August 9, 2013
ROGER W. TITUS, District Judge.
For many employers, conducting a criminal history or credit record background check on a potential employee is a rational and legitimate component of a reasonable hiring process. The reasons for conducting such checks are obvious. Employers have a clear incentive to avoid hiring employees who have a proven tendency to defraud or steal from their employers, engage in workplace violence, or who otherwise appear to be untrustworthy and unreliable. However, under Title VII of the Civil Rights Act of 1964, a specific hiring policy may constitute an unlawful employment practice if it has a disparate impact on the basis of race, color, religion, sex or national origin and the employer fails to demonstrate that the challenged practice is job-related for the position in question and consistent with business necessity. 42 U.S.C. § 2000e-2(k)(1)(A)(i).
As the agency responsible for investigating possible violations of the Act and enforcing anti-discrimination laws in the employment realm, the EEOC has brought this action against the Defendant, Freeman, alleging that it has implemented a hiring policy that, though facially neutral, has a discriminatory effect on African-American and male applicants. The present case is only one of a series of actions recently brought by the EEOC against employers who rely on criminal background and/or credit history checks in making hiring decisions. For example, in two recent complaints filed against discount retailer Dollar General Corp. and car manufacturer BMW, the EEOC claimed that those employers improperly used criminal background checks to bar potential employees, resulting in a disparate impact on African-American applicants. Scott Thurm, Employment Checks Fuel Race Complaints, Wall St. J., June 12, 2013, at A1, available at http://online.wsj.com/article/ SB10001424127887323495604578539283518855020.html.
Because of the higher rate of incarceration of African-Americans than Caucasians, indiscriminate use of criminal history information might have the predictable result of excluding African-Americans at a higher rate than Caucasians. Indeed, the higher incarceration rate might cause one to fear that any use of criminal history information would be in violation of Title VII. However, this is simply not the case. Careful and appropriate use of criminal history information is an important, and in many cases essential, part of the employment process of employers throughout the United States. As Freeman points out, even the EEOC conducts criminal background investigations as a condition of employment for all employees, and conducts credit background checks on approximately 90 percent of its positions. ECF No. 114-44 at 7-8.
Thus, it is not the mere use of any criminal history or credit information generally that is a matter of concern under Title VII, but rather what specific information is used and how it is used. Because of this, it is simply not enough to demonstrate that criminal history or credit information has been used. Rather, a disparate impact case must be carefully focused on a specific practice with an evidentiary foundation showing that it has a disparate impact because of a prohibited factor.
Proof of disparate impact requires reliable and accurate statistical analysis performed by a qualified expert. As the Supreme Court has noted, "the inevitable focus on statistics in disparate impact cases" results in a very "high standard of proof" that can be difficult for plaintiffs to meet. Watson v. Fort Worth Bank and Trust, 487 U.S. 977, 992, 999 (1988). Merely pointing to "statistical disparities in the employer's work force" is not sufficient; the plaintiff must provide "statistical evidence of a kind and degree sufficient to show that the practice in question has caused the exclusion of applicants for jobs or promotions because of their membership in a protected group." Id. at 994. Even if the plaintiff is able to proffer such evidence, neither "courts [n]or defendants [are] obliged to assume that plaintiffs' statistical evidence is reliable, " but can challenge the techniques or data used in the analysis. Id. at 996. Moreover, even if meaningful and reliable statistics are presented, a plaintiff is required to separate out and identify the "specific" employment practice that is allegedly responsible for the disparate impact, particularly where employers combine objective and subjective hiring criteria. Id. at 994.
While some specific uses of criminal and credit background checks may be discriminatory and violate the provision of Title VII, the EEOC bears the burden of supplying reliable expert testimony and statistical analysis that demonstrates disparate impact stemming from a specific employment practice before such a violation can be found. For the reasons explained below, the EEOC has failed to do so in this case. Accordingly, summary judgment shall be granted to Defendant.
Defendant is a provider of integrated services for expositions, conventions, corporate events, meetings, and exhibit programs, with annual revenues exceeding $1.3 billion. ECF No. 114-3 at ¶¶ 3-4. The family-owned company employs over 3, 500 full-time and 25, 000 part-time and seasonal workers, with offices in major cities throughout the United States. Id.
Like many employers, Defendant has experienced problems with embezzlement, theft, drug use, and workplace violence by its employees. ECF Nos. 114-38 at 6-9; 114-39 at 12, 16-18. In 2001, Defendant began conducting background checks on applicants so that it could better evaluate the trustworthiness, reliability, and effectiveness of prospective employees. Id. The background checks were designed with five goals in mind: "(1) avoid exposure to negligent hiring/retention lawsuits; (2) increase the security of Defendant's assets and employees; (3) reduce liability from inconsistent hiring or screening practices; (4) proactively reduce the risk of employee-related loss; and (5) mitigate the likelihood of an adverse incident occurring on company property that could jeopardize customer or employee confidence." ECF Nos. 114-1 at 5; 114-38 at 6-8, 18; 114-39 at 9-10, 13; 114-4 at 1.
From July 20, 2006 to August of 2011, the types of background checks performed by Defendant varied with the nature of the job sought. ECF Nos. 114-41; 114-38 at 20; 114-40 at 6. For "general employees, " i.e., those who did not hold credit sensitive jobs, the check included only a criminal history investigation and social security verification. ECF No. 114-41. For "credit sensitive" positions, the check also included a credit history review. Id. A position was deemed credit sensitive if the employee holding that position had access to client or company credit card information, handled money, checks, or similar valuable items, had budgetary authority, had authority to make agreements with respect to customer invoices, or made purchases from vendors. ECF Nos. 114-38 at 17, 22; 114-39 at 21-24. Finally, for company officers, general managers, and department heads, the Defendant performed an education and certification verification in addition to the above checks. ECF No. 144-41. In total, Defendant regularly ran credit checks for 44 job titles, compared to 109 positions that did not require a credit check. ECF No. 114-7 at 10-13.
Defendant's standard employment application form asked: "Have you ever pleaded guilty to, or been convicted of, a criminal offense." ECF No. 144-4. If the applicant responded in the affirmative, they were given space to describe the date and circumstances. Id. The form contained the following advisement:
A conviction does not automatically mean you will not be offered a job. What you were convicted of, the circumstances surrounding the conviction and how long ago the conviction occurred are important considerations in determining your eligibility. Give all the facts, so that a fair decision can be made.
