Equal Employment Opportunity Commission v. Kaplan Higher Education Corporation
Filing
110
Memorandum Opinion and Order: Defendants' Motion for Summary Judgment (Doc. 79 ) and the Motion to Exclude the Reports and Testimony of Dr. Kevin R. Murphy (Doc. 82 ) are GRANTED. The EEOC's Motion for Partial Summary Judgment (Doc. 80 ) and Motion to Exclude Defendant's Expert Michael G. Aamodt (Doc. 90 ) are DENIED as MOOT. Judge Patricia A. Gaughan on 1/28/13. (LC,S)
UNITED STATES DISTRICT COURT
NORTHERN DISTRICT OF OHIO
EASTERN DIVISION
Equal Opportunity Employment
Commission,
Plaintiff,
Vs.
Kaplan Higher Learning Edu. Corp.,
et al.,
Defendants.
)
)
)
)
)
)
)
)
)
)
)
CASE NO. 1:10 CV 2882
JUDGE PATRICIA A. GAUGHAN
Memorandum of Opinion and Order
INTRODUCTION
This matter is before the Court upon Defendants’ Motion for Summary Judgment (Doc.
79) and the EEOC’s Motion for Partial Summary Judgment (Doc. 80). Also pending are the
Defendants’ Motion to Exclude the Reports and Testimony of Dr. Kevin R. Murphy (Doc. 82)
and the EEOC’s Motion to Exclude Defendant’s Expert Michael G. Aamodt (Doc. 90). This is
an employment case alleging disparate impact. For the following reasons, the motion to exclude
plaintiff’s expert is GRANTED. Defendants’ Motion for Summary Judgment is also
1
GRANTED. The EEOC’s Motion to Exclude Defendant’s Expert Michael G. Aamondt and its
Motion for Partial Summary Judgment are DENIED as MOOT.
FACTS
Plaintiff, the Equal Employment Opportunity Commission (“EEOC”), brings this lawsuit
against defendants, Kaplan Higher Education Corporation (“KHEC”), Kaplan, Inc. (“KI”), and
Iowa College Acquisition Corporation d/b/a/ Kaplan University (“KU”), alleging that
defendants’ use of credit reports in the hiring process has an unlawful disparate impact on Black
applicants.1
The facts of this case are undisputed.
Defendants are groups of educational institutions and operate in a highly regulated
industry. The Department of Education (“DOE”) provides financial aid to many students
enrolled at KU and KHEC. As such, employees working in defendants’ financial aid
departments must utilize the National Student Loan Data System (“NSLDS”), which is DOE’s
student aid database. DOE requires its participants to have quality controls in place that limit
access to student and parent information.
Prior to 2004, defendants discovered breaches of their systems in which business officers
misappropriated student payments. As a result, defendants took additional steps to ensure
compliance with financial rules and regulations, as well as DOE guidelines. Shortly thereafter,
defendants began utilizing credit history in order to ascertain whether applicants for certain
emloyment positions were under “financial stress or burdens” that might compromise their
1
The parties refer to the disparately impacted group as “Black
applicants.” For the sake of consistency, the Court will employ the
parties’ terminology.
2
ethical obligations. The affected positions included:
•
positions involving substantial operating control;
•
positions involving access to financial aid or student accounts; and
•
positions involving responsibility for company financials or significant cash
handling.
Defendant KU’s use of credit information
In 2008, KU hired General Information Services (“GIS”) to perform credit checks on
candidates for hire who had been given conditional offers of employment. GIS reviewed
applicants’ credit reports for the following information:
•
Instances in which the social security number does not match the social security
number the credit bureau has on file;
•
No credit and 25 years or older;
•
Any bankruptcy;
•
Overdue child support payments totaling $2,000 or more;
•
Current garnishments on earnings;
•
Outstanding civil judgments totaling $2,000 or more;
•
Any individual or combined past due balances totaling $2,000 or more that are at
least 60 days overdue;
•
Outstanding collections totaling $2,000;
•
Outstanding individual or combined tax liens totaling $2,000 or more; and
•
Aggregate charge-offs of $1,000 within the preceding one year or $2,500 within
the preceding five years.
In the event a candidate’s credit history contained any of the aforementioned information,
GIS flagged the applicant “Review” and returned a summary to KU identifying the issues. GIS
3
did not inform KU of the race of any applicant.
Over the years, the manner in which KU utilized the information provided by GIS
changed.
