JOHN F. LEHMAN, JR., SECRETARY OF THE NAVY, ET AL., PETITIONERS V.
YVONNE G. TROUT, ET AL.

   No. 83-706

   In the Supreme Court of the United States

   October Term, 1983

   The Solicitor General, on behalf of the Secretary of the Navy and
the Commanding Officer of the Navy Regional Data Automation Center,
petitions for a Writ of Certiorari to review the judgment of the
United States Court of Appeals for the District of Columbia Circuit in
this case.

   Petition for a Writ of Certiorari to the United States Court of
Appeals for the District of Columbia Circuit

                      PARTIES TO THE PROCEEDINGS

   The petitioners are John F. Lehman, Jr., Secretary of the Navy, and
Capt. James C. Richardson, Jr., Commanding Officer of the Navy
Regional Data Automation Center.  The named respondents representing
the class are Clara A. Perlingiero and Yvonne G. Trout.  In addition,
the individual claims of Marie Bach, Charlene Hardy, and Joan S.
Creighton were consolidated with the class action in the district
court, and Bach's appeal was consolidated with the class action in the
court of appeals.  Hardy and Creighton did not appeal from the court
of district court's denial of their claims;  Trout's individual appeal
was rejected and this Court denied her individual petition for
certiorari on April 18, 1983 (No. 82-1429).

            TABLE OF CONTENTS
   Opinions below
   Jurisdiction
   Statute involved
   Statement
   Reasons for granting the petition
   Conclusion
   Appendix A
   Appendix B
   Appendix C

                            OPINIONS BELOW

   The opinion of the court of appeals (App., infra, 1a-39a) is
reported at 702 F.2d 1094.  The opinion of the district court (App.,
infra, 40a-70a) is reported at 517 F. Supp. 873.

                             JURISDICTION

   The judgment of the court of appeals was entered on March 11, 1983.
 Rehearing was denied on May 31, 1983 (App., infra, 73a).  The time
for filing a petition for a writ of certiorari was extended to and
including October 28, 1983, by Justice Brennan on August 15, 1983.
The jurisdiction of this Court is invoked under 28 U.S.C. 1254(1).

                           STATUTE INVOLVED

   Section 717(a) of Title VII of the Civil Rights Act of 1964, 42
U.S.C. (Supp. V) 2000e-16(a), provides in pertinent part:

         All personnel actions affecting employees or applicants for
      employment * * * in military departments as defined in section
      102 of Title 5, * * * shall be made free from any discrimination
      based on race, color, religion, sex, or national original.

                          QUESTIONS PRESENTED

   1. Whether, in light of this Court's subsequent decision in United
States Postal Service Board of Governors v. Aikens, No. 81-1044 (Apr.
4, 1983), the court of appeals erred in affirming the judgment of the
district court despite concluding that the statistical analysis upon
which that court had relied in finding discrimination in violation of
Title VII of the Civil Rights Act of 1964, as amended (42 U.S.C. 2000e
et seq.), included data that could not form a basis for imposing
liability.

   Our primary submission is that, because of the errors in the
district court's analysis identified by the court of appeals, no
violation of Title VII has been established under the standard
prescribed in Aikens, and the judgment of the court of appeals should
accordingly be summarily vacated and the case remanded for
reconsideration in the light of Aikens.  If this Court nevertheless
decides to set this case for plenary review, two additional questions
are presented:

   2. Whether intentional discrimination is established in a class
action suit involving professional technical positions by a
statistical analysis that fails to account adequately for minimum
objective qualifications for those positions.

   3. Whether, in remedial hearings after a class-wide violation of
Title VII has been found, the burden shifts to the employer to
demonstrate "by clear and convincing evidence" that its employment
decisions with regard to particular individuals were based on
legitimate, non-discriminatory factors.

                               STATEMENT

   1. Four cases involving individual and class sex discrimination
claims under Title VII were brought against petitioners (the Secretary
of the Navy and the Commanding Officer of the Navy Regional Data
Automation Center) and consolidated for trial in the United States
District Court for the District of Columbia.  Respondents' class
allegations were that the Navy intentionally discriminated in
employment practices against "all female professional technical
employees employed by the (Naval Command Support Activity)
'NAVCOSSACT' or the (Navy Regional Data Automation Center) 'NARDAC" at
any time between June 6, 1972 and June 4, 1979." /1/ App., infra, 41a.
 The employees involved performed computer-related tasks.

