Citations
- 215 F. Supp. 3d 140
Full opinion text
MEMORANDUM AND ORDER
Casper, United States District Judge
I. Introduction
Plaintiff Equal Employment Opportunity Commission (“EEOC”) has filed this lawsuit against Defendants Texas Roadhouse, Inc., Texas Roadhouse Holdings LLC and Texas Roadhouse Management Corp. (collectively, “Texas Roadhouse”) alleging a pattern or practice of age discrimination under the Age Discrimination in Employment Act (“ADEA”). The EEOC alleges that between 2007 and 2014, Texas Roadhouse engaged in a pattern or practice of discrimination by which its standard operating procedure was to discriminate against individuals over age 40 — the protected age group (“PAG”) — for front-of-house (“FOH”) positions nationwide. D. 35 ¶¶ 26-28. As explained below, the Court ALLOWS in part and DENIES in part Texas Roadhouse’s motion to strike the reports and anticipated testimony of Dr. David L. Crawford, D. 584, DENIES EEOC’s motion to strike portions of the expert report and proffered testimony of Dr. Ali Saad, D. 593, DENIES EEOC’s motion to strike expert report and proposed testimony of Dr. Eric Dunleavy, D. 600, and DENIES Texas Roadhouse’s motion for summary judgment, D. 587.
II. Standard of Review
A. Motion to Strike Expert Testimony
Pursuant to Fed. R. Evid. 702, a qualified expert witness can testify “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.” United States v. Mooney, 315 F.3d 54, 62 (1st Cir. 2002) (quoting Fed. R. Evid. 702). This rule “assign[s] to the trial judge the task of ensuring that an expert’s testimony both rests on a reliable foundation and is relevant to the task at hand.” Cipollone v. Yale Indus. Prod., Inc., 202 F.3d 376, 380 (1st Cir. 2000) (quoting Daubert v. Merrell Dow Pharm., Inc., 509 U.S. 579, 597, 113 S.Ct. 2786, 125 L.Ed.2d 469 (1993)). “[T]he district court must perform [this] gatek-eeping function by preliminarily assessing ‘whether the reasoning or methodology is ... valid and ... properly can be applied to the facts in issue’ ” by examining multiple factors through a flexible, case-specific inquiry. Seahorse Marine Supplies, Inc. v. Puerto Rico Sun Oil Co., 295 F.3d 68, 80-81 (1st Cir. 2002) (quoting Daubert, 509 U.S. at 592-93, 113 S.Ct. 2786). Ultimately, the purpose of the inquiry is “to determine whether the testimony of the expert would be helpful to the jury.” Cipollone, 202 F.3d at 380. As long as the expert’s testimony is found to rest upon reliable grounds, “the traditional and appropriate means of attacking shaky but admissible evidence” is through “[vigorous cross-examination, presentation of contrary evidence, and careful instruction on the burden of proof.” Milward v. Acuity Specialty Prods. Grp., Inc., 639 F.3d 11, 15 (1st Cir. 2011) (quoting Daubert, 509 U.S. at 590).
B. Motion for Summary Judgment
The Court will grant summary judgment when there is no genuine dispute on any material fact and the undisputed facts show that the moving party is entitled to judgment as a matter of law. Fed. R. Civ. P. 56(a). “An issue is genuine if ‘it may reasonably be resolved in favor of either part/ at trial, and material if it ‘possesses] the capacity to sway the outcome of the litigation under the applicable law.’ ” Iverson v. City of Boston, 452 F.3d 94, 98 (1st Cir. 2006) (alteration in original) (internal citations omitted). The movant “bears the burden of demonstrating the absence of a genuine issue of material fact.” Rosciti v. Ins. Co. of Pa., 659 F.3d 92, 96 (1st Cir. 2011) (quoting Carmona v. Toledo, 215 F.3d 124, 132 (1st Cir. 2000)). If the moving party meets this burden, then the non-movant must “with respect to each issue on which she would bear the burden of proof at trial, demonstrate that a trier of fact could reasonably resolve that issue in her favor.” Borges ex rel. S.M.B.W. v. Serrano-Isern, 605 F.3d 1, 5 (1st Cir. 2010). “The test is whether, as to each essential element, there is sufficient evidence favoring the nonmoving party for a jury to return a verdict for that party.” DeNovellis v. Shalala, 124 F.3d 298, 306 (1st Cir. 1997) (internal quotation mark and citation omitted). In deciding a summary judgment motion, the Court views the record in the light most favorable to the non-moving party, drawing all reasonable inferences in his favor. Noonan v. Staples, Inc., 556 F.3d 20, 25 (1st Cir. 2009). This standard is no different in a pattern-or-practice discrimination case.