From January of 2008 through January of 2011, KU approved all job applicants for
financial aid positions unless the applicant’s credit report showed defaulted student loans. These
applicants were rejected because DOE requirements prevented these individuals from accessing
the NSLDS and, according to KU, access to NSLDS was an essential function of financial aid
positions. During this time period, KU rejected 23 of 405 applicants because of defaulted
student loans.2
For non-financial aid positions, if GIS assigned an applicant’s credit report as “Review,”
a hiring manager could request an override of GIS’s determination from KU’s Controller. This
procedure varied over time. From January through May of 2008, the time during which Paul
Klier acted as Controller, any applicant whose file GIS flagged as “Review,” was automatically
rejected for hire. Sometime in mid-2008, Howard Rogoff assumed some of Klier’s duties,
including reviewing the applicant files flagged by GIS. Rogoff testified that rather than reject all
applicants flagged as “Review,” he implemented a “holistic” approach. Rogoff looked at
whether an applicant had “multiple” items on the GIS checklist. Rogoff determined whether, in
his judgment, an applicant’s credit history created an “increased risk” to the organization. From
September of 2008 through March of 2010, KU rejected 25 of 81 non-financial aid applicants.
In March of 2010, Rogoff amended the criteria he used in assessing the applicants marked
2
Thirteen other applicants were rejected for non-credit reasons such
as felony convictions.
4
“Review.” He relaxed the criteria he used to determine a candidate “ineligible.” For example,
instead of rejecting an applicant with $2,000 in past due balances that were at least 60 days
overdue, he considered only balances of greater than $5,000 that were at least 90 days overdue.
Other similar changes were made in connection with Rogoff’s assessments. Between March of
2010 and January of 2011, Rogoff rejected only one candidate based on the credit report. In
January of 2011, HireRight, Inc. replaced GIS as the entity providing credit reports. The criteria
used to flag an applicant as “Review” differed from those used by GIS. At this time, KU
reviewed all applicants that HireRight flagged as “Review.”
Defendant KHEC’s use of credit information
Unlike KU, KHEC did not extend conditional offers of employment. As a result, many
candidates were not offered positions even though KHEC had not yet learned the results of the
credit check. In addition, even if a credit report was graded “Review” by GIS, human resources
approved the candidate for hire if the report was flagged for no credit history and the candidate is
under 25 years old. If GIS assigned applicant as “Review” for any other reason, human
resources forwarded the file to Controller Kevin Corser. Corser reviewed all of the information
contained in the report in order to determine whether the individual might commit fraud or theft.
If Corser recommended against hiring an applicant, the recommendation was forwarded to
KHEC’s Group Vice President in charge of the particular school to which the applicant applied.
The Group Vice President made the final determination. In September of 2009, Rogoff replaced
Corser as the Controller responsible for reviewing credit reports. Rogoff also forwarded his
recommendation to the Group Vice President for final determination. Like KU, KHEC also
contracted with HireRight in January of 2011. KHEC, however, used HireRight for positions
5
above the Director Level. For lower-level positions, KHEC continued to use GIS. In November
of 2011, KHEC’s Chief Financial Officer, Jerry Dervin, assumed responsibility for reviewing
credit reports supplied by both HireRight and GIS. Dervin used the information “to some
degree” and considers the position and whether the applicant has “nonmedical charged-off debt
in excess of $5,000.”
Plaintiff EEOC’s use of credit reports
The EEOC also runs credit checks on job applicants. It appears that the Office of
Personnel Management (“OPM”), screens applicants for jobs with the EEOC. According to the
EEOC’s Personnel Suitability and Security Program Handbook, credit checks are required for 84
of the 97 positions at the EEOC. The handbook bases the need for credit checks on the notion
that “overdue just debts increase temptation to commit illegal or unethical acts as a means of
gaining funds to meet financial obligations.” According to plaintiff, it has never “unfavorably
adjudicated employees based on their debt issues.”
This lawsuit
The EEOC filed this lawsuit3 alleging that defendants’ use of credit history in making
hiring decisions violates certain provisions of Title VII, i.e., 42 U.S.C. § 2000e-2(a)(1), (a)(2)
and (k), because the practice has a disparate impact on Black applicants.
Defendants move for summary judgment and plaintiff moves for partial summary
judgment. Both motions are opposed. In addition, both parties move to strike the opposing
side’s expert reports. These motions are also both opposed.
3
Plaintiff amended the complaint after the Court dismissed certain
claims as time-barred.
6
STANDARD OF REVIEW
Rule 56(a) of the Federal Rules of Civil Procedure, as amended on December 1, 2010,
provides in relevant part that:
A party may move for summary judgment, identifying each claim or defense—or the part
of each claim or defense—on which summary judgment is sought. The court shall grant
summary judgment if the movant shows that there is no genuine dispute as to any
material fact and the movant is entitled to judgment as a matter of law.