   At trial, respondents relied on a computerized multiple regression
analysis to show that women received lower salaries than men, in an
attempt to show women were paid less than one would expect in the
absence of discrimination.  A multiple regression analysis is a
statistical technique designed to estimate the effects of several
independent variables on a given result.  Properly used, such an
analysis can show how much influence factors such as sex, experience,
and education had on perceived disparities in salary levels.  Here,
the independent variables included in respondents' study were the
following:  (1) level (but not type) of formal education, (2) years
between date of last education and entry into the federal service --
in effect, the age at which the employee was hired, (3) length of
service (but not type of service) with NAVCOSSACT and NARDAC, and (4)
length of service (but not type of service) with the federal
government.  App., infra, 6a.  After accounting for these four
variables, the analysis attributed the remaining salary differences to
the sex of the employees considered.  Based on this analysis, which
reflects the effects of pre-1972 actions on salaries within the
plaintiff class as well as the effects of actions taken by agencies
other than NAVCOSSACT and NARDAC, respondents' expert concluded that
there was statistically significant adverse treatment of women at
NAVCOSSACT and NARDAC.

   Petitioners introduced evidence to show that this conclusion
resulted from certain flaws in respondents' analysis, including the
fact that respondents' study relied on data regarding employment
decisions that could not form the basis of Title VII liability.  /2/
Accordingly, petitioners' evidence excluded -- and was the only
evidence to exclude -- employment decisions that the court of appeals
ultimately held were not actionable -- i.e., the effect of pre-1972
actions and initial placement decisions.  See page 6, infra,.  /3/ On
the basis of this evidence, petitioners contended that when such
irrelevant factors as pre-1972 actions and initial placement decisions
were eliminated and persons with comparable computer experience were
compared, women advanced in a statistically indistinguishable way from
men.

   2. Following a trial, the district court ruled that petitioners had
discriminated against the class of women professional technical
employees in initial grade placement and promotions, basing its
finding largely on the regression analysis presented by respondents.
/4/ Essential to the district court's ultimate conclusion was its
rejection of two arguments made by petitioners.  First, the court
ruled (App. infra, 46a-48a) that petitioners' objection to
respondents' failure to distinguish between pre- and post-1972 actions
was not valid because discrimination occurring before 1972 "even if
coupled with neutral employment practices since then, produced
actionable continuing discriminatory effects after 1972 * * *." /5/
Second, the court rejected (App., infra, 48a) petitioners' contention
that respondents' regression analyses were flawed because they placed
responsibility on petitioners for decisions regarding initial
placement of employees when hired by, or transferred to, NAVCOSSACT
and NARDAC.  /6/

   Applying similar reasoning, the district court also rejected
petitioners' statistical analyses, in large part because they did not
take into account pre-1972 discrimination, App., infra, 57a, and
because they "assumed that the employees studied were placed in an
appropriate salary grade at the outset, and (the analyses) therefore
could not and did not detect any bias in the hiring and placement
process." App., infra, 56a (footnote omitted).  Relying on its
conclusion that pre-1972 actions and initial placement decisions were
actionable, the district court concluded that respondents' multiple
regression analysis established a prima facia case, and that "applying
the burden-of-production rule articulated by the Supreme Court in
(Texas Department of Community Affairs v.) Burdine, (450 U.S. 248
(1981),) * * * the (petitioners) have not rebutted (respondents')
prima facie case of discrimination" (App., infra, 59a).

   3. A divided court of appeals (MacKinnon, J., dissenting) affirmed
in part and reversed in part the district court's ruling on the class
claims presented in this case.  It reversed the district court's
holding that petitioners, rather than other federal agencies
responsible for formulating federal personnel regulations, were
responsible for determining the grade levels of new employees.  /7/
The appellate court also ruled that the district court's theory for
permitting the inclusion of pre-1972 statistics in respondents'
analysis "is flatly inconsistent with the Supreme Court's
pronouncements in Hazlewood School District v. United States, 433 U.S.
at 309-10, * * * and United Air Lines, Inc. v. Evans, 431 U.S. 553,
558 (1977)" (App., infra, 15a-16a).

