving that African-Americans were substantially underrepresented among the prospective jurors to whom questionnaires were mailed in 2001-2002,” that “even if African-Americans were numerically underrepresented from June, 2001, through mid-Fail, 2002, among prospective jurors, defendant has failed to establish that the circumstances were such that that underrepresentation was unconstitutional as defined by the Supreme Courts of the United States and Michigan,” that “even if there *265was unconstitutional underrepresentation in the total number of prospective jurors, there was no underrepresentation of African-Americans in the venire from which defendant’s jury was selected,” and, finally, that “any underrepresentation was the product of chance, not any bias, even an innocent and accidental bias, in the jury selection process. Hence, systematic exclusion has not been proven.”
Defendant appeals again, arguing that he was denied his Sixth Amendment right to be tried by an impartial jury drawn from a fair cross section of the community because there was only one African-American in the jury venire of 42 people.
II. STANDARD OF REVIEW
We review de novo questions regarding systematic exclusion of minorities from jury venires. People v Hubbard (After Remand), 217 Mich App 459, 472; 552 NW2d 493 (1996).
III. ANALYSIS
The issue in this case is whether defendant was denied his Sixth Amendment right to be tried by an impartial jury drawn from a fair cross section of the community because there was only one African-American in the jury venire of 42 people.
The Sixth Amendment applies to the states through the Due Process Clause of the Fourteenth Amendment. Duncan v Louisiana, 391 US 145, 149; 88 S Ct 1444; 20 L Ed 2d 491 (1968). In addition, the Michigan Constitution guarantees the right to trial by jury. Const 1963, art 1, § 14. In Taylor v Louisiana, 419 US 522, 528; 95 S Ct 692; 42 L Ed 2d 690 (1975), the United States Supreme Court stated “that the selection of a petit jury *266from a representative cross section of the community is an essential component of the Sixth Amendment right to a jury trial.” While the “fair-cross-section requirement does not entitle the defendant to a petit jury that mirrors the community,” it “guarantees an opportunity for a representative jury by requiring that jury wheels, pools of names, panels, or venires from which juries are drawn must not systematically exclude distinctive groups in the community and thereby fail to constitute a fair cross section of the community.” Hubbard, 217 Mich App at 472-473.
In Duren v Missouri, 439 US 357, 364; 99 S Ct 664; 58 L Ed 2d 579 (1979), the United States Supreme Court articulated the showing that a defendant must make to establish a prima facie violation of the Sixth Amendment fair-cross-section requirement:
(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.
Once a defendant establishes a prima facie violation of the fair-cross-section requirement, “the government may overcome the right to a proper jury by proffering a significant state interest that manifestly and primarily advances those aspects of the jury selection process that would result in the disproportionate exclusion of a distinctive group . . . .” Hubbard, 217 Mich App at 473; see also Duren, 439 US at 367-368.
As we observed in our previous opinion in this case, defendant satisfied the first prong of Duren because “African-Americans are considered a constitutionally *267cognizable group for Sixth Amendment fair-cross-section purposes.” Hubbard, 217 Mich App at 473.
The second prong of Duren “is satisfied where it has been shown that a distinctive group is substantially underrepresented in the jury pool.” Id. at 474. Although it recently had the opportunity to specify the preferred method of measuring if representation of a distinctive group in the jury pool is fair and reasonable, the United States Supreme Court has not done so. See Berghuis v Smith, 559 US _, _; 130 S Ct 1382, 1393-1394; 176 L Ed 2d 249, 261 (2010) (“[W]e would have no cause to take sides today on the method or methods by which underrepresentation is appropriately measured.”). In People v Smith, 463 Mich 199, 203; 615 NW2d 1 (2000), our Supreme Court observed that federal courts since Duren have applied three different tests to measure whether representation of a distinctive group in the jury pool is fair and reasonable: the absolute-disparity test, the comparative-disparity test, and the standard-deviation test. Recognizing that all three tests are subject to criticism, our Supreme Court stated the following regarding the appropriate method to measure underrepresentation:
We thus consider all these approaches to measuring whether representation was fair and reasonable, and conclude that no individual method should be used exclusive of the others. Accordingly, we adopt a case-by-case approach. Provided that the parties proffer sufficient evidence, courts should consider the results of all the tests in determining whether representation was fair and reasonable. [Id. at 204.]