Fed .R.Civ.P. 56(a).
Rule 56(e) provides in relevant part that “[i]f a party fails to properly support an assertion
of fact or fails to properly address another party's assertion of fact as required by Rule 56(c), the
court may ... consider the fact undisputed for purposes of the motion ... [and] grant summary
judgment if the motion and supporting materials—including the facts considered
undisputed-show that the movant is entitled to it.” Fed.R.Civ.P. 56(e).
Although Congress amended the summary judgment rule, the “standard for granting
summary judgment remain unchanged” and the amendment “will not affect continuing
development of the decisional law construing and applying” the standard. See, Fed.R.Civ.P. 56,
Committee Notes at 31.
Accordingly, summary judgment is appropriate when no genuine issues of material fact
exist and the moving party is entitled to judgment as a matter of law. Celotex Corp. v. Catrett,
477 U.S. 317, 322-23 (1986) (citing Fed. R. Civ. P. 56(c)); see also LaPointe v. UAW, Local
600, 8 F.3d 376, 378 (6th Cir. 1993). The burden of showing the absence of any such genuine
issues of material facts rests with the moving party:
[A] party seeking summary judgment always bears the initial
responsibility of informing the district court of the basis for its
motion, and identifying those portions of “the pleadings,
7
depositions, answers to interrogatories, and admissions on file,
together with affidavits,” if any, which it believes demonstrates the
absence of a genuine issue of material fact.
Celotex, 477 U.S. at 323 (citing Fed. R. Civ. P. 56(c)). A fact is “material only if its resolution
will affect the outcome of the lawsuit.” Anderson v. Liberty Lobby, 477 U.S. 242, 248 (1986).
Once the moving party has satisfied its burden of proof, the burden then shifts to the
nonmoving party. The court must afford all reasonable inferences and construe the evidence in
the light most favorable to the nonmoving party. Cox v. Kentucky Dep’t. of Transp., 53 F.3d
146, 150 (6th Cir. 1995) (citation omitted); see also United States v. Hodges X-Ray, Inc., 759
F.2d 557, 562 (6th Cir. 1985). However, the nonmoving party may not simply rely on its
pleading, but must “produce evidence that results in a conflict of material fact to be solved by a
jury.” Cox, 53 F.3d at 150.
Summary judgment should be granted if a party who bears the burden of proof at trial
does not establish an essential element of his case. Tolton v. American Biodyne, Inc., 48 F.3d
937, 941 (6th Cir. 1995) (citing Celotex, 477 U.S. at 322). Accordingly, “the mere existence of a
scintilla of evidence in support of plaintiff’s position will be insufficient; there must be evidence
on which the jury could reasonably find for the plaintiff.” Copeland v. Machulis, 57 F.3d 476,
479 (6th Cir. 1995) (quoting Anderson, 477 U.S. at 52 (1986)). Moreover, if the evidence is
“merely colorable” and not “significantly probative,” the court may decide the legal issue and
grant summary judgment. Anderson, 477 U.S. at 249-50 (citation omitted).
ANALYSIS
In order to establish a prima facie case of employment discrimination based on a
disparate impact theory, the plaintiff must: (1) identify a specific employment practice that is
8
being challenged; and (2) establish, through statistical means, that the identified employment
practice “caused the exclusion of applicants...because of their membership in a protected group.”
Watson v. Fort Worth Bank & Trust, 487 U.S. 977, 994 (1988). See also, Grant v. Metro. Gov’t
of Nashville & Davidson Cnty., 446 Fed.Appx.737, 740 (6th Cir. Aug. 26, 2011).
Assuming arguendo that the use of credit checks in the manner employed by defendants
constitutes a “specific employment practice,” the Court finds that plaintiff nonetheless fails to
present a prima facie case of disparate impact discrimination because plaintiff fails to provide
reliable statistical evidence of discrimination.
EEOC’s expert
Plaintiff offers the testimony and report of Dr. Kevin R. Murphy in support of its position
that defendants’ use of credit reports disparately impacted Black applicants in violation of Title
VII. Defendants move to exclude the testimony and report on the grounds that Dr. Muphy’s
method of determining race is scientifically unsound. In addition, defendants argue that Dr.
Murphy’s sample is not representative of the applicant pool. Defendants further claim that Dr.
Murphy’s analysis fails to account for non-discriminatory variables, including socio-economic
status, that could account for any disparate impact. The EEOC disputes these arguments.
a. “Race raters”
In order to determine the race of a particular applicant, the EEOC subpoenaed records
from the Departments of Motor Vehicles (“DMVs”) from 38 states and the District of Columbia.