III. Factual Background
The following facts are drawn from the parties’ statements of material facts, D. 589, D. 616, D. 617, D. 644, and supporting documents and'are undisputed unless otherwise noted.
A. The Texas Roadhouse Corporate Structure
As of December 30, 2014, Texas Roadhouse owned and operated 368 Texas Roadhouse restaurant locations and franchised another 79 restaurants. D. 617 ¶ 2; D. 644 ¶ 2. Its founder, chairman and CEO is W. Kent Taylor (“Taylor”). Id.
The “Support Center,” Texas Roadhouse’s headquarters, D. 589 ¶ 5; D. 616 ¶ 5; D. 617 ¶ 10; D. 644 ¶ 10, dispatches the Market Partners, the Regional Market Partners, Training Managers and Regional Human Resources and Marketing employees to its locations all over the country. D. 617 ¶¶ 10-11; D. 644 ¶¶ 10-11.
Each individual Texas Roadhouse restaurant generally has one salaried Service Manager (“SM”) with primary responsibility for managing FOH operations, one salaried Kitchen Manager (“KM”) with primary responsibility for managing back-of-the-house (“BOH”) operations and one Managing Partner with primary responsibility for the day-to-day operations of the entire restaurant. D. 589 ¶¶ 16-17; D. 616 ¶¶ 16-17; D. 617 ¶ 9; D. 644 ¶ 9. The Managing Partner is the overall manager for a single restaurant location. D. 589 ¶ 16; D. 616 ¶ 16. Managing Partners must follow Texas Roadhouse recipes and conform to the company guidelines and policies. D. 617 ¶ 6; D. 644 ¶ 6; D. 618-20 at 99-100.
Texas Roadhouse restaurants are grouped into “markets” of up to fifteen restaurants per market, each of which is assigned to a Market Partner. D. 589 ¶ 18; D. 616 ¶ 18. Each Market Partner supervises the operation of all of the restaurants in his market and is responsible for ensuring adherence to all aspects of the Texas Roadhouse concept, strategy and standards of quality, which includes weekly visits and oversight of the hiring and development of each restaurant’s management team. D. 589 ¶¶ 18-19; D. 616 ¶¶ 18-19; D. 617 ¶ 12; D. 644 ¶ 12; D. 618-10 at 5-6; D. 618-26 at 30-31. Market Partner duties generally include overseeing the operations of the restaurant locations, hiring and developing each restaurant’s management team, overseeing the hiring of staff lower-level managerial positions in the restaurants alongside one of the Support Center’s Field Support Staffing Manager and are generally on-site during hiring for a new store opening. D. 617 ¶¶ 12-13, 15-16; D. 644 ¶¶ 12-13, 15-16; D. 618-21 at 11-12. Each Market Partner’s job includes ensuring that the Managing Partners are focused on and executing the company-wide operational goals, including training and employee image. D. 589 ¶ 14; D. 616 ¶ 14; D. 617 ¶¶ 3, 5; D. 644 ¶¶ 3, 5. A Market Partner has the authority to fire a Managing Partner for bad performance or not adhering to company standards. D. 617 ¶ 14; D. 644 ¶ 14. Market Partners themselves can be terminated for failing to follow the directions and policies of the company. D. 617 ¶ 23; D. 644 ¶ 23.
A set of markets are further grouped into regions, each assigned to a Regional Market Partner. D. 589 ¶20; 616 ¶20. Regional Market Partners supervise Market Partners. D. 589 ¶21; D. 616 ¶21; D. 617 ¶ 21; D. 644 ¶ 21. One aspect of the Regional Market Partners’ job is to ensure that Managing Partners are executing the company’s overall operational goals, including training and employee image. D. 617 ¶¶ 3, 5; D. 644 ¶¶ 3, 5. The majority of Regional Market Partners make visits to restaurants to observe what is going on in each restaurant. D. 617 ¶ 22; D. 644 ¶ 22.
Steve Ortiz (“Ortiz”) was the former Chief Operating Officer of Texas Roadhouse. D. 617 ¶ 24; D. 644 ¶ 24. In his role, he supervised all of the employees in operations including all of the Managing Partners, Market Partners, Regional Partners and the Vice President of Training, Juanita Coleman. Id. Ortiz reported to the CEO, Taylor. D. 617 ¶ 27; D. 644 ¶27. During the 2007 to 2014 period, Ortiz was responsible for building the Texas Roadhouse brand, which included training and supervising the Managing Partners, Market Partners and Regional Market Partners. D. 618-4 at 10; D. 618-20 at 31-32; D. 618-21 at 17. In addition to Ortiz, Taylor was a hands-on CEO who visits at least three to four Texas Roadhouse locations per month and provides feedback on a variety of issues he identifies at those restaurants directly to Managing Partners and Market Partners. D. 617 ¶¶ 31-35; D. 644 ¶¶ 31-35. Beyond Ortiz and Taylor, Training Managers conduct regular visits to their assigned restaurants and human resources staff conduct “culture checks” at store locations when a particular restaurant has received complaints or when a Market Partner requests such a check. D. 617 ¶¶ 43-44; D. 644 ¶¶ 43-44.