Because the United States Supreme Court did not adopt a specific test to measure underrepresentation in Berghuis, we are bound to follow the case-by-case approach articulated by our Supreme Court in Smith. *268On remand, there was evidence offered regarding all three tests. We will therefore address each test in turn.
The absolute-disparity test measures the difference between the percentage of the distinctive group in the population eligible for jury duty and the percentage of that group who actually appear in the venire. Ramseur v Beyer, 983 F2d 1215, 1231 (CA 3, 1992). This Court has previously recognized that the absolute-disparity method of measuring underrepresentation is of questionable usefulness when applied to a group that makes up a small percentage of the population, Hubbard, 217 Mich App at 476-477, and in this case African-Americans who were 18 years of age or older made up a small percentage of the Kent County population when defendant’s jury was selected. The evidence indicated that 8.25 percent of eligible voters in Kent County were African-American. This Court has stated that an absolute disparity between 2 percent and 11.2 percent is statistically insignificant and does not constitute substantial underrepresentation. Id. at 475; see also Ramseur, 983 F2d at 1232 (disparities of 2 percent to 11.5 percent do not constitute substantial underrepresentation). Populations that fall within this percentage range can never be statistically significant because “the percentage disparity can never exceed the percentage of African Americans in the community.” United States v Rogers, 73 F3d 774, 776-777 (CA 8, 1996). Even if the Kent County juror-selection system excluded all African-Americans from jury service, a successful Sixth Amendment fair-cross-section challenge would be impossible because the total percentage of African-American voters in the Kent County population constitutes a percentage that is less than that which is considered statistically significant for Sixth Amendment fair-cross-section purposes. See Hubbard, 217 Mich App at 477.
*269Dr. Paul Stephenson, chairman of the Department of Statistics at Grand Valley State University, calculated the absolute disparity in this case to be 6.031 percent; however, he observed that an analysis of absolute disparity is not a viable method of identifying or measuring underrepresentation of the African-American community in this case for the reasons already explained in this opinion. While the absolute disparity of 6.03 percent in this case does not indicate substantial under-representation, we conclude, like we did in Hubbard, “that the absolute disparity test is an ineffective measure of acceptable disparity” because of the low percentage of African-Americans who were eligible to vote in Kent County. Id. Thus, we decline to find the absolute-disparity test controlling in this case.
The comparative-disparity test “measures the diminished likelihood that members of an underrepresented group, when compared to the population as a whole, will be called for jury service.” Ramseur, 983 F2d at 1231-1232. The diminished likelihood is calculated by dividing the absolute disparity by the percentage of the population made up by the distinctive group in question. Dr. Stephenson applied the comparative-disparity test to the venire from which defendant’s jury was chosen and calculated the comparative disparity to be 73.1 percent. This means that the venire for defendant’s trial had 73.1 percent fewer African-Americans than could have been expected in Kent County. In rendering his calculation, Dr. Stephenson relied on the 2000 United States Census, which indicated that 8.25 percent of the population of Kent County 18 years of age or older was African-American.
*270In our previous opinion in this case, this Court articulated its belief that the comparative-disparity test was not controlling because of the fact that the population of African-Americans in Kent County was small and therefore a small change in the jury pool would distort the proportional representation. Bryant, unpub op at 3. We acknowledge the difficulties in applying this method to a group that makes up a small percentage of the population.
In choosing the appropriate test to apply in this case, we are mindful that “[e]ach test is imperfect.” Berghuis, 559 US at _; 130 S Ct at 1393; 176 L Ed 2d at 261. We are further cognizant that some of the concerns with applying the comparative-disparity test to a group that makes up a small percentage of the population also exist with applying the absolute-disparity test, Hubbard, 217 Mich App at 476-477, that courts typically apply the standard-deviation test in Fourteenth Amendment cases, but not in Sixth Amendment cases, and that no court in the country has accepted application of the standard-deviation test alone as determinative in Sixth Amendment challenges to jury-selection systems, Smith, 463 Mich at 204. We must apply some test to measure the representation of African-Americans in defendant’s venire, and the comparative-disparity method at least yields a calculation that is indicative of the underrepresentation of African-Americans in defendant’s venire. We agree with the United States Court of Appeals for the Eighth Circuit that as between the absolute- and comparative-disparity tests, the comparative-disparity test is most appropriate to measure underrepresentation in cases in which the percentage of African-Americans in the relevant community is low. Rogers, 73 F3d at 776-777. In Rogers, the Eighth Circuit opined:
*271While ... both [absolute and comparative disparity] provide a simplified statistical shorthand for a complex issue, the comparative disparity calculation provides a more meaningful measure of systematic impact vis-a-vis the “distinctive” group: it calculates the representation of African Americans in jury pools relative to the African-American community rather than relative to the entire population. [Id. at 777.]