Fourteen states and the District of Columbia provided records that identified an applicant’s race.
The remaining 24 states provided copies of the driver’s license photos pertaining to the
applicants. In order to determine the race of applicants from these states, Dr. Murphy assembled
9
a team of five “race raters,” who were asked to review each photograph and determine whether
the individual is “African-American,” “Asian,” “Hispanic,” “White,” or “Other.” Individuals
considered “multi-racial” were adjudged “Other.” Dr. Murphy required that four of the five
“race raters” agree (80%) in order to consider that applicant’s race. In all, the “race raters” were
shown 891 photographs.4 In 11.7% of the photographs, the “race raters” were unable to achieve
an 80% consensus with regard to the applicant’s race. The “race raters” worked individually in
assigning race, not as a group. Thus, each “race rater” came to an independent determination of
race without conferring with other “race raters.” The group was comprised of individuals with
advanced degrees in cultural anthropology, education, human development, psychology and
economics. Although the “race raters” have experience involving multiple racial populations,
none of them have prior experience in determining race via visual means. In some instances,
multiple photographs of the same individual were provided. In that case, Dr. Murphy or a staff
member chose the “clearest” photograph to provide to the “race raters.” The “race raters” were
provided with the applicant’s name along with the photograph.
Defendants argue that the judgment of these “race raters,” as to an applicant’s race is
nothing more than guesswork, which results in inherently unreliable data. According to
defendants, Dr. Murphy fails to identify any scientific study supporting the use of “race raters.”
Nor does Dr. Murphy point to any evidence suggesting this type of analysis has been tested,
subjected to peer review, or accepted by the scientific community. Defendants further argue that
4
It appears that 15 additional photographs were obtained after Dr.
Murphy filed his initial report. These photographs were analyzed
by a different group of three “race raters,” which included Dr.
Murphy and two staff members. A 100% consensus was reached
for these 15 photographs.
10
Dr. Murphy did nothing in advance to establish an error rate for the methodology. Rather, after
his final analysis, he compared the results of the photo review of 57 applicants for which he had
also possessed race data from the DMV or other supporting race data. According to defendants,
this sample set is too small to make accurate predictions about a data set containing 906
photographs. This is especially so given that “race” is the most important data field in this case.
Defendants also claim that bias may result because the “race raters” possessed the names of the
applicants. In other words, knowledge of the applicant’s name may affect what race the
applicant is assigned. Defendants argue that Dr. Murphy should have omitted from
consideration individuals with multiple photographs rather than allowing a subjective
determination of which photograph is the “clearest.”
The EEOC argues that in response to defendants’ motion challenging the use of “race
raters,” Dr. Murphy performed a new analysis to rebut defendants’ argument. According to the
EEOC, the new analysis shows that even excluding all of the applicants for which race was
identified based on DMV photographs, Dr. Murphy still concludes that the use of credit checks
has a disparate impact on Black applicants. In addition, the panel reached a consensus with
regard to 88.3% of the photographs. The EEOC claims that defendants fail to cite to any
research or case law to support its argument that the conclusions reached by the “race raters” are
unreliable.
The EEOC points out that the comparison of the 57 applicants for which both
photographic and “other” indications of race were present confirms the reliability of the panels’
conclusions. According to the EEOC, the cross-checking of the 57 applicants is not a “sample”
as that term is used in scientific research. Rather, it is “anecdotal corroboration” of the
11
reliability of the study. The EEOC claims that even if the data relied on by the “race raters” is
disputed, i.e., the Court refuses to consider the race of the applicants as depicted in Dr. Murphy’s
study, the applicants’ race simply remains a “disputed issue of fact.” Therefore, whether Dr.
Murphy can offer expert testimony regarding the race of the individual applicants is not
dispositive. The EEOC claims that defendants failed to comply with the EEOC’s Uniform
Guidelines on Employee Selection Procedures, which require employers to “maintain data from
which the adverse impact of their selection procedures can be determined.” As such, “defendant
filed no record evidence showing that it collected or maintained the required...data.”
In response, defendants argue that EEOC’s guidelines are not binding on employers and
do not excuse the EEOC’s failure to obtain race information directly from the applicants.
