United States Court of Appeals
                        FOR THE EIGHTH CIRCUIT
                               _____________

                                No. 97-1143
                               _____________

Ian Maitland,                            *
                                         *
             Appellant,                  *
                                         *
       v.                                *
                                         *
University of Minnesota; Regents of the *
University of Minnesota; Wendell R.      *
Anderson; M. Elizabeth Craig; Jean B. * Appeal from the United States
Keffeler; Elton A. Kuderer; H. Bryan     * District Court for the
Neel, sued as H. Bryan Neel III; Mary J. * District of Minnesota.
Page; Lawrence Perlman; Thomas R.        *
Reagan; David K. Roe; Darrin M.          *
Rosha; Stanley D. Sahlstrom; Ann J.      *
Wynia; Julie Bleyhl, sued as Julie A.    *
Bleyhl; William E. Hogan, III; Hyon T. *
Kim; William Peterson, sued as           *
William R. Peterson; Nils Hasselmo,      *
University of Minnesota President,       *
                                         *
             Appellees.                  *
                                   _____________

                             Submitted: June 10, 1998
                                 Filed: September 16, 1998
                              _____________

Before BOWMAN, Chief Judge, BEAM and LOKEN, Circuit Judges.
                            _____________

BOWMAN, Chief Judge.
       Ian Maitland, a member of the faculty at the University of Minnesota, appeals
pro se from the order of the District Court granting summary judgment to the
University, its president, and present and former members of its Board of Regents
(collectively, the University) on Maitland's claim under Title VII, 42 U.S.C. §§ 2000e
to 2000e-17 (1994 & Supp. II 1996), that he was subjected to employment
discrimination based on his gender, and on his claim under 42 U.S.C. § 1983 (Supp.
II 1996) that the University violated his equal protection rights. We reverse in part,
vacate in part, and remand.

                                           I.

       This case has its genesis in what is known as the Rajender consent decree. The
settlement therein was entered into in 1980 by the University and a class of women
academic employees at the University to settle a gender discrimination class action
lawsuit that began in the 1970s. During the 1980s, female faculty members filed a
number of claims of salary discrimination pursuant to the procedure set forth in the
Rajender decree. In 1989, the University entered into a second consent decree in order
to settle those claims. Although Maitland was not a party to the decree, he was
permitted to express his objections to it. This decree provided for the distribution of
three million dollars in salary increases, in three phases, to class members--all of whom
were women. Any academic employee, male or female, was permitted to file a claim
to seek a salary increase under the "manifest inequity" provision of the settlement.
Maitland did not avail himself of this opportunity.

       Instead, Maitland filed an administrative claim, and then brought suit in federal
court alleging that the University was discriminating against him on the basis of his
gender. The district court granted summary judgment to the University, holding that
Maitland's action was barred by § 108 of the Civil Rights Act of 1991, 42
U.S.C. § 2000e-2(n)(1) (1994) (providing that an employment practice implemented
pursuant to the terms of a consent decree cannot be challenged by a Title VII lawsuit

                                          -2-
if certain procedures are in place to protect the rights of those who are not parties to the
decree). The court further held that Maitland was estopped from pursuing his claims.
We reversed. See Maitland v. University of Minn., 

(8th Cir. 1994).

       On remand, Maitland amended his complaint to seek damages under § 102(a)
of the Civil Rights Act of 1991, 42 U.S.C. § 1981a (1994), and to advance an equal
protection claim under 42 U.S.C. § 1983.1 The District Court again granted summary
judgment for the University, this time on the merits of Maitland's claims. The court
held that there remained no genuine issue of fact to be decided at trial and that the
University was entitled to judgment as a matter of law. The court further held that, in
any case, Maitland would be ineligible to recover the damages he sought under the
Civil Rights Act of 1991. Finally, the court held that "Defendants" (presumably the
individual defendants acting in their personal capacities) were entitled to qualified
immunity. Maitland v. University of Minn., Civil No. 4-93-25, Mem. Op. and Order
at 13 (D. Minn. Dec. 6, 1996). Maitland appeals.2

                                            II.

       We review the District Court's decision to grant summary judgment de novo.
See Hindman v. Transkrit Corp., 

, 990 (8th Cir. 1998). We view the
record in the light most favorable to Maitland, giving him the benefit of all reasonable
inferences that may be drawn from the evidence. We will affirm the District Court only




       1
       A § 1983 claim was in Maitland's original lawsuit but had been dismissed
without prejudice.
       2
        The court also held that "the Eleventh Amendment prohibits lawsuits requesting
monetary damages from the University of Minnesota and individual state officers in
their official capacities." Maitland v. University of Minn., Civil No. 4-93-25, Mem.
Op. and Order at 14 (D. Minn. Dec. 6, 1996). Maitland does not appeal this holding.

