selection procedure of one judge, which permitted viewing of potential forepersons prior to selection, “clearly” satisfied third requirement of prima facie test
How later courts described this case
- selection procedure of one judge, which permitted viewing of potential forepersons prior to selection, “clearly” satisfied third requirement of prima facie test
- "The statute lists eight possible deficiencies in state court fact-findings (rather than the Townsend six); but their relevance is to the question whether the state findings are to be 'presumed' correct. Further, if none of the eight deficiencies is shown, the effect of this is, not to negate the power of the judge to call for a hearing, but, confusingly, to shift to the petitioner the burden to show at a hearing that the state findings were erroneous."
- "Townsend was not, however, completely superseded by the [1966] amendment, for the Supreme Court decided when a federal evidentiary hearing is mandatory while the habeas corpus statute, as amended, merely establishes a presumption that the state court judgment is correct unless the applicant establishes one of a number of specific reasons to disregard it.”
- "The statute lists eight possible deficiencies in state court fact-findings (rather than the Townsend six
Written by the judges who cited it.
The opinion
REAVLEY, Circuit Judge,
with whom RONEY, GEE and FAY, Circuit Judges, join, dissenting:
Vigorous counsel for Guice and Claxton had an unimpeded opportunity when they were before the state court to make any available proof of racial discrimination in the selection of the grand jury foreman. They requested no further hearing in the federal district court. They made no complaint on that ground when they appealed to this court. After a panel held that their proof was lacking, on rehearing they then asked for an evidentiary hearing in which they could try again.
The en banc court, adopting a statistical model based on population figures that would establish a prima facie case of discrimination if no more than thirteen black grand jury foremen were chosen out of the last thirty-one, now grants their request— laboring to justify the grant by saying that counsel must have failed to make their proof in the state court because of “the confused nature of the presentation of the claims to the trial court.” 661 F.2d at 506.
I would not mandate an evidentiary hearing in the federal court after one has already been held in the state court unless the petitioner asks for it. To my mind, the record before the district judge did not necessitate an additional hearing. I would affirm his judgment.
Moreover, the statistical analysis adopted by the majority is mistaken. They suggest that the petitioners can prove that blacks have been underrepresented in the selection of grand jury foremen in Madison Parish over a significant period of time, see Rose v. Mitchell, 433 U.S. 545 , 99 S.Ct. 2993 , 61 L.Ed.2d 739 (1979) (quoting Castenada v. Partida, 430 U.S. 482, 495 , 97 S.Ct. 1272, 1280 , 51 L.Ed.2d 498 (1977)), 1 by adopting as *509 part of their proof an improper statistical model based on the cumulative binomial probability distribution. Under this model, the petitioners would need to put into evidence only the number of times a grand jury, foreman was selected in Madison Parish and the racial composition, percent black or percent non-black, of the parish in order to demonstrate the probability that a specified outcome — in the instant case the successive selection of thirty-one non-black grand jury foremen — will occur.
Traditionally, in lawsuits alleging discrimination, if the probability of an outcome occurring is less than .05, one out of twenty, and no other explanation for the result is available, the courts have ruled out the possibility that the event was due to chance alone and have inferred that the outcome was due to impermissible discrimination in the selection process. See D. Baldus & J. Cole, Statistical Proof of Discrimination §§ 9.02, 9.03 (1980). The majority follow this approach and take for the pool from which grand jury foremen are chosen the population composed of 60% blacks and 40% non-blacks of Madison Parish, Louisiana. They calculate that the probability of ten successive selections of a non-black from this population is approximately 10,-000 to one. They then conclude that if Judge Adams had made thirty-one selections, “absent a racially discriminatory motivation [a] qualified black person could have been found and appointed” grand jury foreman. 2 661 F.2d at 505.