According to defendants, the EEOC learned that defendants do not retain race data on applicants
in 2009, giving them ample time to obtain race information by other means. Defendants argue
that the EEOC could have simply contacted the actual candidates themselves in order to
ascertain the relevant race data.5
The admissibility of expert testimony is governed by Fed. R. Evid. 702, which provides,
If scientific, technical, or other specialized knowledge will assist
the trier of fact to understand the evidence or to determine a fact in
issue, a witness qualified as an expert by knowledge, skill,
experience, training, or education, may testify thereto in the form
of an opinion or otherwise, if (1) the testimony is based upon
sufficient facts or data, (2) the testimony is the product of reliable
principles and methods, and (3) the witness has applied the
principles and methods reliably to the facts of the case.
5
The EEOC argues that phone surveys are not an acceptable method
of determining race because the response rates to the surveys
themselves suffer from racial disparities.
12
Under Daubert v. Merrell Dow Pharmaceuticals Inc., 509 U.S. 579, 597, 113 S.Ct. 2786
(1993), the trial judge serves as a “gate keeper” to determine whether an expert's testimony is
reliable and relevant. “The trial judge has considerable leeway in deciding how to go about
determining whether particular expert testimony is reliable.” U.S. v. Sanders, 59 Fed.Appx. 765,
767 (6th Cir. March 7, 2003)(citing Kumho Tire Co., Ltd. v. Carmichael, 526 U.S. 137, 152, 119
S.Ct. 1167)).
The Sixth Circuit has noted,
Daubert set forth a non-exclusive checklist for trial courts to use in assessing the
reliability of scientific expert testimony. Fed.R.Evid. 702 (advisory notes). The specific
factors explicated by the Daubert court are (1) whether the expert's technique or theory
can be or has been tested; (2) whether the technique or theory has been subject to peer
review and publication; (3) the known or potential rate of error of the technique or theory
when applied; (4) the existence and maintenance of standards and controls; and (5)
whether the technique or theory has been generally accepted in the scientific community.
Avery Dennison Corp. v. Four Pillars Enterprise Co., 45 Fed.Appx. 479, 483 (6th Cir. Sept. 3,
2002).
Upon review, the Court finds that the expert reports and testimony provided by Dr.
Murphy are inadmissible because plaintiff fails to present sufficient evidence that the use of
“race raters” is reliable. Simply put, plaintiff offers no evidence sufficient to satisfy any of the
Daubert factors.
Plaintiff offers no indication that the use of “race raters” has been or could be tested.
Even assuming that race rating is a technique susceptible to testing, plaintiff wholly fails to
provide any known or potential rate of error in the technique employed by the “race raters.”
When challenged by defendants in this regard, plaintiff points out that Dr. Murphy
“corroborated” the conclusions reached by the race raters. Specifically, he avers that there were
13
47 individuals for whom the DMV produced both photographic and race identification
information. Out of these 47 individuals, there was 95.7% agreement between the conclusions
reached by the “race raters” and the race identification provided by the DMV. In addition, there
were both photographs and race information provided by “PeopleSoft” for 10 applicants. Of
these 10, the “race raters” and the data contained in “PeopleSoft” matched “at least 80%” of the
time. In response, defendants argue that a sample size of 57 is much too small to provide any
degree of validation to the “race rating” process. Plaintiff acknowledges that the sample is not
being utilized as a “scientific sample.” Rather, Dr. Murphy indicates in his declaration as
follows:
In his November 30, 2012 Declaration, Dr. Saad criticizes this evidence as a “sample
size” that is “too small.” However, this evidence is not a “sample” as that term is used in
scientific research. Instead, the evidence represents the cross-references from my data set
where photos were rated for individuals who were also race identified by some other
source (e.g. DMV spreadsheet or “PeopleSoft”). While that evidence does not, alone,
establish the reliability of my photo rating methodology, it does support the conclusion
that the methodology is reliable.
As characterized by plaintiff, the cross-referencing the conclusions of the race raters with other
data is used only to show “anecdotal corroboration.” The Court finds, however, that “anecdotal
corroboration” is insufficient to substitute for a “rate of error” as discussed in Daubert. In order
to establish the reliability of the analysis, the Court must be convinced that the rate of error is
within acceptable parameters. Here, however, plaintiff acknowledges that the “sample” is not
intended to support a scientific analysis. Accordingly, the Court finds that plaintiff fails to show
that the “race rating” process has a statistically acceptable rate of error.