                                            -3-
if the record "show[s] that there is no genuine issue as to any material fact and that the
moving party is entitled to a judgment as a matter of law." Fed. R. Civ. P. 56(c).

        During the course of the second stage of the Rajender litigation, that which led
to the salary settlement, the parties prepared a database that included salary and other
personnel information from the files of the University's academic employees, both male
and female. Each side then prepared its own statistical pay study using multiple
regression analysis, but the results differed because the parties disagreed on the
relevance of certain variables such as rank, retention, length of service, and market
forces. The Striebel study, prepared for the plaintiffs, demonstrated a disparity in the
salaries of male and female faculty members of between 4.1% and 10.3%. This model
excluded factors such as rank and market (that is, how the University's faculty salaries
compared with those of people in similar positions at other institutions). The
statistician preparing the study felt that these factors themselves reflected employment
discrimination on the basis of gender and therefore would distort the results if included.
The study commissioned by the University, the Goodman study, showed a statistically
insignificant difference in pay (2%), but included the suspect factors. Neither study
included a performance variable per se.

       The parties, together with the special masters conducting the settlement
negotiations, agreed to use a compromise model, which included some of the contested
variables, in order to resolve the litigation. That model reflected a 6% differential
between the salaries of male and female academic employees and provided the basis
for the affirmative action salary plan set forth in the consent decree, although the
University never acknowledged that it had discriminated against its female faculty
members in setting salaries. Maitland contends that the University discriminated
against him when it implemented the plan because there was no salary disparity that
required remediation.




                                           -4-
       An affirmative action hiring plan may be successfully challenged as a violation
of Title VII unless the consideration of an otherwise improper factor (the gender of the
prospective employee) "was justified by the existence of a 'manifest imbalance' that
reflected underrepresentation of women in 'traditionally segregated job categories.'"
Johnson v. Transportation Agency, Santa Clara County, Cal., 

, 631 (1987)
(quoting United Steelworkers, AFL-CIO-CLC v. Weber, 

, 197 (1979)).
Further, to pass muster under Title VII, such an employment plan must not
"unnecessarily trammel[] the rights of male employees" and it must be "intended to
attain" a balance, "not to maintain one." 

639. We think this analysis applies
equally to an affirmative action salary plan, that is, a plan that favors one gender over
another in awarding salary increases in order to correct an allegedly historical
imbalance.

        Similarly, in equal protection jurisprudence, the test for the constitutionality of
an affirmative action plan is whether "a purportedly remedial affirmative action plan is
bona fide," that is, whether it "is indeed remedial." Donaghy v. City of Omaha, 

, 1458 (8th Cir. 1991) (hiring case), cert. denied, 

(1992). The
plan's defender is required to demonstrate a "conspicuous [gender] imbalance" if the
plan is to be held constitutional. 

plan must be narrowly tailored to
achieve the remedial goal. See 

see no reason why this test for the
constitutionality of an affirmative action hiring plan should not be the same for an
affirmative action salary plan.

       This case is a bit complicated because the plan in question was implemented
pursuant to a consent decree; it cannot be characterized as an entirely voluntary
affirmative action plan, but neither can it be said to have been "imposed after making
a formal finding of intentional discrimination." 

The question is what
difference, if any, it might make to our analysis that the plan had a court's imprimatur.
Our Court has held, in an equal protection case, that the employer's required showing
in defense of its plan remains the same notwithstanding the court's sanction, and that

                                           -5-
"the entry of an affirmative action consent decree does not guarantee that the decree
serves a remedial purpose or is narrowly tailored." 

heightened
judicial oversight inherent in a properly entered decree" may help the employer meet
his burden. 

was an equal protection hiring case, we think
similar consideration is due an affirmative action salary plan implemented pursuant to
a consent decree and challenged under Title VII.

        We now zero in on the specific--and undeniably material--question raised by this
appeal: whether there is undisputed evidence that shows, at this stage of the litigation,
that in 1989 when the salary plan was implemented there existed a "manifest" (Title
VII) or "conspicuous" (equal protection) imbalance in the salaries of male and female
faculty members at the University. It is clear that the relevant evidence is far from
undisputed. In the record are the widely varying conclusions of three different
statistical models and, while a salary imbalance of 6% or 10% might well be found to
be manifest and conspicuous, a statistically insignificant imbalance of 2% likely would
not.