In reaching this conclusion, the majority necessarily make one of two assumptions. Either they assume that no legitimate reason exists to choose one venireman rather than another to serve as foreman and, as a result, the choice of a black or a non-black foreman should occur at random. Or they assume that on each of the past thirty-one grand jury venires the proportion of blacks who are as qualified to serve as foreman as the most qualified person on the venire, and are thus eligible for service as foreman, is constant and is the same as the proportion of blacks in Madison Parish. The majority, however, are justified in assuming neither. Louisiana has decided to select from the grand jury venire the most qualified person to serve as foreman. Thus, because the selection of a foreman is based on qualifications, there is no reason to presuppose that the selection of blacks and non-blacks to serve as foreman will occur at random. Moreover, there is no reason to assume that on each of the previous thirty-one venires the proportion of qualified blacks who are eligible for selection as grand jury foreman is constant and equals the proportion of blacks in the population of Madison Parish.
The distribution upon which the majority builds its statistical model, the cumulative binomial probability distribution, is, in turn, based on a Bernoulli process. A Bernoulli *510 process has four characteristics. First, an event can have only two possible outcomes. Second, the probability of both outcomes is constant. Third, the occurrence of a particular outcome in one trial cannot affect the outcome in a subsequent trial. And, fourth, successive outcomes occur without any fixed pattern. See C. Clark & L. Schkade, Statistical Analysis for Administrative Decisions 89 (3d ed. 1979).
Continuous tosses of a coin is the classical example of a Bernoulli process. A flip of a coin answers in either heads or tails. The chance of either a head or a tail on any toss is constant. No result on a prior toss affects a subsequent toss. 3 Outcomes of heads or tails occur at random.
The cumulative binomial probability distribution is defined by two parameters. One parameter is a proportion, usually represented by the letter p, that indicates the probability of either one of the two possible outcomes of a Bernoulli trial. The other parameter, usually represented by the letter n, indicates the number of trials. C. Clark & L. Schkade, Statistical Analysis for Administrative Decisions 89, 93 (3d ed. 1979).
Absent illegal discrimination or the use of impermissible selection criteria, employment decisions and selection for service on a jury venire should approximate a Bernoulli process. The person hired will be either black or non-black. If racially neutral, the hiring or selection of one person should not affect whether the next person hired is black or not. Also, if the choices are made from the same pool of eligible persons without regard to race, the probability of choosing a black or a non-black is constant for each choice and should occur at random.
Moreover, in hiring unskilled labor and in drawing at random for service on a jury venire, both of the parameters of the cumulative binomial probability distribution are easily determined. The parameter representing the number of persons selected for employment or jury service, a, is known. The parameter representing the proportion of blacks and non-blacks in the eligible population from which the selections are made, p, is easily measured.
Because hiring unskilled labor and making random selections for jury service should resemble a Bernoulli process and both parameters of the distribution are determinable, the cumulative binomial probability distribution is often used by statisticians to demonstrate illegal discrimination in the selection process in those settings. When we examine a discrepancy between the racial make-up of the jurors or laborers chosen and the pool of eligibles, the explanation is between two alternatives: either the selection procedure was biased or the outcome was due to chance alone. If the discrepancy is so great as to occur only one time out of twenty, the .05 level, the courts infer that the disparity is due to an impermissible selection process. See D. Baldus & J. Cole, Statistical Proof of Discrimination §§ 9.02, 9.03 (1980).
In building its legal analysis on a statistical model based on the cumulative binomial probability distribution, the majority premise that the selection of grand jury foremen in Madison Parish resembles a Bernoulli process. They also premise that the value of the p parameter is equal to .6, an eligible population composed of 60% blacks, 4 and that the value of the n parameter is *511 equal to 31, the number of times a grand jury foreman was selected in Madison Parish. The majority then use the cumulative binomial probability distribution to show the extreme improbability of selecting thirty-one successive white grand jury foremen and conclude that “the record demonstrated the likelihood that [the petitioners] could prove facts entitling them to relief.” 661 F.2d 506 .
The selection of grand jury foremen in Madison Parish and throughout most of Louisiana has some resemblance to a Bernoulli process. A grand jury foreman is either black or non-black. The selection of a grand jury foreman from one grand jury venire does not significantly affect the selection of a foreman from a subsequent grand jury venire. 5 Because only one foreman is selected from each venire, the probability of choosing a black or a non-black is constant for each venire.