Likewise, plaintiff fails to show that the process of “rating race” by visual means has
been the subject of peer review or publication. Although the parties cite cases in which other
14
studies have been done involving the use of “race raters,” there is no indication that these studies
were subject to peer review. Moreover, the courts considering the issue disfavor reliance on
similar studies. See, e.g., United States v. Mesa-Roche, 288 F.Supp.2d 1172, n.50 (noting that
study in which teams of three persons studied photos to determine race was “seriously flawed”
because instances in which team members could not agree were discarded, thus resulting in the
overrepresentation of blacks); See also, United States v. Duque-Nava, 315 F.Supp.2d 1144, at
n.47 (same); Smith v. Chrysler Financial Co., 2004 WL 3201002 (D.N.J. Dec. 30,
2004)(rejecting data compilation process which consisted of examining black and white
photocopies of driver’s licenses because the analysis had “such great potential for error.”).
Plaintiff relies on United States v. Alcaraz-Arellano, 302 F.Supp.2d 1217, 1230 (D. Kan.
2004), in which the court cited a study indicating that “there is little doubt that there is high
reliability in determining race with regard to Blacks and Caucasians.” The Court agrees with
defendants, however, that plaintiff fails to provide a complete picture of the case in that the court
went on to state that distinguishing between “Whites and Hispanics is a ‘difficult task’ and
subject to error.” Id. In this case, the “race raters” were asked to distinguish among five race
categories, including a category identified as “other,” which itself included individuals deemed
“multi-racial.” Contrary to plaintiff’s reliance on Alcaraz-Arellano, the “race raters” were not
asked to identify applicants as only either Black or Caucasian.
In addition, plaintiff fails to establish that the “race rating” analysis was subjected to
proper controls. The Court is greatly concerned that Dr. Murphy was involved in both the
statistical analysis related to the race data, and the determination of the race data itself. Where
multiple photos existed of the same applicant, Dr. Murphy (or his staff) determined which photo
15
was “the clearest” and forwarded that photo on to the “race raters.” Even more troubling, Dr.
Murphy sat on one of the panels that determined the race of 15 of the applicants. Accordingly,
in those instances, Dr. Murphy both determined the underlying fact of race and also analyzed the
significance of his own determinations in concluding that defendants’ use of credit reports
disparately impacted Black applicants. In addition, the “race raters” were provided the names of
the applicants, which may create unintended bias on the part of the panel. For example, in a
close case, an individual may be likely to “rate” someone as “Hispanic” as opposed to
“Caucasian” if her surname is Gonzalez, a traditionally Hispanic surname.6 This may be so even
if the applicant is in fact Caucasian and married into an Hispanic family, or simply has the
surname Gonzalez and is not in fact Hispanic at all.
Plaintiff also presents no evidence that determining race by visual means is generally
accepted in the scientific community. In fact, the EEOC itself discourages employers from
visually identifying an individual by race and indicates that visual identification is appropriate
“only if an employee refuses to self-identify.7” In fact, even if an employer’s visual belief as to
an employee’s race differs from that disclosed by the employee, the employer must rely on the
race as identified by the employee. The EEOC argues that defendants’ “consternation” over the
EEOC’s own guidelines is “exaggerated.” According to the EEOC, it implemented these
guidelines not because of the accuracy of visual identification, but to facilitate and respect
6
Surprisingly, the EEOC suggests that Dr. Murphy’s race rating
team properly utilized the names on the photographs to corroborate
race. The Court does not believe that determining race (even
solely for corroboration) based on an individual’s name is
appropriate.
7
Http://www.eeoc.gov/employers/eeo1/qanda-implementation.cfm
(question 14)
16
‘individual dignity.’ Regardless of the reason supporting the pronouncement, it is clear that the
EEOC itself frowns on the very practice it seeks to rely on in this case and offers no evidence
that visual means is a method accepted by the scientific community as a means of determining
race8
The Court is also concerned at the lack of experience of the panelists. Although plaintiff
argues that each individual “race rater” has extensive experience with “multiple racial
populations,” not one of them has experience in identifying race via visual means.9 In response,
plaintiff argues that the racial composition of the applicant pool is a factual question, not one
requiring expert analysis. As such, the Court could allow the jury to determine the race of each
applicant. Assuming arguendo that such a process would be admissible at trial, plaintiff’s expert
report and testimony rely on the data as determined by the “race raters.” In that defendants have
moved for summary judgment, plaintiff must be able to provide some evidence at this juncture
demonstrating that the use of credit reports has a disparate impact on Black applicants. Because
plaintiff’s expert relied on data obtained by unreliable means, the Court finds that whether the
jury could ultimately “correct” the process employed by the “race raters” is irrelevant.
Plaintiff also argues that even if the Court disregards all of the applicants for which race
was determined by the “race raters,” there is still sufficient information to conclude that the use
of credit reports disparately impacts Black applicants. Specifically, Dr. Murphy avers as
8
Although not dispositive, it is noteworthy that the EEOC did not
contact the applicants directly and instead chose to assign a race to
the applicants, a process which is inconsistent with its policies
promoting personal dignity.