For the reasons we have just outlined, we conclude that the comparative-disparity test is the most appropriate test to measure underrepresentation in this case.2
Seventy-three and one-tenth percent is a significant comparative disparity and is sufficient to demonstrate that the representation of African-Americans in the venire for defendant’s trial was unfair and unreasonable. See id. (holding that comparative disparity of more than 30 percent satisfies the second prong of Duren)-, see also Smith, 463 Mich at 219 (CAVANAGH, J., concurring) (stating that comparative disparities of 40 percent have been held to be borderline). In any event, the comparative disparity in this case, 73.1 percent, is substantially higher than the 30 or 40 percent that has been deemed sufficient to demonstrate an unfair and unreasonable representation of minorities in a jury *272venire. Thus, under the comparative-disparity test, defendant has established that African-Americans were underrepresented on the venire from which his jury was selected.
On remand, there was some evidence regarding the standard-deviation test, which explains the probability that any disparity was the result of random chance. Smith, 463 Mich at 219 (CAVANAGH, J., concurring). Standard deviation is calculated “by multiplying the number of prospective jurors in the jury pool by the percentage of the distinct group in the population by the percentage of the population that is not in the distinct group, and then taking the square root of that product.” Id. at 220. At one of the postremand evidentiary hearings, Dr. Chidi Chidi, an expert presented by defendant, testified that the appearance of one African-American in the venire failed the standard-deviation test. In a report that was admitted into evidence on remand, Dr. Chidi calculated the standard deviation to be 27.86. This figure is close to the standard deviation of 29 condemned in Castaneda v Partida, 430 US 482, 496 n 17; 97 S Ct 1272; 51 L Ed 2d 498 (1977). However, Dr. Stephenson’s opinion regarding use of the standard-deviation test was that the “test uses a normal approximation of a binomial random variable. In this case, the normal approximation is not valid, and therefore, the standard deviation test is not appropriate.” Furthermore, our Supreme Court has noted that the standard-deviation test is not typically used in Sixth Amendment cases and that “ ‘no court in the country has accepted [a standard-deviation analysis] alone as determinative in Sixth Amendment challenges to jury selection systems.’ ” Smith, 463 Mich at 204, quoting United States v Rioux, 97 F3d 648, 655 (CA 2, 1996). For these reasons, we conclude that in this case, the standard-*273deviation test has little value in measuring the under-representation of African-Americans in Kent County jury venires.
The third prong of Duren requires proof that under-representation of African-Americans “is due to systematic exclusion of the group in the jury-selection process.” Duren, 439 US at 364. Systematic exclusion is exclusion “inherent in the particular jury-selection process utilized.” Id. at 366. Systematic exclusion is not shown by one or two incidents of disproportionate venires. Hubbard, 217 Mich App at 481. In Duren, the United States Supreme Court concluded that underrepresentation of women in every weekly venire for nearly a year constituted underrepresentation that was systematic. Id. at 366.
In this case, there was evidence of a significant problem with the jury-selection process, and the prosecution has conceded that this problem lasted for a significant duration. The problem with the jury-selection process in Kent County was twofold. First, the Secretary of State provided Kent County with a list of 453,414 individuals who were eligible to vote in Kent County, but the information technology department of the Kent Circuit Court erroneously reduced this list to only 118,000 individuals. Second, it is not disputed that a computer program used in Kent County did not select jurors at random across all zip codes, as it was supposed to do. As a result of the problem with the computer program, jurors were overselected from zip codes with small minority populations and underselected from zip codes with large minority populations. The prosecution previously conceded that “there was indeed a problem in the jury selection process in Kent County which occurred from late 2001 to July 2002,” explaining:
*274“[I]n essence... a computer program used to select potential jurors chose a disproportionately large number of jurors from areas with lower zip codes, which had the unintended effect of selecting fewer jurors from areas of the county where African-Americans live. The assumption is that this led to an artificial shortfall of African-American jurors, though to what extent has never been determined.” [Bryant, unpub op at 4.]