B. Texas Roadhouse Training
As mentioned, Texas Roadhouse has a company-wide set of “operational goals” that include “training” and “employee image.” D. 617 ¶ 3; D. 644 ¶3. Texas Roadhouse has devoted resources to recruiting and training managers and hourly employees and its trainings include teaching managers on hiring for their store locations. D. 617 ¶ 48; D. 644 ¶ 48. Texas Roadhouse provides ten to twelve months of training to new Managing Partners which includes four months in the Manager in Training (“MIT”) program and attending the Legendary Learning and MP101 programs at the Support Center that includes receiving “Kent’s Top Ten,” a document written by Taylor that instructs managers about the mood, look and character of Texas Roadhouse restaurants. D. 617 ¶¶ 49, 54; D. 644 ¶¶49, 54. The Legendary Learning program is a five-day program where Managing Partners, service managers, kitchen managers and others are taught Texas Roadhouse’s operational goals and the company’s culture. D. 617 ¶ 55; D. 644 ¶ 55. Texas Roadhouse further trains managers in training centers — store locations with management teams that train new managers — across the country in the MIT program before sending managers to their assigned locations. D. 617 ¶¶ 57-58; D. 644 ¶¶ 57-58. When new managers complete their training, Texas Roadhouse confirms that each person has the knowledge and ability to be a manager according to criteria in the Texas Roadhouse MIT Workbook. D. 617 ¶ 60; D. 644 ¶60. Texas Roadhouse expects Managing Partners to operate his or her location in a manner consistent with the training provided. D. 617 ¶ 52; D. 644 ¶ 52.
As part of their overall training, Managing Partners receive training on how to hire for different positions including FOH positions comprised of hosts, servers, server assistants and bussers and bartenders. D. 589 ¶ 58; D. 616 ¶ 58; D. 617 ¶ 61; D. 644 ¶ 61. Managers in training are validated on legendary hiring, behavioral interviewing and “hiring for Image and Heart.” D. 617 ¶ 62; D. 644 ¶ 62. Texas Roadhouse also provided interviewing and hiring handouts to all attendees of Legendary Learning and Service Manager University. D. 617 ¶ 63; D. 644 ¶ 63. Texas Roadhouse also provides hiring training to managers at Market meetings. D. 617 ,¶ 64; D. 644 ¶ 64. Texas Roadhouse believes it is important that all Texas Roadhouse managers with hiring responsibility receive and apply consistent hiring practices. D. 617 ¶¶ 66-68; D. 644 ¶¶ 66-68.
At some point throughout this training, Texas Roadhouse provides managers with at least some training concerning anti-discrimination laws. D. 589 ¶ 35; D. 616 ¶ 35. In addition, the Texas Roadhouse Interviewing and Hiring Guide for managers also provides examples of improper questions based upon age. D. 589 ¶ 50; D. 616 ¶ 50.
C. Texas Roadhouse’s Management of Hiring
The People Department has developed the Company’s Equal Employment Opportunity Policy and Harassment and Discrimination Policy, D. 589 ¶ 23; D. 616 ¶ 23, and Texas Roadhouse has published written non-discrimination policies in various documents made available to its employees, D. 589 ¶¶ 25, 29; D. 616 ¶¶25, 29. Company template applications also include language that indicates age discrimination is prohibited and that the company is an equal opportunity employer. D. 589 ¶ 31, 33; D. 616 ¶ 31, 33.
Texas Roadhouse believes that hiring for FOH positions is important because those positions are a reflection of the company’s brand and culture. D. 617 ¶ 78; D. 644 ¶ 78. In its goals and training, Texas Roadhouse emphasizes that a key to success at each store location was focusing on image which includes hosts who are happy and attractive, bartenders who are all-American types and servers who are great looking, although the parties dispute the exact contours of this “image.” D. 617 ¶¶ 80-81; D. 644 ¶¶ 80-81. Hiring poor image employees could result in a Managing Partner receiving a 30-day notice, a precursor to termination. D. 617 ¶ 84; D. 644 ¶ 84.