9
It is unclear to the Court how an advanced degree in economics,
for example, would make an individual particularly adept at
determining race.
17
follows:
In Kaplan’s November 30, 2012 motion to exclude me as an expert Kaplan challenges
generally the reliability of race-identifying an individual based on his or her DMV photo.
However, even if one excludes from my data set all individuals who were race identified
by DMV photo alone, the conclusion does not change. In response to Kaplan’s
November 30, 2012 motion, I removed from my data set the individuals who were race
identified by DMV photo alone. Then I re-analyzed the data set. As explained in Exhibit
D to this Declaration, the conclusion was the same. This reanalysis still shows that
Kaplan’s credit checks have a significant disparate impact on African Americans.
(Doc. 92 at ¶ 18).
In response to this argument, defendants appear to argue that production of various parts
of Dr. Murphy’s report is untimely. Defendants do not mention this analysis in particular, and
appear to argue that the “last” report provided by Dr. Murphy occurred on November 8, 2012.
Thus, it is not clear to the Court whether this particular analysis was provided to defendants on
November 8, 2012, or whether defendants saw the analysis for the first time on December 21,
2012, along with plaintiff’s brief in opposition. Regardless, the Court finds that this new
analysis is untimely and will not consider it for purposes of summary judgment. Aside from the
untimeliness, the Court is greatly concerned that there is no indication that the sample is random.
Rather, it appears that the sample would consist of only those specific individuals for which race
data was provided. There is no indication, for example, that the data is fairly distributed among
geographic areas or is in any other way “representative” of the applicant pool as a whole. As
discussed more thoroughly below, the “correction” made by Dr. Murphy to exclude all “race
rated” applicants does not save the report and testimony.
Because the use of “race raters” fails to satisfy any of the Daubert factors, the Court finds
18
defendants’ motion to exclude to be well-taken.10
b. Representative sample
Defendants argue that even if the Court considered “race rating” as performed in this case
to be a scientifically sound method for determining race, Dr. Murphy’s testimony and report
should nonetheless be excluded because the sample he relied on is not representative of the
applicant pool as a whole. According to defendants, Dr. Murphy’s data set consisted of 4,670
individuals, which was not a complete set of all applicants in the relevant time period. Of those
4,670 individuals, Dr. Murphy sampled 1,090 of the applicants. Defendants argue that Dr.
Murphy possessed information regarding 800 additional applicants, yet did not include these
individuals in his analysis. In addition, defendants claim that Dr. Murphy acknowledged that his
sample was not random and that, instead, he included only those individuals for whom he
possessed three categories of information. As a result, the data results obtained by Dr. Murphy
are skewed and unreliable. By way of example, defendants’ expert avers that the rate at which
individuals failed to credit check when reviewing the entire set of data is 13.3%. Yet, Dr.
Murphy’s subset of data has an overall failure rate of 23.8%. Similarly, there is a significant
difference in the numbers of individuals from each state:
Dr. Murphy included 214 Georgia residents in his analysis. In other words, Georgia
residents made up 19.6% of Dr. Murphy’s sample of 1,090. His entire set of 4,670,
however, included 474 residents, and Georgia residents comprised only 10.1% of the
individuals reflected in the GIS data. Dr. Murphy similarly over- or under- utilized other
States in his analysis. For instance, Wisconsin residents made up only .6% of the
individuals in Dr. Murphy’s sample, but comprised 7.3% of the applicants in his entire
10
In a footnote, defendants argue that Dr. Murphy’s reliance on the
race raters analysis is hearsay and should be excluded on that basis
as well. The Court need not address this issue as the expert report
and testimony are inadmissible for the other reasons set forth
herein.
19
data set.
(Doc. 84 at ¶ 11).
Defendants also point to other mistakes and errors Dr. Murphy made in conducting his
analysis.
In response, plaintiff argues that defendants fail to provide evidence that the inclusion of
additional applicants would have resulted in a finding of “no disparate impact.” Plaintiff further
argues that Dr. Murphy’s data set was based on appropriate criteria. Specifically, to be included
in Dr. Murphy’s data set, Dr. Murphy required evidence showing: the outcome of the applicant’s
credit check, the applicant’s race, and the applicant’s “identification data,” such as listed name.
Plaintiff argues that defendants’ expert did not opine that these requirements were “unnecessary”
or that their use rendered the data unreliable. Plaintiff also claims that each time defendants’
expert criticized an aspect of Dr. Murphy’s report, Dr. Murphy amended his analysis and
obtained the same result, namely, that the use of credit checks has a disparate impact on Black
applicants. Plaintiff does not address the specific argument that the data sample used by Dr.