        The divergent outcomes of the studies are the result of the different variables
that were included in the models. Maitland claims certain variables should have been
included in the Striebel study, and that the failure to do so skewed the results. It is true,
as the University and the District Court point out, that a regression analysis does not
become inadmissible as evidence simply because it does not include every variable that
is quantifiable and may be relevant to the question presented. See Bazemore v. Friday,

, 400 (1986) (per curiam) (Brennan, J., concurring).3 But we are


       3
       In Bazemore v. Friday, 

, 386, 387 (1986) (per curiam), the Court
held that "for the reasons stated in the concurring opinion of JUSTICE BRENNAN," the
court below erred "in holding that petitioners' regressions were unacceptable as
evidence of discrimination." Therefore, Justice Brennan's concurrence states the
position of the entire court.

                                            -6-
reviewing a decision to grant summary judgment, and the admissibility of the evidence
is not the question here; whether that evidence might be believed by a jury is, and
"failure to include variables will affect the analysis' probativeness." 

That is, if a regression analysis omits variables, it is for the finder of fact to
consider the variables that have been left out of an analysis, and the reasons given for
the omissions, and then to determine the weight to accord the study's results--in this
case, whether those results show a manifest or conspicuous imbalance in salaries.

       There were three comparative salary studies performed in connection with the
Rajender salary litigation, and three different results were reached because different
variables were used. Thus it was not "mere conjecture or assertion" by Maitland that
including certain variables that he believes to be relevant would give a different result
concerning the existence vel non of a manifest or conspicuous imbalance between male
and female academics. Palmer v. Shultz, 

, 101 (D.C. Cir. 1987), quoted
in Catlett v. Missouri Highway & Transp. Comm'n, 

, 1266 (8th Cir.
1987), cert. denied, 

(1988). By concluding that the Striebel study
"logically justified the non-use of variables," the District Court chose one study on
which to rely and thus was able to find the necessary salary imbalance to justify the
plan. Maitland, Civil No. 4-93-25, Mem. Op. and Order at 8. The court, however,
thereby resolved a genuine issue of material fact, and so summary judgment (for either
side) was inappropriate. Cf. Smith v. Virginia Commonwealth Univ., 

, 676
(4th Cir. 1996) ("[Defendant] cannot rely on Bazemore at summary judgment to
establish as a matter of law that the multiple regression analysis was sufficient to
determine manifest imbalance.").4

      4
       In Smith v. Virginia Commonwealth University, the court particularly faulted
the regression analysis at issue (there was only one) for its "failure to account for the
performance factors," noting that "Bazemore and common sense require that any
multiple regression analysis used to determine pay disparity must include all the major
factors on which pay is determined." 

, 676 (4th Cir. 1996). We do not
disagree, but note that none of the studies in this case, including the study on which

                                           -7-
        Moreover, we conclude that the fact that the salary plan was implemented
pursuant to a court-approved consent decree is of little consequence here. In approving
the consent decree, the district court made it clear that the settlement was the result of
a compromise between the parties, and said, "[T]he agreement increases base salaries
of female academic employees in response to statistical studies which provide a basis
to find a possible imbalance between male and female academic salaries." In re:
Rajender Salary Settlement, Civil File No. 3-89-464, Order Approving Settlement
Agreement and Consent Decree ¶ 14 (D. Minn. Oct. 12, 1989) (emphasis added). It
is apparent that the court found no "manifest" or "conspicuous" salary imbalance. In
fact, the court acknowledged that the settlement left unresolved the same disputed fact
questions at issue in this case:

      The two [statistical] reports raise significant questions about the proper
      variables, the appropriate groups to include in an analysis, the theory of
      statistical analysis, appropriate sample size and many other matters. It is
      uncertain whether plaintiffs' statistical analysis and other evidence would,
      in fact, establish an illegal disparity between men's and women's salaries.
      It is also uncertain whether the University's statistical analysis and other
      evidence would, in fact, establish that there is no illegal salary
      discrepancy between men and women at the University of Minnesota.

In these circumstances, where the court frankly acknowledges there is an
open question as to whether discrimination even exists, the fact that the affirmative
action salary plan was implemented pursuant to a consent decree does not bolster the



Maitland relies, included a performance variable. Accordingly, it is not clear that
Maitland presents, as he suggests, a genuine issue of material fact on the issue of
whether performance variables should have been included in the models. Cf. 

(reversing summary judgment in part because of dispute about performance variables:
defendant said that such variables were "not suitable for statistical analysis" and
plaintiffs' expert "stated that the study was not valid without adding the performance
factors").

                                           -8-
District Court's conclusion at the summary judgment stage of this case that there was
a manifest imbalance in faculty salaries.

       Maitland has demonstrated that there is a genuine issue of material fact on the
question of whether there was a manifest or conspicuous imbalance in faculty salaries
based on gender. We therefore do not consider the remaining steps of the analyses:
whether the salary plan unnecessarily trammeled the rights of male faculty and whether
it was designed to attain or maintain a balance (Title VII), or whether the plan was
narrowly tailored to meet the remedial objective (equal protection). The District Court's
grant of summary judgment in favor of the University on this issue is reversed.