   Notwithstanding these rulings, the court of appeals affirmed the
district court's decision that respondents had established a prima
facie case that petitioners had violated Title VII with regard to
promotions (App., infra, 17a-18a) and that petitioners had not
rebutted that case (App., infra, 18a-19a).  /8/ The court acknowledged
(App., infra, 16a) that, because they group nonactionable employment
decisions with actionable ones, respondents' analyses "hold the
potential for some distortion";  it nonetheless concluded that
respondents had established a prima facie case because their analyses
"included a substantial amount of data covering post-1972 employment
decisions affecting class members." /9/

   The court of appeals also affirmed the district court's decision
regarding the burden of proof that would be placed on the employer in
remedial hearings held to determine liability to individual class
members.  Reiterating its ruling in McKenzie v. Sawyer, 684 F.2d 62,
76-78 (D.C. Cir. 1982), the court held (App., infra, 21a-22a):
"Because such class members are presumptively entitled to relief upon
showing that they were potential victims of the defendants'
discriminatory practices * * * the defendants must 'rebut the
plaintiffs' individual showing by clear and convincing evidence'.
McKenzie, 685 F.2d at 78."

   Judge MacKinnon dissented from the majority's imposition of
class-wide liability.  He argued that the district court had
incorrectly allocated the burden of proof by requiring defendants to
"disprove the plaintiffs' allegations" (App., infra, 29a;  emphasis in
original).  Moreover, he concluded (id. at 31a) that the district
court "was so concerned with the allocation of the 'stages' of proof,
focusing upon whether the plaintiffs established a prima facie case
and whether the defendant rebutted that case, that it lost sight of
the basic principle that plaintiffs were ultimately required to prove
intentional discrimination by a preponderance of the evidence."

   Petitioners filed a petition for rehearing, suggesting rehearing en
banc.  The court of appeals denied the petition, with Judges Tamm,
Wilkey, MacKinnon and Bork dissenting (App., infra, 75a).

                   REASONS FOR GRANTING THE PETITION

   The outcome of most Title VII litigation involving class-wide
claims of discrimination is determined by the court's evaluation of
statistical evidence and analyses.  It is, therefore, crucial that
such evidence be carefully analyzed, bearing in mind that the ultimate
responsibility of the court is to apply the substantive principles of
law underlying Title VII and, as this Court emphasized in United
States Postal Service Board of Governors v. Aiken, supra, to determine
whether, at the conclusion of the trial, plaintiffs have established
that they have been discriminated against in violation of Title VII.
The uncritical acceptance of statistical evidence, without regard to
settled principles of law carefully developed by Congress and this
Court's prior decisions, will inevitably undermine the objectives of
Title VII.  The courts below imposed liability relying on a
statistical analysis whose conclusion was based, in critical part, on
evidence that did not relate to employment decisions actionable under
Title VII.  The data offered in this case showed only that women in
respondent's class generally occupy lower pay levels than men in a
unit of the Navy that has recently increased substantially the number
of its women employees, with resultant added concentration of women at
trainee and entry level positions.

   The district court and the court of appeals rendered their decision
before Aikens was decided, and committed the same fundamental error as
the lower courts in Aikens.  /10/ Had the court of appeals focused
"directly on the question of discrimination," Aikens, slip. op. 5, and
decided whether plaintiffs, at the conclusion of the evidence, had
established the existence of discriminatory employment practices, it
would necessarily have concluded that there was no basis for affirming
the district court's imposition of liability.  The foundation of the
district court's evaluation of the evidence was simply swept away by
the court of appeals' ruling that the effects of pre-1972 actions and
initial placement decisions could not form the basis for imposing
liability.  The district court relied on statistical evidence that
factored in employment decisions that the court of appeals
specifically found could not form the basis for imposing liability
under Title VII, and rejected petitioners' statistical evidence
precisely because it excluded such decisions.  Once the court of
appeals corrected those errors in the district court's legal analysis,
there remained no reliable evidence upon which the appellate court
could properly conclude that respondents had established, with respect
to the class claims, any violation of Title VII by petitioners.
Accordingly, the court of appeals should have either reversed the
district court's decision or remanded the case to the district court
for an evaluation of all the evidence under correct legal standards.
Pullman-Standard v. Swint, 456 U.S. 273, 292 (1982).