The similarity to a Bernoulli process, however, ends there. Because Louisiana has decided to select their grand jury foremen based on their qualifications to serve, there is nothing inherent in the selection of a Louisiana grand jury foreman that suggests randomness so that the racial makeup of the past thirty-one foremen should occur without any fixed pattern and statistically mirror the population at large. So, when the majority assume that absent racial discrimination the proportion of blacks who have served as grand jury foremen should be statistically equivalent to the black representation in Madison Parish, they take a step too far.
Louisiana law does not comtemplate the selection of the grand jury foreman at random. Rather, it places in the district judge the responsibility to select the most qualified person to serve. This scheme is demonstrated by the Code of Criminal Procedure. The Code provides that “the court shall select one person from the grand jury venire to serve as foreman.” La.Code Crim. Pro.Ann. art. 413 (West 1967). By contrast, it then provides for the selection of the other members of the grand jury at random when it appoints the sheriff to “draw indiscriminately and by lot from the envelope containing the remaining names on the grand jury venire a sufficient number . . . to complete the grand jury.” Id.
This scheme of entrusting the selection of the foreman to the judge, so that the most qualified person on the grand jury venire may be chosen as foreman, while leaving the remainder of the grand jury to chance, is, in light of the responsibilities and powers of the grand jury, a legitimate choice for Louisiana to make. The Louisiana grand jury is a puissant body. It plays a key role in the instigation of criminal prosecutions by providing the accused with “an independent determination of probable cause,” State v. Williams, 310 So.2d 528, 533 (La. 1975) (emphasis in original), prior to returning a true bill, see La.Code Crim.Pro. Ann. arts. 443-444 (West 1967). In fulfilling its responsibilities, the Louisiana “grand jury may, on its own initiative, investigate any matter coming to its attention, from whatever source,” Hewitt v. Webster, 118 So.2d 688, 693 (La.App.1960), and decide for itself the times and places of its meetings throughout the parish, La.Code Crim.Pro. Ann. art. 435 (West Supp.1981). It has the power to subpoena witnesses to appear before it. Id. art. 439 (West 1967). The willful failure of a person under subpoena to reappear before the grand jury as directed by the foreman may constitute contempt of court. In re Grand Jury Subpoenas, 363 *512 So.2d 651, 655 (La.1978). To increase its effectiveness, Louisiana law provides that “the proceedings of the grand jury [are] cloaked in secrecy,” In re Grand Jury Subpoenas, 387 So.2d 1140, 1141 (La.1980); see La.Code Crim.Pro.Ann. art. 434 (West Supp. 1981), and forbids a witness called before the grand jury to appear with counsel, 387 So.2d at 1142 . See generally Comment, Selected Problems of the Louisiana Grand Jury, 52 Tul.L.Rev. 707, 719-31 (1978).
In sum, there is neither anything random about the process of selecting a grand jury foreman in Louisiana, nor any reason for the majority to indulge in the presupposition that the selection of grand jury foremen should, absent discrimination, produce random results. Louisiana has decided to choose the best possible person from the grand jury venire to serve as foreman. That determination alone refutes the assumption of randomness that underlies a Bernoulli process and disallows an inference of racial discrimination by statistically testing the numbers against the cumulative binomial probability distribution.
The majority violate another tenet of a Bernoulli process when they assume that a value of the p parameter equal to the proportion of blacks in Madison Parish is the same as the probability of selecting a black for service as a grand jury foreman from each of the thirty-one grand jury venires. Even accepting the majority’s baseless assumption that all persons in Madison Parish are equally eligible to be selected by the district judge as a foreman, the use of a p equal to the percentage of blacks in the parish population is simply an improper use of statistics. Selections for grand jury foreman are made from the grand jury venire, not from the population of Madison Parish. Variability may exist between the proportion of blacks on the different grand jury venires; variability also may exist between the proportion of blacks on the venires and in the parish population. The proportion of blacks on a grand jury venire is, of course, equal to the probability of selecting a black to serve as a foreman using a racially neutral selection process. Because these probabilities will differ from one venire to the next, however, the requirement of a Bernoulli process that the probability of an outcome remain constant throughout all selections is violated. Only with thirty-one separate values of the p parameter, each equal to the proportion of blacks on one of the thirty-one separate grand jury venires, can any valid statistical calculation be made.