While the parties dispute the extent of top-down control over hiring by Texas Roadhouse, both sides agree that there is some deviation in practice as to hiring practices at the local level. For instance, Managing Partners at the local restaurant level do not consider applications for uniform periods of time such that applications may be considered anywhere between five days to a year depending on the particular Managing Partner. D. 589 ¶ 67; D. 616 ¶ 67. In addition, different Managing Partners may use different criteria to hire a server as opposed to a host, or a server assistant or a bartender. D. 589 ¶ 69; D. 616 ¶ 69.
D. The Experts’ Statistical Analysis
Over the course of 2007 to 2014, there were 181,583 FOH hires across the relevant Texas Roadhouse locations, 1.62% of which were FOH hires of those individuals over age 40 (the PAG hires). D. 617 ¶ 144; D. 644 ¶ 144. For each of the four FOH positions, the percent of PAG hires were: 1.49% for servers; 1.85% for server assistants; 0.28% for hosts; and 3.08% for bartenders. Id. For most of this time period, Texas Roadhouse predominantly accepted paper applications; halfway through 2013, Texas Roadhouse predominantly used electronic applications it received through the website Snagajob. D. 617 ¶ 147; D. 644 ¶ 147.
The parties each offer statistical analysis from their own experts and dispute the reliability and admissibility of the other experts (as discussed below when addressing their respective Daubert motions). The EEOC’s expert, Dr. David Crawford (“Crawford”) applied statistical tests to each position for each store-year for which he had sufficient data against three different benchmarks, census data, paper applications and electronic applications. D. 617 ¶ 150; D. 644 ¶ 150. As a result, he concluded that if hiring had been age-neutral, the probability of the aforementioned PAG hire rate results for each position would be equivalent of more than seven standard deviations (i.e. a likelihood of more than 1 in 781,000,000,000). D. 617 ¶ 150; D. 644 ¶ 150. Even after Crawford revised his calculations, his results were still statistically significant at greater than seven standard deviations. D. 617 ¶ 151; D. 644 ¶ 151. Using the census benchmark data, Crawford also calculated shortfalls in the number of PAG individuals who would have been hired if there had been age-neutral hiring. D. 617 ¶ 154; D. 644 ¶ 154. As a result, he found that the total shortfalls for servers, server assistants, hosts and bartenders were 14,604; 10,477; 4,685; and 260, respectively. D. 617 ¶ 154; D. 644 ¶ 154.
Crawford additionally concluded that when considering position-specific comparisons of hires to the paper application benchmark, there were shortfalls in PAG hiring of 95.6%, 97.4% and 90.0% of the store-years for servers, hosts and server assistants. D. 617 ¶ 159; D. 644 ¶ 159; D. 622-20 at 3-5; D. 586-4 ¶¶ 22, 23. He further found that the probability of those results are less than 1 in 781,000,000,000 for servers, 1 in 7,048,151,460 for hosts and 1 in 93 for server assistants if there were a process of age-neutral hiring. D. 617 ¶ 159; D. 644 ¶ 159; D. 586-3 ¶ 22; D. 586-4 ¶ 22. Crawford also found shortfalls when comparing the PAG hires for FOH positions when compared against electronic Snaga-job applications; he specifically found shortfalls in 83.5% of the store-years tested and stated that the probability of those results being generated by an age-neutral hiring process is less than 1 in 781,000,-000,000. D. 617 ¶ 160; D 644 ¶ 160; D. 586-2 ¶ 82(a).
Texas Roadhouse’s expert, Saad, explained some drawbacks and errors in Crawford’s analysis, including frailties in Crawford’s calculations due to store-by-store variations and his use of external census data instead of the applicant flow data otherwise available through the paper and electronic applications. See, e.g., D. 595-2 ¶¶ 9-13, 36-42, 48-49, 58-60, 66, 73-74. Saad also formulated an alternative statistical . analysis that utilizes the theory of “duration dependence” and applied that theory to the application data using a Cox proportional hazard ratio. See id. ¶¶ 84-96, 103-04, 107-08. After applying this analysis — which he said more properly accounted for distinctions between the under 40 and over 40 applicant pools — Saad found that there was no statistically significant shortfall across the entire eight-year period and that the years of 2007, 2008, 2013, and 2014 had significant surpluses in PAG hiring. Id. ¶¶ 104-118.