   The district court and the court of appeals agreed that the
determination of whether defendants had engaged in class-wide
discrimination "revolves primarily around the statistics submitted by
the parties and the analysis of the statistics by the parties'
experts" (App., infra, 42a, 6a).  Statistical evidence "must be
regarded with a substantial degree of caution," Pegues v. Mississippi
State Employment Service, 699 F.2d 760, 766 (5th Cir. 1983), petition
for cert. pending, No. 83-139, because the usefulness of statistics
"depends on all the surrounding facts and circumstances."
International Brotherhood of Teamsters v. United States, 431 U.S. 324,
340 (1977).  No valid conclusions regarding petitioners' employment
practices can be drawn unless the statistical evidence is analyzed in
accordance with legally correct interpretations of substantive Title
VII Law.  See Pegues, 669 F.2d at 767-768;  Pouncy v. Prudential
Insurance Co. of America, 668 F.2d 795, 803 (5th Cir. 1982);  Wilkins
v. University of Houston, 654 F.2d 388, 395 (5th Cir. 1981).  In this
case, the court of appeals premised liability on statistics which
admittedly were not meaningfully tied to correct legal principles.
When analyzed in light of those principles and of the government's
evidence revealing their flaws, those statistics simply do not show
that NAVCOSSACT and NARDAC discriminated against women in promotions
during the relevant time frame.  Yet the court of appeals, repeating
the same error that court had made in Aikens, based its decision on
whether respondents' evidence was adequate to establish a prima facie
case in the absence of evidence by the government proving the flaws in
respondents' statistical presentation (App., infra, 10a-21a;  Cf. id.
at 31a-33a (dissenting opinion).

   It is undisputed that the women in respondents' class in general
occupy lower pay grades than men.  It is, however, both erroneous and
contrary to the purposes of Title VII for a court to leap from that
fact alone to the conclusion that discrimination has caused the
disparity.  Recent recruitment efforts have brought more members of
previously under-represented groups into the federal workforce in
technical fields such as the computer operations involved here.  /11/
Substantial numbers of women and minorities are accordingly
necessarily clustered in entry level positions, and are paid
accordingly.  Until these inexperienced workers gain the requisite
experience and training, and until vacancies occur at higher level
positions, it cannot legitimately be expected that minorities and
women will be equally distributed throughout an agency's workforce.
See Valentino v. United States Postal Service, 674 F.2d 56, 72 n.29
(D.C. Cir. 1982).  Unwarranted speculation, such as that of the court
below, that the disparity between the numbers of men and women in
higher level positions is the result of discrimination penalizes
agencies for recruiting women and minorities at entry-level positions,
thereby discouraging such recruitment efforts and, in the long run,
tending to decrease the opportunities of these groups for advancement
in new fields.

   Moreover, the court of appeals' substantive rulings entirely
eliminated the legal basis for the district court's conclusion that
respondents' regression analysis established facts necessary to impose
liability on petitioners.  The court of appeals correctly ruled that
two critical components of respondents' analysis were not legally
relevant.  First, it held that the district court had erred in ruling
that petitioners' pre-1972 actions "even if coupled with neutral
employment practices since then, produced actionable continuing
discriminatory effects after 1972" (App., infra, 47a).  As this Court
said in United Air Lines v. Evans, 431 U.S. 553, 558 (1977) (emphasis
in original):  "the emphasis should not be placed on mere continuity;
the critical question is whether any present violation exists."
Accordingly, here as in Evans, where the present system is neutral,
pre-1972 acts of discrimination, alone, do not convert it into a
violation of Title VII.  Second, the court of appeals properly
concluded that petitioners had established at trial "that initial
grade placements were made by either the CSC (Civil Service
Commission) or the OPM (Office of Personnel Management) or by
employing agencies that were not made defendants in this case" (App.,
infra, 17a).  /12/ There was thus no basis for holding defendants here
responsible for initial placement decisions.