Moreover, there is simply no reason to believe, as the majority assume, that a value of p equal to the proportion of blacks in the population of Madison Parish represents the probability of selecting a black from all of the thirty-one venires. Because Louisiana has decided to place the selection of the grand jury foreman in the district judge so that the most1 qualified person is selected, a sub-population of persons most qualified for service as a foreman exists on each grand jury venire. It is from this sub-population composed of veniremen who are eligible for selection as foreman that the p parameter must be determined. We pointed this out in United States ex rel. Barksdale v. Blackburn, 639 F.2d 1115, 1124 (5th Cir. 1981) (en banc), when we said, “statistics describing the presumptively eligible black juror population, rather than the general black population, provide the proper starting point for an inquiry into racial disparities in the Parish [venires].” 6
As applied to the instant case, Barksdale requires the petitioners to show the proportion of blacks, on each of the thirty-one *513 previous grand jury venires, who were as qualified to serve as foreman as the most qualified person on the grand jury venire. Only with this data can the statistician determine the p parameters to calculate the probability of the choice of thirty-one successive non-black grand jury foremen in Madison Parish. And this data is simply unknown and unknowable. 7
Thus, the majority’s calculation of an extremely small number to represent the probability of selecting thirty-one consecutive white grand jury foremen does not suggest the possibility that the petitioners have suffered from discrimination in the exclusion of blacks as grand jury foremen inasmuch as it reflects the use of an inappropriate statistical model. The consequences of the majority’s decision reveal the error in their analysis. If, as in Madison Parish, thirty-one random selections for grand jury foremen are made from a population that is 60% black, the probability is one in 10,000 that less than nine blacks will be chosen; the probability is less than one in 2000 that less than ten blacks will be chosen; and the probability is one in 200 that less than twelve blacks will be chosen. The prima facie case is made if not more than thirteen blacks have served as grand jury foremen! 8
I end this statistical discourse as the majority ends theirs: declaring the inappropriateness of these statistics. They then fall back to opine that in these years, “absent racially discriminatory motivation ... [a] qualified black person could have been found and appointed” grand jury foreman. Majority opinion at 505. I regard this speculation as wide as their statistics from justifying an additional hearing for the petitioners. I would affirm the judgment of the district court.
. Under Castenada v. Partida, 430 U.S. 482 , 97 S.Ct. 1272 , 51 L.Ed.2d 498 (1977), a claimant who alleges an equal protection violation because members of his race have been systematically excluded from grand jury service must prove three elements of a prima facie case.
First, . . . that the group [to which he belongs] is one that is a recognizable, distinct class, singled out for different treatment under the laws, as written or as applied .... Next, the degree of underrepresentation must be proved .... Finally ... a selection procedure that is susceptible of abuse or is not racially neutral supports the presumption of discrimination ....
*509 Id. at 494 , 97 S.Ct. at 1280 . I, of course, agree with the majority that the petitioners have proved the first and third elements of their prima facie case. I said so in the prior panel opinion.
Since both petitioners are black, there is, therefore, “no question .. . that .. . [they] are members of a group recognizable as a distinct class capable of being singled out for different treatment under the laws.” Moreover, the method of selecting grand jury foremen in Madison Parish is certainly a “procedure that is susceptible of abuse.”
Guice v. Fortenberry, 633 F.2d 699, 704 (5th Cir. 1981) (quoting Rose v. Mitchell, 443 U.S. 545 , 99 S.Ct. 2993 , 61 L.Ed.2d 739 (1979)) (citation omitted).
. The issue in this appeal is not whether Judge Adams could have selected one or more qualified black persons to serve as grand jury foreman any one of the thirty-one times he appointed a new foreman. Rather, the issue in this appeal is whether the petitioners have made out a prima facie case of an equal protection violation by showing that the procedures used to select grand jury foremen have resulted in a substantial underrepresentation of their race. Rose v. Mitchell, 433 U.S. 545 , 99 S.Ct. 2993 , 61 L.Ed.2d 739 (1979) (quoting Castenada v. Partida, 430 U.S. 482, 494 , 97 S.Ct. 1272, 1280 , 51 L.Ed.2d 498 (1977)). Whether persons of one race or another “could” have been selected for service as grand jury foremen plays no part in this determination. Indeed, the majority recognize as much at the outset of their opinion when they focus on whether “blacks were systematically excluded from service as grand jury foremen both on the grand jury that indicted [the petitioners] and on prior grand juries.” 661 F.2d at 497.