Id. Applicants were also required to sign a form authorizing a vendor, PreScreen America (PSA), to conduct the background investigation. ECF No. 114-3 at ¶ 12. The authorization form contained the same questions as the employment application regarding prior criminal offenses. ECF No. 114-5.
Generally, the background check was run after the applicant was offered and accepted a position, but before he or she began work. ECF No. 114-38 at 15-16. For credit checks, PSA obtained credit histories from TransUnion, a national credit bureau; for criminal checks, PSA collected information on convictions and their equivalents and active criminal warrants, but not arrests. ECF No. 114-45 at 5-7. Defendant limited its consideration of convictions to those that occurred within seven years of the application date. Id. at 8.
Defendant used a multi-step evaluation process to review the information obtained by PSA and determine whether an applicant was qualified to begin work. First, Defendant considered whether the applicant was truthful about his or her criminal convictions on the application and authorization forms. Under one of the few bright-line rules in Defendant's policy, an applicant who failed to disclose a conviction, seriously misrepresented the circumstances of a criminal offense, or made any other materially dishonest statement on the application, was automatically disqualified. ECF No. 108-15 at ¶ 16; ECF No. 114-3 at ¶ 17. Second, Defendant examined any outstanding arrest warrants. Applicants with pending warrants were given a reasonable opportunity to resolve the matter and have the warrant withdrawn; failure to do so made it unlikely, but not impossible, for the applicant to be hired. ECF No. 108-15 at ¶ 14. Third, Defendant considered the existence of any criminal convictions which the applicant committed, or was released from confinement for, within the past seven years. Id. Finally, Defendant evaluated whether the criminal conduct underlying a particular conviction made an applicant unsuitable for employment. ECF No. 108-7 at 3-5. Types of convictions that were of particular concern to Defendant, and would generally disqualify an applicant, included those involving violence, destruction of private property, sexual misconduct, felony drug convictions, or job-related misdemeanors. ECF No. 108-15 at ¶ 15. In general, initial decisions by Defendant's office manager not to hire an applicant because of a particular conviction were reviewed and approved by Defendant's senior vice president for human resources or vice president of benefits and compliance. ECF No. 114-43 at 15-19.
For credit checks during the relevant time period covered by the EEOC's complaint, Defendant's policy consisted of a list of hiring criteria. ECF Nos. 108-15 at ¶ 7; 108-17. Applicants whose credit histories revealed any of the following issues were excluded from employment for a credit-sensitive position:
1) More than two accounts of $300 or more that were 90 days past due;
2) More than three collection accounts that were not medically related;
3) More than two paid charge-offs in the prior 12 months;
4) Any unpaid charge-offs in the prior 12 months;
5) A car repossessed in the prior three years;
6) A house foreclosure in the prior three years;
7) Filed for bankruptcy in the prior seven years;
8) A judgment in the prior seven years;
9) A default on student loans;
10) Any unsatisfied liens;
11) Any satisfied liens in the prior three years;
12) Any delinquency in paying child support.
The EEOC has not challenged any of the specific criteria or procedures described above, but has merely alleged that Defendant's policy of conducting criminal and credit background checks, as a whole, produces a disparate impact on protected classes.
On January 17, 2008, applicant Katrina Vaugn filed a discrimination complaint with the EEOC, asserting that Defendant violated Title VII by rejecting her for employment based on information about her credit history. ECF No. 27-2, Ex. 1. Based on Ms. Vaughn's charge, on September 30, 2009, the EEOC filed a complaint in this Court under Sections 706 and 707 of Title VII of the Civil Rights acts of 1964, as amended, 42 U.S.C. §§ 2000e-5(f)(1) and (3) and 2000e-6. ECF No. 1.
In its complaint, the EEOC alleged that since February of 2001, Defendant engaged in a "pattern or practice" of discrimination against African-American job applicants by using poor credit history as a hiring criterion (the "credit class"), and against African-American, Hispanic, and male job applicants by using criminal history as a hiring criterion (the "criminal class"). Id. at ¶¶ 8-11. The EEOC asserted that these hiring criteria have a significant disparate impact on the identified suspect classes and are not job-related or consistent with business necessity. Id. The EEOC sought an injunction prohibiting Defendant using any credit history or criminal history information when hiring employees. Id. at ¶ A. It also sought "make whole" relief for affected class members, including back pay with prejudgment interest and other affirmative relief necessary to eradicate the effects of unlawful employment practices, such as instatement and front pay in lieu thereof. Id. at ¶¶ B-E.
A period of contentious discovery and a flurry of motions activity followed the filing of the complaint. Subsequent decisions by this Court limited the scope of the claims currently at issue and put an end to several of the EEOC's initial contentions. On April 27, 2010, this Court granted Defendant's motion to dismiss all claims relating to hiring decisions made before March 23, 2007, the date 300 days before the January 17, 2008 charge of discrimination that served as the legal predicate for the EEOC's claims. ECF No. 19. On January 31, 2011, this Court also granted Defendant's partial motion for summary judgment and dismissed all claims relating to decisions based on criminal history information and made prior to November 30, 2007, the date 300 days before the EEOC notified Defendant on September 25, 2008 that it was expanding its investigation to include potential race discrimination based on use of criminal history records. ECF No. 43. Finally, on August 24, 2012, the parties stipulated to a dismissal with prejudice of the EEOC's claim that Defendant unlawfully discriminates against Hispanics. ECF No. 95.
As a result of these developments, the EEOC's current "credit class" consists of 51 African-Americans who allegedly were unlawfully excluded from employment between March 23, 2007 and August 11, 2011. ECF No. 114 at 2. The EEOC's current "criminal class" consists of 83 African-Americans and males who allegedly were unlawfully excluded from employment between November 30, 2007 and July 12, 2012. Id.