   The inclusion of data regarding pre-1972 actions and initial
placement decisions necessarily rendered respondents' regression
analysis, and the district court's conclusions based on it,
unreliable.  See EEOC v. Federal Reserve Bank, 698 F.2d 633, 645 (4th
Cir. 1983), petition for cert. pending sub. nom. Cooper v. Federal
Reserve Bank, No. 83-185.  Liability cannot be premised upon the
results of a multiple regression analysis that inextricably
intertwines the effects of actionable employment decisions with the
effects of nonactionable employment decisions.  It is not possible to
isolate the portion of such results, if any, attributable to decisions
on which liability may properly be grounded.  The statistical results
upon which the trial court relied reflected without differentiation
the combined effects of three factors:  (1) timely promotion decisions
upon which it would be legally permissible to premise liability;  (2)
untimely promotion decisions upon which it would not be legally
permissible to premise liability;  and (3) timely and untimely initial
placement and decisions of others upon which it would not be legally
permissible to premise petitioners' liability.  Based on the evidence
relied upon by the trial court, it is not legally or logically
possible to determine which, if any, portion of the combined results
of these three factors is due to personnel actions upon which
liability can be based.  Indeed, the district court specifically
stated that there was "(n)o practical method of separating the effects
of pre-1972 and post-1972 discrimination in regression analysis"
(App., infra, 48a n.20);  but see note 14, infra.  /13/ It was,
therefore, improper for the court of appeals to hold without further
refinement or analysis that the relevant evidence established an
inference of discrimination sufficient to carry respondents' burden of
proof.  Wilkins v. University of Houston, 662 F.2d 1156, 1157 (5th
Cir. 1981).

   Under Pullman-Standard v. Swint, 456 U.S. 273, 292 (1982), a remand
is required where the district court findings are rendered unreliable
by an erroneous view of the law, such as the district court exhibited
here, unless "the record permits only one resolution of the factual
issue." See also United States Postal Service Board of Governors v.
Aikens, supra.  The evidence in this record when evaluated under the
correct legal standards regarding the effects of pre-1972 actions and
initial placement decisions does not permit the conclusion that the
trier of the fact would be required to resolve the issues in
respondents' favor.  Thus, the state of the record required the court
of appeals to remand the case to the district court for further
proceedings under proper legal standards.  In Aikens, this Court
ordered remand because "we cannot be certain that (the district
court's) findings of fact in favor of the Postal Service were not
influenced by its mistaken view of the law" (slip op. 5).  In the
instant case, the district court's acceptance of respondents'
statistics and its rejection of petitioners' statistics are expressly
based upon the very legal errors recognized by the court of appeals
(pages 3-6, supra).  Accordingly, the principle of Aikens, a
fortiorari, requires a remand to the district court here.  /14/

   The courts below rendered their decisions prior to Aikens.  The
court of appeals' decision cannot stand in light of this Court's
direction in Aikens that all of the evidence must be evaluated at the
conclusion of the case in order to determine whether plaintiffs
carried their burden of establishing by a preponderance of the
evidence that defendants intentionally discriminated against them.  We
therefore submit that the court of appeals' decision should be
summarily vacated and the case remanded with appropriate directions
for further proceedings.  /15/

                              CONCLUSION

   The judgment of the court of appeals should be vacated and the case
remanded for further consideration in light of United States Postal
Service Board of Governors v. Aikens, supra.

   REX E. LEE

      Solicitor General

   J. PAUL MCGRATH

      Assistant Attorney General

   CAROLYN B. KUHL

   ROBERT E. KOPP

      Attorneys

   OCTOBER 1983

   /1/ NAVCOSSACT was responsible for the Chief of Navy Operations'
computer activities in the 1960's and 1970's.  Among its duties were
the development and operation of highly classified computer programs
and systems relating to national defense.  These activities included
wargaming, telecommunications, and strategic command and control.  The
name of NAVCOSSACT was changed to NARDAC on March 17, 1977, when the
Navy effected a major reorganization of its data processing
activities.  In order to consolidate these activities, data-processing
personnel and support resources of the Navy Materiel Command Support
Activity, the Navy Accounting and Finance Center and the Naval
District, Washington, D.C., were transferred to NARDAC.