. A familiar example of the occurrence of one outcome affecting a subsequent outcome is the successive drawing of playing cards from a deck without replacing the selected card prior to the next draw. With a standard deck of 52 cards, the probability of choosing a black card, a spade or a club, on the initial draw is .5, or one out of two. Conversely, the probability of choosing a non-black card is also .5, or one out of two. If, however, the selected card is not returned to the deck, the selection of a card affects a subsequent drawing: the probability of selecting a black card (or selecting a non-black card) is changed. In this example, if a non-black card is initially drawn from a deck of 52 and not returned, then the probability of selecting a black card on the next, draw increases to .5098, or 26 out of 51. The probability of selecting a non-black card becomes .4902, or 25 out of 51.
. As discussed in the text accompanying notes 6-7, infra, I disagree with the majority’s assumption that the proportion of blacks in Madison Parish is the same as the proportion of eligible blacks qualified to be selected as a grand jury foreman.
. In Louisiana the selection of grand jury foremen resembles, in part, drawing cards from a deck without returning the cards prior to the next draw. See note 3, supra. The Code of Criminal Procedure requires that the parish jury commission delete from the general venire, the venire from which the grand jury venire is drawn and the grand jury foreman is selected, the names of those persons who have served since the previous selection of the general venire. La.Code Crim.Pro.Ann. arts. 410, 411 (West Supp.1981); see La.Sup.Ct.R. 25, § 4 (West 1981). Thus, like the playing card previously drawn and not yet returned to the deck, it is possible that a few persons in the population may be ineligible for selection as the next grand jury foréman. This aberration is, however, statistically insignificant. M. Finkelstein, Quantitative Methods in Law 32 (1978).
. The use of a sub-population whose members each possess traits enabling them to be selected is also employed in testing for racial discrimination in employment. Wilkins v. University of Houston, 654 F.2d 388 , 396 n.9 (5th Cir. 1981) (quoting Hazelwood School Dist. v. United States, 433 U.S. 299 , 308 n.13, 97 S.Ct. 2736 , 2741 n.13, 53 L.Ed.2d 768 (1977)) (“statistical challenges to a defendant’s hiring practices for positions requiring special qualifications must be based not on comparisons with the general population but on comparisons with ‘the smaller group of individuals who possess the necessary qualifications’ ”); see also Dorsaneo, Statistical Evidence in Employment Discrimination Litigation: Selection of the Available Population, Problems, and Proposals, 29 Sw. L.J. 859 , 866-71 (1975).
. We know absolutely nothing about the qualifications of the people on these several venires or on the venire from which the grand jury came that indicted the petitioners. In view of the peculiar responsibilities and demands placed on a foreman of a grand jury, it is nothing but wild speculation to assume that the absence of a black foreman was due to the systematic exclusion of blacks as grand jury foremen.
. The cumulative binomial probabilities computed to the nearest .0001 by the MINITAB II statistical package, see T. Ryan, B. Joiner & B. Ryan, MINITAB Student Handbook 91-101 (1976), for 31 random selections (n = 31) from a population that is 60 percent black (p = .6) are as follows:
Number of blacks chosen in 31 random selections (“k”): Probability of choosing k or fewer number of blacks:
1 0.0000
2 0.0000
3 0.0000
4 0.0000
5 0.0000
6 0.0000
7 0.0000
8 „ 0.0001
9 0.0005
10 0.0017
Number of blacks chosen in 31 random selections (“k"): Probability of choosing k or fewer number of blacks:
11 0.0050
12 0.0135
13 0.0320
14 0.0677
15 0.1284
16 0.2194
17 0.3399
18 0.4805
19 0.6248
20 0.7546
21 0.8566
22 0.9262
23 0.9670
24 0.9874
25 0.9960
26 0.9990
27 0.9998
28 1.0000
The prima facie case at the .05 level is attained when the probability decreases from .0677 to .0320 as the number of blacks selected as grand jury foremen drop from 14 to 13.