After numerous extensions to the deadlines for expert disclosures, the parties completed their expert disclosures in accordance with the Court's amended scheduling order, which mandated that the EEOC's Rule 26(a)(2) expert disclosures be completed by July 18, 2012,  Defendant's 26(a)(2) expert disclosures be completed by November 23, 2012, and the EEOC's rebuttal expert disclosures be completed by January 9, 2013. ECF No. 100. In support of their disparate impact allegations, the EEOC submitted initial and amended reports containing a statistical analysis of Defendant's hiring patterns prepared by expert Kevin R. Murphy. ECF No. 108-8 (Murphy Amended Report). The EEOC supported Murphy's analysis with a report prepared by Beth M. Huebner, which purported to replicate Murphy's analysis and results based on the same data utilized by Murphy. ECF No. 108-9 (Huebner Report).
On December 18, 2012, Defendant filed a motion to preclude the testimony of Murphy and Huebner, pointing to an overwhelming number of inaccuracies in the underlying data that they both used. See ECF No. 108. On January 22, 2013, the EEOC filed with its opposition a "supplemental" report and declaration by Murphy. ECF No. 121. In this new report, Murphy attempted to "moot" Defendant's criticisms of his original report by offering "updated" statistical charts based on a new analysis that supposedly fixed all of the identified errors and still produced the same results indicating disparate impact. ECF No. 121-1. On March 2, 2013, Defendant filed a reply analyzing the new report and contending that it was still plagued by troublesome errors that rendered it unreliable. ECF No. 130.
Three days after filing its motion to preclude the EEOC's expert reports, Defendant filed a motion for summary judgment on December 21, 2012, arguing that, having failed to present any reliable statistical evidence, the EEOC could not demonstrate the existence of disparate impact. ECF No. 114. The EEOC responded on February 6, 2013, and Defendant replied on March 13, 2013. ECF Nos. 126, 138.
Finally, on March 11, 2013, the EEOC filed a motion for leave to file a sur-reply in response to the motion to exclude expert testimony, or, in the alternative, to strike Defendant's reply and exhibits, claiming that the reply introduced new arguments and evidence. ECF No. 137. The EEOC did not attach to its motion a proposed sur-reply, but only vaguely outlined in its supporting memorandum what it hoped to accomplish with such a filing. ECF No. 137-1. Defendant responded on March 21, 2013, and the EEOC filed a reply on April 8, 2013. ECF Nos. 139, 141.
Oral argument on all three motions was presented by the parties at a hearing before this Court on June 19, 2013. ECF No. 147.
STANDARD OF REVIEW
Summary judgment is proper if there are no issues of material fact and the moving party is entitled to judgment as a matter of law. Celotex Corp. v. Catrett, 477 U.S. 317, 322 (1986); Francis v. Booz, Allen & Hamilton, Inc., 452 F.3d 299, 302 (4th Cir. 2006). A material fact is one that "might affect the outcome of the suit under the governing law." Spriggs v. Diamond Auto Glass, 242 F.3d 179, 183 (4th Cir. 2001) (quoting Anderson v. Liberty Lobby, 477 U.S. 242, 248 (1986)). A dispute of material fact is only "genuine" if sufficient evidence favoring the non-moving party exists for the trier of fact to return a verdict for that party. Anderson, 477 U.S. at 248-49. However, the nonmoving party "cannot create a genuine issue of material fact through mere speculation or the building of one inference upon another." Beale v. Hardy, 769 F.2d 213, 214 (4th Cir. 1986). "A party opposing a properly supported motion for summary judgment may not rest upon the mere allegations or denials of [his] pleadings, ' but rather must set forth specific facts showing that there is a genuine issue for trial.'" Bouchat v. Baltimore Ravens Football Club, Inc., 346 F.3d 514, 522 (4th Cir. 2003) (alteration in original) (quoting Fed.R.Civ.P. 56(e)).
The Court may only rely on facts supported in the record, not simply assertions in the pleadings, in order to fulfill its "affirmative obligation... to prevent factually unsupported claims or defenses' from proceeding to trial." Felty v. Grave-Humphreys Co., 818 F.2d 1126, 1128 (4th Cir. 1987) (citing Celotex, 477 U.S. at 323-24). When ruling on a motion for summary judgment, "[t]he evidence of the nonmovant is to be believed, and all justifiable inferences are to be drawn in his favor." Anderson, 477 U.S. at 255 (emphasis added). However, "if the evidence is merely colorable or not significantly probative, it may not be adequate to oppose entry of summary judgment." Thompson Everett, Inc., v. Nat'l Cable Adv., 57 F.3d 1312, 1323 (4th Cir. 1995).
To prevail on a claim of disparate impact, a plaintiff must show that a certain class of applicants is disproportionately and adversely impacted by a particular employment practice on the basis of their race, color, religion, sex or national origin. See Title VII of the Civil Rights Act of 1964, 42 U.S.C. § 2000e-2(k). The plaintiff bears the burden of proving discriminatory impact by showing statistical disparities between the number of protected class members in the qualified applicant group and those in the relevant segment of the work-force. Wards Cove Packing Co. v. Atonio, 490 U.S. 642, 650 (1988). Upon such a showing, the burden then shifts to the employer to prove that the allegedly discriminatory policies or practices are job-related for the position in question and consistent with business necessity. See 42 U.S.C. § 2000e-2(k). A plaintiff must do more than merely raise a statistical inference of discrimination before the burden shifts to the employer; it must actually prove the discriminatory impact. See Garcia v. Spun Steak Co., 998 F.2d 1480, 1486 (9th Cir. 1993). If the plaintiff cannot make a prima facie showing of disparate impact, it is appropriate to grant summary judgment for the employer. Id.
The EEOC attempts to make a statistically sufficient demonstration of disparate impact through the expert reports of Kevin R. Murphy and Beth M. Huebner. These experts analyzed data produced by Defendant in an attempt to show that African-American applicants fail Defendant's credit background checks at a significantly higher rate than other races, and that male and African-American applicants fail Defendant's criminal background checks at a significantly higher rate than females and non-blacks.
Defendant argues that the experts' conclusions are based on unreliable data and are rife with analytical errors, in addition to being untimely, and thus are inadmissible to demonstrate the existence of disparate impact. Defendant also argues that the experts do not isolate and identify which aspect of Defendant's credit and criminal record check processes allegedly causes the disparate impact, thereby failing to make out a prima facie case under Title VII. Finally, Defendant contends that the national statistics proffered by the EEOC cannot, on their own, demonstrate that the challenged policies cause a disparate impact on suspect classes. Because this Court finds Defendant's numerous objections to be well-taken based on the evidence before it, the expert testimony proffered by the EEOC will be precluded, and summary judgment will be granted in favor of Defendant.