E. Purported Anecdotes of Discrimination and Non-Discrimination
In addition to the statistical analysis upon which it relies to show discriminatory intent, the EEOC also relies upon direct or circumstantial evidence that age bias un-dergirded a nationwide discriminatory hiring practice at Texas Roadhouse. This evidence includes images of young-looking people used in management training, the “Examiner” and other written materials distributed to managers and the cover sheet used for some applications. See, e.g., D. 616 ¶¶ 93, 95, 99, 106; D. 644 ¶ 93, 95, 99, 106. It also includes testimony from a Texas Roadhouse trainer and service manager regarding the image for which Texas Roadhouse hired and testimony from individuals over 40 who unsuccessfully applied for FOH positions and heard age-related comments made during their interviews. See, e.g., D. 617 ¶¶ 170, 175, 176; D. 644 ¶¶ 170,175,176. '
By contrast, Texas Roadhouse presents declarations from 372. Managing Partners who attest that they understood that Texas Roadhouse maintained non-discrimination policies that prohibited age discrimination and that they were not aware of any company practices to refuse to hire PAG applications for FOH positions. D. 589 ¶¶ 70, 73; D. 616 ¶¶70, 73. In addition, roughly 45 Market Partners and Regional Market Partners have filed declarations stating that they understand that Texas Roadhouse maintains non-discriminatory policies that prohibit against age discrimination in hiring and that they were not part of any practice at Texas Roadhouse to discourage or reject PAG applicants from FOH positions. D. 589 ¶ 75; D. 616 ¶75. Finally, 234 workers age 40 and over who were hired for FOH positions at Texas Roadhouse declare that they have never witnessed or experienced any age discrimination. D. 589 ¶¶ 140, 142; D. 616 ¶¶ 140, 142.
IV. Procedural History
In March 2009, the EEOC initiated an agency charge of age discrimination against Texas Roadhouse. D. 589 ¶ 108; D. 616 ¶ 108. The EEOC then instituted this action on September 30, 2011. D. 1. On August 2, 2016, Texas Roadhouse moved for summary judgment. D. 587. Texas Roadhouse also moved to strike the reports and testimony of the EEOC’s expert, Crawford. D. 584. The EEOC filed a motion to strike portions of the expert report and proffered testimony of Texas Roadhouse’s expert, Saad, D. 593, and also moved to strike the expert report and proposed testimony of Texas Roadhouse’s other expert, Dr. Eric Dunleavy (“Dun-leavy”), D. 600. The Court heard the parties on the pending motions on October 5, 2016 and took these matters under advisement. D. 648.
V. Motions to Strike Expert Testimony
A. Texas Roadhouse’s Motion to Strike Crawford’s Proffered Opinion
1. Use of Census Data Benchmarks
To determine whether the Texas Roadhouse hiring decisions were age-neutral, Crawford, an economist engaged by the EEOC, analyzed the rates of actual PAG hires at Texas Roadhouse, D. 586-2 ¶¶ 14-17, 20, against three different sets of data: (1) census data for occupation codes similar but not exactly matched to those of FOH positions at Texas Roadhouse, id ¶¶ 14, 29; D. 586-9; (2) the Snagajob electronic applicant pool used by Texas Roadhouse, D. 586-2 ¶¶ 14, 32; and (3) a random sample of the unsuccessful paper applications produced by Texas Roadhouse, id. ¶¶ 14, 43, 49. In so doing, Crawford analyzed the comparisons across the three different benchmarks to assess the likelihood that age-neutral hiring would have resulted in the number of actual PAG hires given the pool. Id. ¶¶ 56-60. In his analysis, he presents his findings for each of the individual FOH positions at issue as well as aggregated findings based upon all FOH positions combined. Id. ¶ 60.
Texas Roadhouse now moves to strike Crawford’s opinions and proffered testimony. D. 584. Texas Roadhouse initially argues that this opinion should be struck because Crawford’s use of census data as a proxy is not reliable under Fed. R. Evid. 702 given the breadth and variety of service jobs reflected in census data that exceeds that of the four FOH categories. D. 585 at 9-16. Texas Roadhouse first maintains that the census data is unreliable because it contains an overbroad range of food service workers at varying types of establishments that is not equivalent to FOH positions at Texas Roadhouse. D. 585 at 12-14. Here, the census data upon which Crawford relies is not general census data but instead age distribution data from the EEO Tabulation of census data for particular occupations in particular geographic areas. D. 586-2 ¶ 28. Namely, Crawford relied upon age distribution data for the job categories of (1) waiters and waitresses; (2) other food preparation and serving related workers; (3) hosts and hostesses; and (4) bartender categories within the EEO Tabulation information which he further broke down by geographic region corresponding to each Texas Roadhouse location. Id. ¶¶ 28-31. Crawford compared data on actual PAG hires at Texas Roadhouse against this census data that includes those coded as waiters and waitresses, other food preparation and serving related workers, hosts and hostesses, and bartenders. Id. These categories include a variety of jobs outside of the FOH positions and from restaurant positions including cafeteria workers, maitre d’s, and lunchroom aides as well as establishments dissimilar to Texas Roadhouse like luxury restaurants and hospital cafeterias. D. 586-9; D. 586-1 at 49-54.