Murphy is neither random nor representative of the applicant pool as a whole.
Upon review, the Court agrees with defendants. Although the Court does not go so far as
to require that a sample necessarily be perfectly random or precisely representative of the whole,
in this case defendants have proffered evidence demonstrating that the sample11 was not taken
randomly. Dr. Murphy testified as follows:
Q: Was it your intention in the process to include everyone for whom you had either a
DMV race status of a picture rated race status?
A: Our intention in the process was to include everyone for whom we had race
11
In reality, Dr. Murphy’s data consists of a “sample of a sample.”
20
information, information about third-party vendor outcome, and information about the
decisions that were made by Kaplan, so it was really a three-part test for who would be
included in our analysis.
Q: And so your sample wasn’t meant to be a random sample, correct?
A: Our sample was meant to include every one for whom we had the three pieces of
information I just recited....
(Doc. 83-5 at p. 207)
It appears that plaintiff argues that since Dr. Murphy consistently applied the same
“criteria” in order to obtain his sample, it is statistically sound. But, there is no indication that
the application of these criteria resulted in a fair cross section of the applicant pool. The
application of “criteria” in and of itself does not mean that the analysis is admissible. For
example, Dr. Murphy may have had race information only from certain localities which might
skew the sample.
In addition, defendants’ expert points out that the results from the sample are not in fact
representative of the applicant pool. As set forth above, the failure rate of the entire applicant
pool amounted to 13.3%, yet, Dr. Murphy’s subset of data has a significantly higher failure rate
of 23.8%. In addition, the disparity of applicant representation by state could arguably have an
impact. Plaintiff, however, did not respond to this argument at all. Thus, while the Court
acknowledges that a statistical analysis need not be “perfect” in order to be admissible, the
evidence presented by defendants’ expert shows that the sample was not randomly taken and is
not representative of the applicant pool as a whole. Because the EEOC fails to present any
evidence indicating that the analysis is nonetheless admissible, the Court finds defendants’
argument to be well-taken. The Court will exclude Dr. Murphy’s expert report and testimony on
this additional ground as well.
21
The Court rejects, however, defendants’ argument that Dr. Murphy’s failure to include
additional applicants for which he possessed data renders the analysis unreliable. Defendants do
not point the Court to any evidence or case law dictating that an expert must rely on all data in
his or her possession. Providing the sample size is sufficient, there appears to be no requirement
that all data must be analyzed. Accordingly, the argument is rejected.
c. Non-discriminatory variances and timeliness
Having concluded that plaintiffs’ expert evidence is inadmissible as set forth above, the
Court need not reach whether it should also be excluded because Dr. Murphy fails to account for
“non-discriminatory variances.” Nor will the Court address defendants’ argument that the
additional conclusions in the reports are untimely, except as previously set forth.
Because plaintiff fails to present admissible evidence showing that the use of credit
reports “caused the exclusion of applicants...because of their membership in a protected group,”
plaintiff cannot set forth a prima facie case of disparate impact discrimination. As such,
summary judgment in favor of defendants is warranted and the Court will not reach defendants’
alternative arguments, i.e., job relatedness, business necessity, and estoppel.
Motions filed by plaintiff
Plaintiff moves to exclude defendants’ expert witness and further moves for partial
summary judgment on defendants’ “affirmative defense.” In these motions, plaintiff argues that
there is no evidence supporting defendants’ arguments regarding job relatedness, business
necessity, and estoppel. The Court, however, will not reach these issues in light of the fact that
plaintiff fails to state a prima facie case. Accordingly, these motions are MOOT.
22
CONCLUSION
For the foregoing reasons, Defendants’ Motion for Summary Judgment (Doc. 79) and the
Motion to Exclude the Reports and Testimony of Dr. Kevin R. Murphy (Doc. 82) are
GRANTED. The EEOC’s Motion for Partial Summary Judgment (Doc. 80) and Motion to
Exclude Defendant’s Expert Michael G. Aamodt (Doc. 90) are DENIED as MOOT.
IT IS SO ORDERED.
/s/ Patricia A. Gaughan
PATRICIA A. GAUGHAN
United States District Judge
Dated: 1/28/13
23
Disclaimer: Justia Dockets & Filings provides public litigation records from the federal appellate and district courts. These filings and docket sheets should not be considered findings of fact or liability, nor do they necessarily reflect the view of Justia.
Why Is My Information Online?