We find that the underrepresentation in this case was the result of the system by which juries in Kent County were selected because jurors from zip codes with small minority populations were routinely overselected and jurors from zip codes with large minority populations were routinely underselected as the result of a glitch or problem with the computer program that selected jurors. Given this problem with the computer program, underrepresentation was inherent in the jury-selection process used in Kent County during the time that the computer glitch existed. It is irrelevant that the problem with the computer program’s failing to randomly select jurors across all zip codes does not appear to be intentional. A party need not show that the underrepresentation of a distinctive group came as a result of intentional discrimination. Duren, 439 US at 368 n 26.
Moreover, there was evidence that the error began in April 2001 and persisted over a period of 16 months. Terry Holtrop, the case-management manager for the Kent Circuit Court, testified that he became aware in April 2001 that there was a problem of underrepresentation of minorities on Kent County juries. Gail VanTimmeren, the jury clerk for the Kent Circuit Court, testified that it was “visually evident” that there were not enough minorities coming in for jury duty and that she had spoken to the administrator “over and over again” about this. VanTimmeren asserted that on a number of occasions, she handpicked individuals who *275appeared to be African-American to be placed on a panel from which a jury would be selected. She asserted that “we significantly, in every single week, were not getting minorities in, and something was wrong.”
There was also testimony and statistical evidence3 presented by Dr. Stephenson that supports our conclusion that the underrepresentation of African-Americans on Kent County jury venires occurred over a significant period. Dr. Stephenson examined census data and determined that in the vast majority of the zip codes that were overrepresented, there were a small number of African-Americans, and that in the zip codes that were underrepresented, there were a large number of African-Americans. Furthermore, Dr. Stephenson testified that “the way that the process was performing did, in effect, over the long run, create a situation where black or African-Americans were going to be underrepresented, in my opinion, in the compilation of jury venires.” Dr. Stephenson also provided data and statistics that permit this Court to calculate the comparative disparity for the three-month period of January through March 2002 at 49.5 percent. This is higher than the 30 percent found to satisfy the second prong of Duren in Rogers, 73 F3d at 777.
In sum, we conclude that defendant has established a prima facie violation of the Sixth Amendment’s fair-cross-section requirement. Because defendant established a prima facie violation, the burden shifts to the prosecution to demonstrate that “a significant state interest [is] manifestly and primarily advanced by those aspects of the jury-selection process . . . that result in *276the disproportionate exclusion of” African-Americans. Duren, 439 US at 367-368. The prosecution filed a brief the day before oral argument in this case but failed to proffer any significant state interest that would be advanced by the errors and computer glitch that resulted in the systematic underrepresentation of African-Americans in Kent County jury venires. In fact, we cannot conceive of any significant state interest that could possibly justify the jury-selection process used in Kent County during the time the computer glitch systematically excluded African-Americans from jury venires.
Reversed and remanded for a new trial before an impartial jury that is drawn from a fair cross section of the community. We do not retain jurisdiction.
This number represents the percentage of Kent County African-American residents who were 18 years of age or older (8.25 percent) minus the percentage of African-Americans appearing in defendant’s venire (2.22 percent).
To the extent that the previous unpublished opinion in this case concluded that the comparative-disparity test is not controlling, we find that the law-of-the-case doctrine does not preclude this Court from concluding, after remand, that the comparative-disparity test is the appropriate test to measure underrepresentation under the facts of the case. First, the law-of-the-case doctrine applies only if the facts remain substantially or materially the same, People v Phillips (After Second Remand), 227 Mich App 28, 32; 575 NW2d 784 (1997), and in this case the trial court conducted several evidentiary hearings on remand, which yielded significant expert testimony regarding each of the three tests. Second, the law-of-the-case doctrine does not limit an appellate court’s power, but is instead a discretionary rule of practice. Schumacher v Dep’t of Natural Resources, 275 Mich App 121, 128; 737 NW2d 782 (2007).
The trial court’s conclusion that statistics are inadequate to demonstrate underrepresentation is incorrect. In Duren, 439 US at 364, the United States Supreme Court found that “[t]he second prong of the prima facie case was established by petitioner’s statistical presentation.”