I. Murphy's Reports and Conclusions Must Be Precluded as Unreliable and Untimely.
A. Standard for admissibility of expert evidence.
Federal Rule of Evidence 702 provides that an expert qualified "by knowledge, skill, experience, training, or education, may testify" to scientific, technical, or other specialized knowledge if it will assist the trier of fact. Such testimony is only admissible 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." Id. Federal Rule of Evidence 104(a) allows for the exclusion of unreliable expert testimony if the proponent cannot establish by a preponderance of the evidence that these requirements are met. See Daubert v. Merrell Dow Pharms., 509 U.S. 579, 592 n. 10 (1993); Cooper v. Smith & Nephew Inc., 259 F.3d 194, 199 (4th Cir. 2001).
Expert statistics are unreliable if they are based on "incomplete data sets and inadequate statistical techniques." Watson v. Fort Worth Bank and Trust, 487 U.S. 977, 996-97 (1988) (plurality opinion). Among other factors, courts "consider the rate of error of the methods employed by the expert, the existence and maintenance of standards used in the expert's methods, and whether the expert's methods have been generally accepted by his or her respective community." Anderson v. Westinghouse Savannah River Co., 406 F.3d 248 (4th Cir. 2005) citing Daubert, 509 U.S. at 594. Coding errors should be considered and can warrant the exclusion of the expert's testimony. See, e.g., Malletier v. Dooney & Bourke, Inc., 525 F.Supp.2d 558, 630 (S.D.N.Y. 2007) (coding errors contributed "to the cumulative effect of the methodological errors" that warranted exclusion of a consumer confusion survey); EEOC v. Sears, Roebuck & Co., 628 F.Supp. 1264, 1304, 1305 (N.D. Ill. 1986) (stating that the EEOC "has made so many general coding errors that its data base does not fairly reflect the characteristics of applicants for commission sales positions at Sears."), aff'd, 839 F.2d 302 (7th Cir. 1988).
Moreover, a court should exclude expert statistical testimony when the data relied upon is connected to the expert's opinion "only by the ipse dixit of the expert. A court may conclude that there is simply too great an analytical gap between the data and the opinion proffered." Gen. Electric Co. v. Joiner, 522 U.S. 136, 146 (1997) (citations omitted); see Raskin v. Wyatt Co., 125 F.3d 55, 67 (2d Cir. 1997) (expert testimony that contains "elementary" error is not helpful); Wilkinson v. Rosenthal & Co., 712 F.Supp. 474, 479 (E.D. Pa. 1989) (same).
B. Murphy's inaccurate database renders his conclusions unreliable.
In its original motion to exclude, Defendant challenged the admissibility of Murphy's report by identifying a number of inaccuracies in his database. Rather than denying or explaining those criticisms in its opposition brief, the EEOC instead chose to file a supplemental report and declaration by Murphy that, after supposedly "fixing" the identified issues, purported to reach the same findings regarding disparate impact. Upon consideration of Defendant's subsequent analysis of Murphy's multiple reports and its comparison of the original data produced by Defendant to that included in Murphy's database, as set forth in Defendant's papers and at the hearing before the undersigned, there appear to be such a plethora of errors and analytical fallacies underlying Murphy's conclusions to render them completely unreliable, and insufficient to support a finding of disparate impact.
1) Murphy had access to, but did not utilize, the materials necessary to create an unbiased, accurate testing database.
Murphy's analysis suffers primarily because it was not based on a random sample of accurate data from the relevant applicant pool and time period. The EEOC attempts to blame any issues with Murphy's testing database on the Defendant's failure to produce sufficient information during discovery. To the contrary, Defendant has clearly demonstrated that it properly supplied the EEOC during discovery with complete background check logs for the entire period covered by the complaint. See ECF No. 148. These logs list applicant names, the branch where application was made, the date a background check was performed, and each applicant's resulting status (coded as Hireable: Y or N). For individuals who were listed as not hireable, a brief explanation was provided (e.g., "failed drug test" or "Resisting arrest, falsified app"). In addition to the logs, Defendant also provided during discovery all EEO datasheets filled out by applicants during the relevant time period, which provide information on the applicant's race and gender, the criminal background check reports prepared by PSA (PSA reports), and applicant flow logs containing applicant names, application date, position sought, branch applied to, race, gender, and application disposition. Finally, Defendant provided EEOC with all of its adverse action notices sent to applicants during the relevant period (i.e., notices sent to individuals who are denied a position because of their credit history or criminal background). See ECF No. 108-3 at ¶¶ 16-17.
Before this present action was initiated, the EEOC conducted an investigation of Defendant's background check policies based on the complaint by Katrina Vaugn in 2008. During the course of that investigation, Defendant produced to the EEOC two Excel spreadsheets, one that contained information about applicants who were credit checked from January 1, 2005 to October 13, 2008, and the other that contained information about applicants who were criminal background checked at 18 of Freeman's 39 branches from January 1, 2007 to October 14, 2008. See ECF No. 130-1 at ¶¶ 2-4. The spreadsheets listed applicants' names and social security numbers, their application dates, the branch where they applied, and the date a background search was requested. Id.; ECF Nos. 130-3; 130-4.
In Murphy's original report, he admits that he had access to information on 58, 892 applicants through the discovery materials. His ultimate testing database, however, included fewer than 2, 014 unique applicants, with many of the 2, 014 entries being duplicates. ECF No. 108-15 at ¶ 19. Murphy instead relied almost entirely on the two Excel spreadsheets in creating his database, with only a few individuals cherry-picked for inclusion from the discovery materials. ECF No. 130-1 at ¶¶ 2-7. As explained below, Murphy's database is so full of material flaws that any evidence of disparate impact derived from an analysis of its contents must necessarily be disregarded.
2) Murphy's database does not cover the time period identified in the EEOC's claims.
The time period covered by the spreadsheets - and hence by Murphy's database - represents only a distorted fraction of the time period relevant in this case. The credit check spreadsheet spans January 1, 2005 to October 13, 2008, whereas the EEOC's credit claims cover applicants from March 23, 2007 to August 11, 2011. Similarly, the criminal check spreadsheet spans January 1, 2007 to October 14, 2008, but the EEOC's criminal claims cover applicants from November 30, 2007 and July 12, 2012. In each case, the time period included in the spreadsheets is both over and under inclusive, making it an inadequate basis for examining disparate impact during the relevant time period.