   /2/ To eliminate the effects of pre-1972 actions and actions taken
by employees other than NAVCOSSACT/NARDAC, petitioners ran the same
regressions used by respondents, but only for those employees hired by
petitioners after 1972.  In no year for which those regressions were
run (1974-1979) was sex a statistically significant variable.  Def.
Exh. 204, at 4-6, Def. Exh. 205;  Tr. 1427-1432, 1436-1437, 1500-1506.

   /3/ This evidence included an extensive "cohort analysis" that
grouped men and women who started at NAVCOSSACT and NARDAC at the same
GS level at similar times and compared their advancements.  In
addition, applicant flow data presented by petitioners showed that
women applying for promotions from September 1976 to 1979 had
statistically the same treatment with regard to promotions as did
their male counterparts.

   In addition, petitioners filed a motion for reconsideration after
trial based upon new statistical analyses.  These new analyses were
intended to further clarify two points:  (1) an evaluation of
NAVCOSSACT and NARDAC personnel files showed that women who were
initially placed at NAVCOSSACT and NARDAC had 49% of the computer
experience of men, and the generally lower initial placement level of
women simply reflected their relative lack of computer experience;
and (2) any showing of adverse treatment of women was eliminated when
initial placement and time-barred acts were removed from the
regression analyses.  The district court rejected this evidence as
untimely.

   /4/ The court also considered five individual claims, ruling in
favor of respondents on two and against them on the remaining three
claims.  The court of appeals affirmed the district court's finding of
discrimination against the two individuals, but modified the relief
awarded.  We do not challenge the rulings regarding the individual
claims in this petition.

   /5/ The court stated that it knew of no practical method of
separating the effects of pre-1972 and post-1972 discrimination in a
regression analysis (App., infra, 48a n.20).

   /6/ The district court also rejected petitioners' contention that
respondents' regression analysis failed to account adequately for the
minimum necessary objective qualifications of the positions at issue.
That is, petitioners contended that the fundamental cause for the
mistaken notion that sex discrimination caused the salary disparity
between men and women in NAVCOSSACT and NARDAC was the failure of
respondents' analyses to account for differences in computer-related
experience.  The district court rejected this argument on the ground
that respondents' variables were sufficiently accurate to establish a
prima facie case.  App., infra, 49a-52a.

   /7/ In accordance with that determination, the court of appeals
concluded that "the District Court erred in holding that the class was
entitled to relief for discriminatory initial placements" (App.,
infra, 17a;  emphasis in original).

   /8/ The court of appeals' opinion focuses almost exclusively on the
statistical evidence presented.  Without discussion, it notes that
"the District Court properly considered the statistical and
nonstatistical evidence on a cumulative basis in assessing the
strength of the (respondents') prima facie case." App., infra, 6a n.7.

   /9/ Regarding petitioners' objection that respondents' analyses
ignored the minimum objective qualifications necessary for placement
and promotion in the professional technical job series at issue, the
court ruled (App., infra, 12a-14a) that respondents' variables were
closely enough related to years of computer experience to provide an
adequate basis for inferring disparate treatment.

   /10/ In our petition for rehearing, we called the court's attention
to Aikens and its application to this case.

   /11/ Evidence at trial (Def. Exh. 198, at 1-18;  Tr. 1409-1425)
showed that the selection of women for trainee positions at
NAVCOSSACT/ NARDAC generally compared favorably with their
availability at the appropriate grade level both nationally and
locally:

      Grade
        5
        7
        9
      Total New Hires (1/1/72-3/12/72)
        26
        64
        16
      % Female Representation
        38.46
        42.19
        18.75
      Nationwide % Female Availability
        20.88
        18.91
        17.33
      Wash., D.C. % Female Availability
        27.09
        24.28
        21.97

   /12/ The grade (i.e., salary) levels of persons entering the
federal service are determined in accordance with OPM (formerly CSC)
regulations that take account of relevant education and experience.
Similarly, OPM regulations limit the extent to which a persons' grade
level may be increased on a transfer from one agency to another.  See
generally 5 C.F.R. Pts. 300-351.