                                          III.

        The District Court also granted summary judgment to the University on
Maitland's claim for damages under § 102(a) of the Civil Rights Act of 1991, 42
U.S.C. § 1981a(a). The court held that the allegedly discriminatory action by the
University was the implementation of the settlement according to its terms, which
predated the Act. Because § 102(a) is not retroactive, see Landgraf v. USI Film Prods.,

(1994), the court concluded that summary judgment for the University
was appropriate. Maitland argues that he has established a "continuing violation" so
that, if he should prevail on the merits of his Title VII discrimination claim, he would
be entitled to recover compensatory damages, whatever he may prove them to be,
suffered after November 1991, the effective date of the Act. We review de novo.

       All female academic employees at the University received a 3% across the board
salary increase in the first phase of the consent decree's affirmative action salary plan
and that, Maitland contends, was discrimination. According to Maitland, he has
suffered damages from that time forward. He describes his claim for damages this way:
"[E]ach University pay check since the implementation of the Settlement has paid less
to me than to similarly situated females. . . . The University has not taken any steps to

                                          -9-
undo the effects of the Settlement increases or to catch up my salary." Brief of
Appellant at 42. Thus, as far as we can discern from Maitland's argument, he is not
seeking compensatory damages as they are described in the 1991 Act. See 42
U.S.C. § 1981a(b)(3). Instead, he is hoping to recover backpay, equitable relief that
always has been available under Title VII to the successful plaintiff (at the discretion
of the court and with some limitations), and that is specifically not available under the
1991 Act. See 

("Compensatory damages awarded under this section
shall not include backpay, interest on backpay, or any other type of relief authorized
under section 706(g) of the Civil Rights Act of 1964 [42 U.S.C. 2000e-5(g)].") Because
Maitland filed a timely Title VII action challenging the 1989 settlement, he is entitled
to pursue his Title VII remedies, including backpay.

     We vacate the ruling of the District Court, as the mischaracterization of the
monetary remedy Maitland seeks renders that ruling of no force in this case.5

                                          IV.

       Finally, Maitland argues that the District Court erred in holding that the
defendants are entitled to qualified immunity.6 We review de novo, considering whether
the facts Maitland cites in support of his § 1983 claim could result in a conclusion that
the defendants violated clearly established law. See Murphy v. State of Ark., 

, 755 (8th Cir. 1997); Smith v. Arkansas Dep't of Correction, 

, 648 (8th
Cir. 1996).


      5
        We express no opinion as to whether Maitland's failure to seek an increase in
salary using the procedures provided for by the consent decree would affect his ability
to recover equitable relief in the form of backpay, should he succeed on the merits of
his Title VII claim.
      6
       We are assuming for the purposes of addressing this argument on appeal that
Maitland in fact has sued the individual defendants in their personal capacities.

                                         -10-
       Based on its "previous analysis discussing permissible gender based
classifications," the District Court held that "Defendants have not violated a clearly
established law." Maitland, Civil No. 4-93-25, Mem. Op. and Order at 13. Further, the
court concluded that the defendants were reasonable in believing "that entering into a
settlement agreement approved by a federal judge" would not "violate[] a clearly
established right of an employee." 

       Our holding in Part II of this opinion requires reversal of the qualified immunity
decision. Maitland is challenging the legality of the consent decree under § 1983, and
we are remanding his claim for trial because there is a disputed issue of material fact
relating to whether there was a conspicuous imbalance in salaries that would justify the
salary plan. The same issue of fact affects the claim of qualified immunity. See Smith
v. Arkansas Dep't of 

. There is no question that the law
was clearly established when the salary plan was implemented: employment
discrimination based on gender may be a violation of the Constitution. The University
entered into the consent decree, and agreed to the salary plan, with knowledge that there
was a multiple regression analysis that showed no statistically significant salary
discrimination--in fact, it was the very study commissioned by the University for use in
resolving the Rajender plaintiffs' claim. We cannot say that it was objectively
reasonable, under clearly established constitutional law, for the defendants to agree to
a salary plan--notwithstanding a court's approval--when they had strong reason to
believe that a likely result of the plan would be to create a salary imbalance in favor of
female academic employees.

      We reverse the District Court's decision to grant qualified immunity to the
defendants.




                                          -11-
                                           V.

       The judgment of the District Court is reversed in part and vacated in part, and the
case is remanded for further proceedings.

      A true copy.

             Attest:

                     CLERK, U.S. COURT OF APPEALS, EIGHTH CIRCUIT.




                                          -12-