Generally, "statistical evidence cannot serve as a basis for proving discrimination beyond the time period analyzed." King v. Gen. Elec. Co., 960 F.2d 617, 626 (7th Cir. 1992). The EEOC justifies Murphy's use of the stale data in place of more current data by citing National Railroad Passenger Corp. v. Morgan, 536 U.S. 101, 113 (2002), Warren v. Halstead Industries, Inc., 802 F.2d 746, 753 (4th Cir. 1986), and Black Law Enforcement Officers Ass'n v. City of Akron, 824 F.2d 475, 483 (6th Cir. 1987) for the principle that time-barred acts can be considered relevant background evidence to a claim of discrimination. ECF No. 121 at 20. Background evidence is admissible under Fed.R.Evid. 401 "as an aid to understanding" the case. Fed.R.Evid. 401 advisory committee's note. It is not, however, an adequate substitute for data from the relevant time period. See Payne v. Travenol Labs., Inc., 673 F.2d 798, 823, 830 (5th Cir. 1982) (remanding case because trial court imposed a remedy for the period 1974 to 1976 when plaintiffs had produced no applicant flow statistics for that period, having cut off their analysis at 1974).
Several principles prevent Murphy from relying on such "background evidence." First, as explained in the next section, Murphy cherry-picked the data that he included from outside the time period covered in the Excel spreadsheets, as opposed to drawing a random sample. Second, Murphy did not perform a mathematical extrapolation to the time periods that were not covered by his analysis, but simply used his randomly assembled database to infer disparate impact for the relevant period. Third, Murphy deliberately ignored the data that was available from the relevant time period in favor of data from outside the time period. In another recent case where the EEOC attempted to introduce a report by Murphy purporting to show evidence of disparate impact, the district court precluded Murphy from testifying because his analysis was based on a "sample of a sample" that "was not taken randomly" and "is not representative of the applicant pool as a whole." EEOC v. Kaplan Higher Learning Education Corp., 2013 WL 322116, at *10-11 & n.11 (N.D. Ohio Jan. 28, 2013). The same rationale can be used here to exclude Murphy's report.
3) Murphy enhanced his disparate impact results by including "fails" from the discovery materials.
Out of the 2, 014 applicants in Murphy's database, only a handful of the entries was derived from the discovery materials produced in this case. In an egregious example of scientific dishonesty, Murphy cherry-picked certain individuals from the other discovery materials in an attempt to pump up the number of "fails" in his database and suggest that his database included sufficient data for the time period past fall of 2008. Specifically, Murphy added only 19 individuals who applied to Defendant after October 14, 2008 to his database, all but one of whom failed the relevant background checks. ECF No. 108-15 at ¶ 7. Because of the small overall number of "fails" in the database, Murphy's 19 additions conveniently increased the fail percentage by over twenty percent, rendering it a meaningless, skewed statistic.
4) Murphy's database includes data generated under the old credit check policy.
By relying almost exclusively on the Excel spreadsheets produced during the investigation rather than the discovery materials produced during this case, Murphy included credit check data from before July of 2006, when Defendant had a stricter credit history policy in place. Accordingly, for any pre-July 2006 credit check "fails" included in Murphy's database, it cannot be known whether those applicants would actually have failed the credit check under the relevant policy. In fact, such pre-July 2006 "fails" constitute over 30% of the total number of credit fails, thus rendering Murphy's analysis of credit fail rates completely untrustworthy.
5) Murphy's database omits all data from half of Defendant's branch offices.
The Excel spreadsheets contained information from only half of Defendant's branch offices; Murphy's database correspondingly contains no data from half of Defendant's branches. Murphy offers no explanation as to why the branches that he included should be considered a representative sample.
6) Muphy's database is rife with material errors and unexplained discrepancies.
The mind-boggling number of errors contained in Murphy's database could alone render his disparate impact conclusions worthless. The pervasiveness of Murphy's errors can be demonstrated by looking at the mistakes that occurred for just one subset of individuals. In a report prepared by its damages expert, Sovan Tun, the EEOC identified, and provided race, gender, and background check information on 41 applicants for whom it is seeking back pay. Out of this group of 41, seven indviduals were not included in Murphy's database. Out of the remaining 34, seven appeared in Murphy's database without a race code, one was incorrectly coded as passing the criminal background check, two were incorrectly coded as failing the criminal background check, one has an incorrect race code, five have incorrect gender codes, nine are listed twice and double-counted in Murphy's results, and three who failed the credit check are not coded with a credit check result. ECF No. 108-15 at ¶¶ 32-43. It is thus impossible to rely on Murphy's conclusions regarding disparate impact, when his results are so clearly skewed by such an abundance of vital errors and mistakes.
7) Murphy's supplemental reports cannot save his flawed analysis.
Murphy claims that his subsequent reports and analyses incorporate all of Defendant's criticisms and still find the existence of disparate impact. But, contrary to Murphy's assertions, his first supplemental report and declaration does not correct all of the errors mentioned above. For instance, he did not remove the 19 cherry-picked individuals from his database. He also did not change the incorrect coding of race and pass/fail status for several individuals. Murphy added only a handful of new applicants to his database in a laughable attempt to better capture the relevant time period; in total, Murphy's augmented analysis includes a mere 12 applicants for the period from October 14, 2008 to January 5, 2010. Amazingly, despite his claims of doing so, Murphy did not exclude the pre-July 20, 2006 credit check data from his new analysis. Finally, Murphy once again managed to introduce fresh errors into his new analysis, including many additional duplicates, material coding errors, and more double-counting. In evaluating fails/passes by race, Murphy double-counts 13 individuals. Suspiciously, 11 of these double-counts are "fails" while only 2 of them are "passes." ECF Nos. 130-3; 130-4. Whether intentional or not, Murphy's continued pattern of producing a skewed database plagued by material fallacies gives this Court no choice but to entirely disregard his disparate impact analysis.
8) Even if Murphy's supplemental reports and declarations were reliable, they are untimely.