   /13/ The district court made no finding that promotion decisions
alone were discriminatory, because it believed, contrary to the court
of appeals' ruling, that petitioners had control over initial
placement decisions.  Indeed, the district court specifically rejected
defendants' extensive statistical evidence, which focused on
promotions and showed no adverse treatment of class members, because
that evidence did not deal with initial placement.  (App., infra, 55a
n.40, 59a n.47.)

   /14/ The respondents' multiple regression analysis could then be
rerun using a model that factors out the effects of irrelevant
employment decisions (i.e., decisions occurring prior to March 24,
1972, and decisions relating to initial placement at time of hire).
This can be easily accomplished with the simple addition of a variable
to the respondents' model that accounts for grade or salary as of
March 24, 1972, and for grade or salary at initial placement for those
placed after March 24, 1972.  See Finklestein, The Judicial Reception
of Multiple Regression Studies in Race and Sex Discrimination Cases,
80 Colum. L. Rev. 737, 745-746 (1980).

   While we would not oppose affording respondents a further
opportunity for evidentiary development in light of the court of
appeals' clarification of the governing legal standards, we submit
that, in fairness, the government should also be given the opportunity
to submit additional evidence, such as that submitted with our motion
for reconsideration.  See note 3, supra.  We submit that such evidence
will establish that women at NAVCOSSACT and NARDAC have received as
many promotions and advanced as similarly situated men at all relevant
times.

   /15/ In light of this submission, we do not urge that the Court
grant plenary review in order to consider the two additional questions
we have presented.  We do not, on balance, believe that these
questions present issues requiring plenary consideration at this time.
 Should the Court set the case for briefing and argument, however, we
believe that these additional questions should also be considered, in
order to present the case in its full factual and legal context.

   The Court recognized in Hazelwood School District v. United States,
433 U.S. 299, 308 (1977), that when special qualifications are
required to fill the jobs at issue, the relevant statistical
comparisons must be among those who possess these qualifications.
Here, respondents' regression analysis failed to account for such
qualifications, since it did not factor in computer-related experience
as an independent variable.  The court below concluded (App., infra,
14a) that the independent variable of length of employment with
NAVCOSSACT/NARDAC was a valid proxy for such experience.  That
conclusion overlooks the fact that the grade level at which a class
member is initially placed when joing NAVCOSSACT/NARDAC reflects her
computer-related experience before coming to the agency.  All
respondents' independent variables relevant to that initial placement
(education, length of government service and, essentially, age) ignore
that experience.  The class members had significantly less
computer-related experience;  that relative lack of experience was not
adequately reflected in respondents' regression analysis.

   The court of appeals followed its prior practice in imposing a
heightened burden of proof on the employer at the remedial stage of
Title VII litigation.  It held that after a class member shows she has
been adversely affected, the employer must rebut the presumption of
discrimination by "clear and convincing evidence" establishing a
legitimate business reason for the adverse action.  App., infra,
21a-22a;  McKenzie v. Sawyer, 684 F.2d 62, 78 (D.C. Cir. 1982).  This
ruling is consistent with the practice in the Seventh and Fifth
Circuits (Stewart v. General Motors Corp., 542 F.2d 445, 453 (7th Cir.
1976), cert. denied, 433 U.S. 919 (1977);  Lee v. Washington County
Board of Education, 625 F.2d 1235, 1239 (5th Cir. 1980)), but it
conflicts with the practice in the Fourth Circuit, where the employer
may justify his actions by a simple preponderence of the evidence.
Sledge v. J. P. Stevens, 585 F.2d 625, 637 (1978), cert. denied, 440
U.S. 981 (1979);  EEOC v. American National Bank, 652 F.2d 1176, 1201
(4th Cir. 1981).  Petitioners are defendants placed in the unusual
posture of having the burden of proof shifted to them at the remedial
stage.  The shift of the burden is, we submit, clearly adequate to
protect the interests served by Title VII under the ordinary
"preponderance of the evidence" standard.  See International
Brotherhood of Teamsters v. United States, 431 U.S. 324, 362 (1977);
California v. Mitchell Brothers' Santa Ana Theater, 454 U.S. 90, 93
(1981);  Herman & MacLean v. Huddleston, Nos. 81-680 & 81-1076 (Jan.
24, 1983).

   Appendix Omitted