This contention, however, does not warrant exclusion of Crawford’s opinion. Even when statistical analysis has involved general population census data to show discriminatory intent, it has not been precluded on Fed. R. Evid. 702 grounds. See E.E.O.C. v. FAPS, Inc., No. 10-cv-3095-JAP DEA, 2014 WL 4798802, at *5-*6 (D.N.J. Sept. 26, 2014) (admitting expert who relied upon local labor market data to examine race discrimination in hiring over Daubert challenge to the reliability of his conclusions); see also Pina v. City of E. Providence, 492 F.Supp. 1240, 1245-46 (D.R.I. 1980) (concluding that plaintiffs established a prima facie case after presenting statistical evidence that compared “the percentage of ranked minorities and the percentage of minorities in the general population” because “the skill involved ... is one that the general population may possess”). Here, the subset of census data was more particularized than general population census information, i.e. figures of persons worked in similar food service positions and were closer, if not perfectly aligned, with FOH positions. Although not perfect, reliance upon this census data is a reliable proxy where, as both parties acknowledge, the actual Texas Roadhouse application data for years 2007 to 2013 is not complete. D. 586-2 ¶¶ 33, 44; D. 617 ¶¶ 188, 190, 193; D. 644 ¶¶ 188, 190, 193; see FAPS, Inc., 2014 WL 4798802, at *5-*6 (admitting the opinion of statistical expert who relied upon local labor market census data because that expert did not find the applicant flow data to be reliable or complete). Finally, failing to use a perfect set of variables that incorporates all relevant factors or excludes all potentially irrelevant variables is not a means for rejecting an expert’s analysis. See Flebotte v. Dow Jones & Co., No. 97-cv-30117-FHF, 2000 WL 35539238, at *4 (D. Mass. Dec. 6, 2000); McMillan v. Massachusetts Soc. for Prevention of Cruelty to Animals, 140 F.3d 288, 302 (1st Cir. 1998). This is because statistical “analyses are admissible even in disparate treatment cases unless they are so incomplete as to be inadmissible as irrelevant.” McMillan, 140 F.3d at 303 (internal quotation marks and citations omitted). That is not the case here. Any concerns raised by overbroad census data go to weight, not admissibility. Currier v. United Techs. Corp., 393 F.3d 246, 250-252 (1st Cir. 2004).
This reasoning applies with equal force to Texas Roadhouse’s argument that Crawford’s opinion should be excluded because the census data does not reliably estimate the actual PAG applicant pool for any given year. D. 585 at 10. Here, Texas Roadhouse emphasizes that for the year 2014, the calculations of PAG applicants based upon the complete Snagajob data differs from the PAG proportions resulting from the census data. D. 585 at 10-11; see D. 586-1 at 56-59. It is undisputed that for 2014, the one year for which there is complete applicant pool data, D. 589 ¶ 101; D. 616 ¶ 101, the estimated applicant figures of 17.7% or 18.5% based upon the census data is substantially higher than the 3.7% representing the applicant pool, a figure based upon the actual Snagajob data, D. 586-1 at 58-59; D. 586-6 at 120-121. That is, the census data calculation may overstate the actual shortfall percentage. D. 585 at 11; D. 586-1 at 58-59. Even when considered with Texas Roadhouse’s other critiques, this argument also does not warrant exclusion of Crawford’s opinion. First, the 2014 data may not be indicative of the applicant pools for the rest of the liability period. See E.E.O.C. v. Am. Nat. Bank, 652 F.2d 1176, 1195-97 (4th Cir. 1981) (explaining that “[ajpplicant flow data limited to one out of seven relevant years cannot be held to rebut a prima facie case based upon gross disparities”). Second, complete applicant pool data is not available for all of 2007 to 2014 and the use of a proxy, even an imperfect proxy, is not grounds for exclusion. McMillan, 140 F.3d at 302; Flebotte, 2000 WL 35539238, at *4. Texas Roadhouse again is not without recourse in challenging Crawford’s results based upon census data: it can raise them on cross-examination, in its final argument to the jury and through the testimony of its own statistical expert.
That it is unclear whether the census data Crawford relied upon includes people new to the job market, D. 585 at 14-15, also goes to the weight, not the admissibility, of his opinion. Here, Texas Roadhouse and the EEOC dispute whether Crawford’s use of census data omits a set of younger applicants that could alter his results. D. 585 at 14-15; D. 613 at 21; D. 595-2 ¶¶ 66, 73-74 (Saad addressing this issue); D. 586-4 ¶¶ 10(p), 71 (Crawford responding to Saad’s critique); Freeman v. Package Mach. Co., 865 F.2d 1331, 1340 (1st Cir. 1988) (concluding that the drawbacks in the expert’s statistics on forced terminations that may have included data of those who were terminated voluntarily was fodder for cross-examination but not for exclusion from the jury).