Under the operative scheduling order in this case, which was amended numerous times to grant the parties extensions for meeting various discovery deadlines, the EEOC should have completed all of its expert disclosures by July 18, 2012. Despite this deadline, the EEOC filed Murphy's amended report on July 26, 2012, and Murphy's supplemental report and declaration almost half a year later on January 22, 2013. The EEOC even attempted to present to the Court yet another declaration/report by Murphy at the hearing on June 19, 2013.
The EEOC should not be allowed to make a mockery of procedural standards by continually offering new expert reports for this Court's consideration, well past the applicable deadline. Despite being titled "supplemental" reports, Murphy's latest filings do not qualify as supplements under Federal Rule of Civil Procedure 26(e), but are instead poorly disguised attempts to counter Defendant's arguments with new expert analyses. Except under limited circumstances not here present, expert disclosures are fixed targets, and not ones movable at will.
Rule 26(e) requires that an expert report be supplemented when a "party learns that in some material respect the disclosure or response is incomplete or incorrect." The rule does not create a "right to produce information in a belated fashion." Goodbys Creek, LLC v. Arch Ins. Co., 07-cv-947, 2009 WL 1139575, at *2 (M.D. Fla. Apr. 27, 2009). "Supplementation under the Rules means correcting inaccuracies, or filling the interstices of an incomplete report based on information that was not available at the time of the initial disclosure." Keener v. United States, 181 F.R.D. 639, 640 (D. Mont. 1998). Indeed, new reports provided under the guise of "supplementation" cannot be produced to "address the criticisms that [defendants] raised in their memorandum in support of summary judgment." Gallagher v. S. Source Packaging, LLC, 568 F.Supp.2d 624, 629 (E.D. N.C. 2008).
Murphy's continually changing analyses are based on materials that were available to him prior to his initial submission deadline, not on newly discovered information. Rule 26(e) is not a "loophole through which a party... who wishes to revise her disclosures in light of her opponent's challenges to the analysis and conclusions therein, can add to them to her advantage after the court's deadline for doing so has passed." Luke v. Family Care & Urgent Med. Clinics, 323 F.Appx. 496, 500 (9th Cir. 2009). Murphy's supplemental reports and declarations are clearly not proper supplementation, but instead fall into that category of counterarguments strictly prohibited by federal courts.
Because the opinions disclosed in Murphy's declaration were untimely and do not qualify as supplements under Rule 26, they must be excluded under Rule 37(c) unless the untimely filing "was either substantially justified or harmless." Campbell v. United States, 470 F.Appx. 153, 157 (4th Cir. 2012). In Campbell, the plaintiff filed an inadequate Rule 26(a)(2) expert report five days late. More than a month later, Plaintiff filed a "supplement, " which included a new expert report supported by many pages of exhibits. The Fourth Circuit held that the new report could not be considered, rejecting the assertion that it was permitted under Rule 26(e)'s "continuing duty" to supplement discovery disclosures. The court stated: "To construe [Rule 26(e)] supplementation to apply whenever a party wants to bolster or submit additional expert opinions would [wreak] havoc in docket control and amount to unlimited expert opinion preparation." Id. at 157 (citing Sharpe v. United States, 230 F.R.D. 452, 462 (E.D.Va. 2005) (alteration in original)). The court found that the plaintiff's late report was not "substantially justified or harmless, " because it merely recast its expert's opinions to comply with Rule 26(a).
In Cochran v. Brinkmann Corp., 08-cv-1790-WSD, 2009 WL 4823858 (N.D.Ga. Dec. 9, 2009), the court considered issues much like those presented here. The plaintiffs opposed the defendant's Daubert and summary judgment motions by submitting affidavits of expert witnesses "to address the grounds on which Defendant's motions to disqualify Plaintiffs' expert witnesses... are based." Id. at *13-15. The court held that the affidavits were untimely and inadmissible, having been filed after the court's deadline for filing Rule 26(a)(2) reports. In keeping with these precedents, Murphy's supplemental reports and declarations must be considered untimely, unjustified in their lateness, and therefore inadmissible.
II. Huebner's report is likewise unreliable and inadmissible.
The report by the EEOC's other expert, Beth Huebner, suffers from the same reliability problems that plague Murphy's report. Huebner's report adds nothing significant to Murphy's analysis, but purports merely to replicate his analysis and confirm his conclusions regarding disparate impact. ECF No. 108-9. Huebner relies on the same flawed database used by Murphy, thereby rendering her conclusions subject to all the criticisms detailed above. Moreoever, the EEOC never offered amended versions of Huebner's report, as they did with Murphy. Thus, it cannot claim that a later version of Huebner's report somehow compensated for all of the underlying data errors that render her original report unreliable.
Based on the inaccuracy of the database used by Murphy and Huebner, the unreliability of their analyses, and the untimeliness of Murphy's supplemental declarations and reports, Defendant's motion to preclude the expert testimony offered by the EEOC shall be granted.
III. National Statistics Alone Cannot Prove That Defendant's Policies Had a Disparate Impact on Certain Classes.
The EEOC argues that even if Murphy and Huebner's reports are inadmissible to demonstrate disparate impact, the national statistics cited in their reports are sufficient evidence of disparate impact. To use general population statistics to create an inference of disparate impact, the general populace must be representative of the relevant applicant pool. See Wards Cove Packing Co. v. Atonio, 490 U.S. 642, 650 (1989) (the "proper comparison [is] between the racial composition of [the at-issue jobs] and the racial composition of the qualified... population in the relevant labor market."). Here, there is no indication that such is the case. The general population pool "cannot be used as a surrogate for the class of qualified job applicants, because it contains many persons who have not (and would not) be" applying for a job with Defendant. Id. at 653-54.
Moreover, the general statistics that the EEOC's experts rely on relate to things that are not even considered under Defendant's hiring criteria, such as arrest and incarceration rates. See ECF Nos. 108-8, 108-9. The complete inapplicability of such numbers to the present situation is particularly notable when considering that both Huebner and Murphy cite to the notable statistical disparity between White and Hispanics with regard to credit ratings, arrests, and convictions, but found no significant difference between those races with respect to Defendant's hiring policies.
With neither national statistics nor expert analysis to support its allegations of disparate impact, the EEOC's case cannot survive. The EEOC bears the burden of establishing a prima facie case, through use of statistics or other evidence, of disparate impact because of a prohibited factor. The burden is not on Defendant to conduct its own analysis to rebut the results produced by the EEOC's flawed report. It is sufficient for Defendant to point out the numerous fallacies in Murphy's report, which raise the specter of unreliability. See EEOC v. Sears, Roebuck & Co., 839 F.2d 302, 313 (7th Cir. 1988) (en banc) ("The cases cited by the EEOC to support its argument that Sears had the burden of rebutting its statistical analysis with more refined, accurate and valid' statistical evidence did not state that the defendant must produce such evidence to succeed in rebutting the plaintiffs' case. Instead, those cases indicated that a defendant could or was entitled to' use such a means of rebuttal" (citations omitted)); see also Dickerson v. U.S. Steel Corp., 472 F.Supp. at 1310, n.2 (taking note that the defendant could produce a report without the alleged flaws, but stating that defendant has no burden to put forward such evidence if it can defeat plaintiff's case by other means). Without any evidence of disparate impact to support its claim, the EEOC has failed to establish one of the primary elements of its claim. Summary judgment must accordingly be entered in favor of Defendant.
IV. The EEOC Has Failed to Isolate a Specific Employment Practice of Defendant's that Allegedly Caused a Disparate Impact.
Even if Murphy and Huebner's reports were admissible, and summary judgment could not be granted for Defendant on that ground, the EEOC and their experts have still failed to identify the specific policy or policies causing the alleged disparate impact. Under Title VII, it is not enough for the plaintiff to show that "in general" the collective results of a hiring process cause disparate impact. Statistical analysis must isolate and identify the discrete element in the hiring process that produces the discriminatory outcome. See 29 U.S.C. §2000e-2(k)(1); Wards Cove, 490 U.S. at 656 (plaintiff must offer evidence "isolating and identifying the specific employment practices that are allegedly responsible for any observed statistical disparities") (citation omitted). Where a hiring process has multiple elements, the plaintiff must identify the element(s) that it is challenging and "demonstrate that each particular challenged employment practice causes a disparate impact, " unless it can demonstrate that "the elements" are not capable of separation for purposes of analysis. 29 U.S.C. §2000e-2(k)(1)(B) (emphasis added).
Defendant argues that by bringing only a general complaint against Defendant's credit and criminal screening policies as a whole, the EEOC has failed to identify the particular policy, process, or rule that allegedly has a disparate impact on certain groups. As described previously, Defendant's background investigation policies are not simple, one-step processes. Rather, they involve different types of checks depending on the specific job an individual is seeking, consideration of both subjective and objective criteria, and examination of a long list of factors, any one of which might control the ultimate employment decision.
The EEOC argues, in turn, that analyzing each sub-factor separately for potential disparate impact is not feasible, meaningful, or necessary. The cases that it cites in support of this position are not applicable, however, as they all predate the Congressional codification of the disparate impact particularity requirement in 1991. See 29 U.S.C. §2000e-2(k)(1)(B).
Several courts have indicated that if a policy can be broken down into discrete parts, the plaintiff must identify which part is responsible for creating racial or gender disparities. In Smith v. City of Jackson, for example, the plaintiffs' claim that their employer's pay plan had a disparate impact on older workers was rejected when they did not identify "any specific test, requirement, or practice within the pay plan that has an adverse impact, " where the pay plan "was based on reasonable factors other than age." 554 U.S. 228, 241 (2005). See also Bennett v. Nucor Corp., 656 F.3d 802, 817-18 (8th Cir. 2011) (rejecting expert's statement that the employer's promotion procedure was "not capable of separation for analysis of individual components of the process, " where the procedure combined discernible objective and subjective components); Easterling v. State of Conn., 783 F.Supp.2d 323, 326, 333 (D. Conn. 2011) (plaintiff demonstrated disparate impact where failure on any single part of a four-part physical fitness test causes the candidate to fail the entire test).
While it is true that "[t]here is no legal requirement to use the smallest possible unit of analysis" in disparate impact cases, Stagi v. National Railroad Passenger Corp., 2010 WL 3273173, at *12-13 (3d Cir., Aug. 16, 2010), the EEOC has made no effort to break down what is clearly a multi-faceted, multi-step policy. Though it is theorectically possible that one or more of Defendant's background check considerations causes a disparate impact on certain classes, the EEOC has failed to demonstrate which such factor is the alleged culprit. Accordingly, the EEOC has failed to establish this element of its case, and summary judgment must be entered for Defendant.
IV. The EEOC Is Not Entitled to File a Sur-Reply.
Defendant supported its motion to exclude with a declaration and exhibits from lay witness Suzanne Bragg, who pointed out the numerous factual and analytical fallacies in Murphy and Huebner's reports. See ECF No. 108-15, Bragg Decl. In its opposition to Defendant's motion, the EEOC submitted an 89-page supplemental report and declaration by Murphy in an attempt to render "moot" Defendant's criticisms of the original report and analyses. See ECF No. 121-1, Murphy Decl. The EEOC also argued that many of the alleged errors in Murphy's dataset were actually errors in Defendant's own data, as provided to the EEOC. See ECF No. 121. Finally, the EEOC contended that, even if Murphy's report was not reliable, it could still establish a prima facie case of disparate impact through its experts' discussion of published statistics on race and gender disparities in the criminal justice system. Id. Defendant subsequently filed a reply attacking the supplemental report and addressing the EEOC's other arguments. See ECF No. 130.
The EEOC contends that Defendant's reply brief should now be stricken, or that the EEOC should be allowed to file a sur-reply, because Defendant's reply brief included "new" evidence and "new" arguments. ECF No. 137. Specifically, Defendant filed with its reply two new declarations by lay witness Suzanne Bragg and attorney John Koerner, each with multiple exhibits, and a new declaration by paralegal Julia McEnroe. See ECF Nos. 130-132. Based on the dates they were signed and their content, all of the declarations appear to have been prepared by Defendant after reviewing the EEOC's opposition to its motion. Id. In her new declaration, Bragg hypothesizes as to how Murphy incorrectly manipulated the data provided by Defendant during discovery to construct his database, and points out the failure of his supplemental report to remedy the issues present in his original and amended reports. See ECF No. 130-1. The EEOC also argues that Defendant asserts for the first time in its reply that the use of published external statistics showing racial disparities in criminal histories is not sufficient to establish disparate impact.