Texas Roadhouse contends that Crawford’s inability to link initially twenty-eight stores with a unique core-based statistical area renders the entirety of the census data comparison as incomplete. D. 585 at 16. In his analysis, Crawford culled the applicable census data by (1) matching the zip codes of each relevant Texas Roadhouse store with a core-based statistical area that was defined the Census Bureau; (2) using the street address for stores cut across multiple core-based statistical areas to assign the store to one core-based statistical area; (3) applying county-based data for each store where core-based statistical data was unavailable; and (4) ultimately using this collected data as benchmark comparisons for the percentage of PAG hiring by Texas Roadhouse for the FOH positions. D. 586-2 ¶¶ 29-30. This provided comparison data for roughly 370 of the 396 stores at issue. Id. ¶ 30; D. 586-4 ¶ 72 (admitting that one minor problem with the available CBSA level data was that 28 stores did not have unique CBSA level data available).
Using alternative proxy data when the primary evidence is not complete — such as the potentially missing paper applications or the piecemeal use of the Snagajob electronic application program — is not grounds for inadmissibility. See Brown v. Nucor Corp., 785 F.3d 895, 903-904 (4th Cir. 2015) (collecting cases). Indeed, “an incremental reduction in probative value — which is a natural consequence of the use of proxy data — does not itself render a statistical study unreliable in establishing a question of discrimination” because to hold otherwise would render “plaintiffs unable to bring a statistics-based employment discrimination claim after a company has intentionally or inadvertently destroyed actual applicant data.” Id. at 906. This is particularly so here where Crawford applied a specified methodology — using street addresses to identify a CBSA for stores that otherwise did not fall squarely within one — to attempt to round out his analysis and find a proxy for the imprecision of comparing Texas Roadhouse stores to the census data. D. 586-2 ¶¶ 29-30.
Texas Roadhouse also contends that Crawford’s opinion should be struck because there is no reason for relying upon 2014 census data because the actual applicant flow data for that year is complete. D. 585.at 15. Even if the paper and electronic data for 2014 were complete and representative, Crawford’s report has separate sections for his results based upon census data comparisons and based upon Snaga-job data comparisons such that his report drew distinctions between any disparate results. This means that if his analysis is not “watertight” and “omitted ... important variables[ ] or was deficient in other respects” Texas Roadhouse can “exploit and discredit the analysis during cross examination.” McMillan, 140 F.3d at 302-03; see Currier, 393 F.3d at 252 (1st Cir. 2004).
Such opinion evidence also does not run afoul of Fed. R. Evid. 403. Here, Texas Roadhouse asserts that the sum of the potential weaknesses in Crawford’s census data will result in undue prejudice against Texas Roadhouse. D. 585 at 16. The Court disagrees. While Texas Roadhouse has highlighted possible flaws in Crawford’s analysis, none rise to the level of causing unfair prejudice or creating inflammatory effect with the jury, particularly where the probative value — statistical support for the EEOC’s allegations of discriminatory animus on the part of Texas Roadhouse — is clear.
2. Statistics as Measure of Store-Level Discrimination
Contrary to Texas Roadhouse’s argument, D. 585 at 17-20, the statistics that Crawford has compiled and analyzed for nationwide Texas Roadhouse statistics are probative. The EEOC alleges that Texas Roadhouse engaged in a pattern and practice of age discrimination that reigned down from its headquarters. Given the nature of the claims, aggregating data for a nationwide view is not improper or unduly prejudicial. Int’l Bhd. of Teamsters v. United States, 431 U.S. 324, 339, 97 S.Ct. 1843, 52 L.Ed.2d 396 (1977); Stagi v. Nat’l R.R. Passenger Corp., 391 Fed.Appx. 133, 145, 148 (3d Cir. 2010) (collecting cases); see Reynolds v. Barrett, 685 F.3d 193, 203 (2d Cir. 2012). Unlike Wal-Mart Stores v. Dukes, 564 U.S. 338, 356-57, 131 S.Ct. 2541, 180 L.Ed.2d 374 (2011), where the Court had to address whether the proposed class members, who were employed at different stores under different managers, had shown commonality of their claims for the purposes of class certification under Fed. R. Civ. P. 23, the very nature of the claim compels a global and unified view of the company’s nationwide pattern and practice.