Allowing a party to file a sur-reply is within the Court's discretion, see Local Rule 105.2(a), but they are generally disfavored. See Chubb & Son v. C & C Complete Servs., LLC, 2013 WL 336718, at *9 (D. Md. Jan. 23, 2013). "Surreplies may be permitted when the moving party would be unable to contest matters presented to the court for the first time in the opposing party's reply." Khoury v. Meserve, 268 F.Supp.2d 600, 605 (D. Md. 2003).
None of the arguments or evidence introduced in Defendant's reply brief can truly be considered "new, " such that the EEOC should be afforded another opportunity to address them. First, the issues with Murphy's amended report that Defendant discusses in its reply are not novel; the vast majority of them are the same ones Defendant identified as to the original report. Instead of addressing those issues head-on in its opposition, the EEOC choose instead to attempt to render them "moot" by having Murphy file a supplemental report. The rationale behind the "ordinary rule in federal courts... that an argument raised for the first time in a reply brief or memorandum will not be considered" is to avoid prejudice to the party who is not afforded a chance to rebut a new argument. Clawson v. FedEx Ground Package Sys., Inc., 451 F.Supp.2d 731, 734-35 (D. Md. 2006). The EEOC did not lack a chance to respond to many of the arguments that it points to now; it simply squandered its chance, and cannot now claim prejudice.
Second, Defendant's "new" arguments and evidence are more correctly characterized as responsive arguments to the claims raised in the EEOC's opposition brief. Where "the arguments made by Defendants in their reply brief are merely responses to new arguments made by Plaintiffs in their response, " a sur-reply is not appropriate. Aguilar v. LR Coin Laundromat, 2012 WL 1569552, at *2-3 (D. Md. May 2, 2012) (denying motion to strike or to allow sur-reply where plaintiffs' opposition introduced arguments about defendants' sales volume, and defendants' reply included an affidavit and tax return not attached to the original motion, but used to address the arguments in the plaintiffs' opposition). For example, Defendant's argument regarding the insufficiency of external published statistics to prove disparate impact could not have come as a surprise to the EEOC. Defendant very briefly broached the topic in its reply only in response to arguments by the EEOC that such statistics could, on their own, establish disparate impact. Although a discussion of the national statistics was included in the EEOC's expert reports, it was not necessarily clear that the EEOC intended to rely on such statistics exclusively, as a back-up method of proving disparate impact, until the EEOC asserted as much in its opposition brief. Defendant was certainly under no obligation to proactively address such an argument in its original motion to exclude. Defendant did not deliberately hold arguments back from its original motion in an attempt to spring them on the EEOC when a response could no longer be made; it merely argued against those points that the EEOC asserted in its opposition.
Third, the EEOC has not clearly identified what it hopes to accomplish through the filing of a sur-reply. Rather than a direct attack on Bragg's declaration, the EEOC's description of its intended sur-reply suggests that it would merely reiterate old arguments (e.g., that the inaccuracies in Murphy's database do not affect his conclusions) or provide unsubstantiated excuses (e.g., that gaps in Defendant's document production hindered the EEOC's ability to analyze relevant data). Although it asserts that it has "not had an opportunity to challenge the accuracy or relevancy" of Bragg's second declaration, see ECF No. 137-1 at 4, the EEOC never points to any facts or conclusions that it specifically believes to be inaccurate. Nor did the EEOC attach a copy of its proposed sur-reply such that the Court could better determine the relevancy and importance of allowing the document to be included in the record. Vague allegations of inaccuracies alone do not sway this Court to permit the filing of a sur-reply. See, e.g., Brown v. Prince George's Cnty., No. DKC 07-2591, 2012 WL 3012573, at *4 (D. Md. July 20, 2012) (refusing to permit sur-reply offered to identify manner in which reply "mischaracterize[d]' the arguments... presented in opposition"); Khoury, 268 F.Supp.2d at 606 (denying the plaintiff's motion for leave to file a sur-reply in order "to correct what [the plaintiff] perceive[d] to be Defendant's misrepresentations" of the record and the law); Thomas v. Artino, 723 F.Supp.2d 822, 833 n.2 (D. Md. 2010) (declining to consider a sur-reply that sought to address "Plaintiff's misstatements of facts and law raised for the first time in [the] reply memorandum"). In short, the EEOC has offered no sufficient reason why a sur-reply would be justified in this case.
Finally, the allowance of a sur-reply would fly in the face of the Court's scheduling order and the federal rules regarding expert discovery. Based on the Murphy declaration provided to the Court at the June 19, 2013 hearing, it appears that any sur-reply filed by the EEOC would consist primarily of yet another untimely "supplemental" report and/or declaration by Murphy, filed well past the deadline for the EEOC's expert disclosures. As discussed previously, Murphy's supplemental reports and declarations are not proper supplements under Rule 26(e). They are not based on new information, but are instead attempts to move the target yet again and get a second (or third or fourth) bite at the apple, changing their underlying analysis to combat whatever arguments are presented by Defendant. The EEOC will not be allowed to abuse the sur-reply form to submit yet another version of Murphy's expert report. In short, procedural requirements governing disclosures of expert reports do not establish "moving targets, " but rather fixed points in time upon which the parties are entitled to rely.
The story of the present action has been that of a theory in search of facts to support it. But there are simply no facts here to support a theory of disparate impact resulting from any identified, specific practice of the Defendant.
Indeed, any rational employer in the United States should pause to consider the implications of actions of this nature brought based upon such inadequate data. By bringing actions of this nature, the EEOC has placed many employers in the "Hobson's choice" of ignoring criminal history and credit background, thus exposing themselves to potential liability for criminal and fraudulent acts committed by employees, on the one hand, or incurring the wrath of the EEOC for having utilized information deemed fundamental by most employers. Something more, far more, than what is relied upon by the EEOC in this case must be utilized to justify a disparate impact claim based upon criminal history and credit checks. To require less, would be to condemn the use of common sense, and this is simply not what the discrimination laws of this country require.
For the foregoing reasons, Defendant's motion for summary judgment [ECF No. 114] shall be granted, Defendant's motion to preclude expert testimony [ECF No. 108] shall be granted, and Plaintiff's motion for leave to file sur-reply [ECF No. 137] shall be denied. A separate order follows.