Moreover, it is imprecise to say that Crawford’s analysis did not address store to store data. Crawford did not simply average the results across the restaurants nationwide. Indeed, as he describes in his report, Crawford “calculate[d] PAG% for each store-year, compute[d] the standardized difference between each store-year PAG% and the corresponding store-year benchmark, and statistically test[ed] whether a set of standardized differences was consistent with age-neutral hiring” by Texas Roadhouse. D. 586-4 ¶ 15. Furthermore, Defendant’s rebuttal expert, Saad, acknowledged this fact, in part, when he said Crawford “uses several benchmarks for the expected rate of hiring, and compares, store by store and by job the actual to the expected numbers of hires based upon those benchmarks, and then presents aggregated findings .... ” D. 595-2 ¶ 4. The underlying store-year data comparisons are provided in Crawford’s charts on a store-by-store and year-by-year comparison, D. 622-18; D. 622-19; should Texas Roadhouse take issue with whether those more granular calculations discredit the EEOC’s theory of disparate treatment, it may do so on cross-examination. Such evidence is at least probative of EEOC’s claim and would also be helpful to the jury as it is asked to determine if the EEOC proved a pattern or practice.
The admission of this evidence is not unduly prejudicial given its probative value and that any “perceived flaws in [Crawford’s] reasoning or calculations may be challenged through the normal adversary process.” Haemonetics Corp. v. Baxter Healthcare Corp., 593 F.Supp.2d 303, 306 (D. Mass. 2009). For this reason, the Court also declines to exclude Crawford’s census data under Fed. R. Evid. 403.
That Crawford’s opinions may not prove all of the EEOC’s case is not a basis for excluding it. Adams v. Ameritech Servs. Inc., 231 F.3d 414, 427-28 (7th Cir. 2000) (“ruling out chance was an important step in the plaintiffs’ proof, even if it was not a single leap from the starting line to the finish line”); see Palmer v. Shultz, 815 F.2d 84, 91 (D.C. Cir. 1987); FAPS, Inc., 2014 WL 4798802, at *14. Consideration of Watson v. Fort Worth Bank & Tr., 487 U.S. 977, 994, 108 S.Ct. 2777, 101 L.Ed.2d 827 (1988) (noting that “the plaintiffs burden in establishing a prima facie case goes beyond the need to show that there are statistical disparities in the employer’s work force”) does not compel another outcome since the EEOC does not rely solely upon Crawford’s opinion or his underlying statistical analysis alone for its claim.
3. The Effect of the 2010 Determination Letter and the 2011 Litigation
Texas Roadhouse’s challenges to certain of Crawford’s opinions as speculative, D. 585 at 21-22, however, stand on different ground. Crawford opines that for several sets of data that “the results ... are consistent with the hypothesis that [Texas Roadhouse] increased its effort to hire older workers after the EEOC’s Letter of Determination in 2010. The results ... are also consistent with the hypothesis that [Texas Roadhouse] increased its effort to hire older workers after the EEOC’s filing of this case, which was in 2011.” D. 586-2 ¶ 62; see also, e.g., id. ¶¶ 62, 83, 96.
Under Fed. R. Evid. 702, expert testimony must be more than unsupported speculation or abstract beliefs. Haemonetics Corp., 593 F.Supp.2d at 305 (quoting Daubert, 509 U.S. at 590, 113 S.Ct. 2786). Indeed, a court may exclude an expert’s opinion when it is based upon conjecture or speculation deriving from an insufficient evidentiary source. United States v. Organon USA Inc., No. 07-cv-12153-RWZ, 2015 WL 10002943, at *3-4 (D. Mass. Aug. 17, 2015).
If the evidence at trial shows that, as a factual matter, increases in 40+ hiring improved after the EEOC’s 2010 letter and/or the initiation of this lawsuit in 2011, the EEOC could certainly argue to the jury that there is some correlation between the two. For Crawford, however, to opine as a matter of his expertise that there is a causal link between the two, there must be a reliable factual basis for the same. It is unclear that he has provided such a basis. For example, Crawford did not know who at Texas Roadhouse was aware of the 2010 letter or whether there was evidence that the Managing Partners — those employees generally in charge of hiring — knew of the 2010 letter. D. 586-1 at 14. Moreover, Crawford admitted that he made no professional conclusion as to whether the hypothesis that Texas Roadhouse increased its efforts to hire FOH workers in the protected age group after the 2010 letter of determination was true. Id. at 15. That is, Crawford explained in his deposition that he “ha[s] not proven and do[es]n’t claim to have proven that the changes before and after the letter were caused by the letter.” Id. In addition, Crawford explained that his “statistical observation is that there was a difference in the rates before and after the filing of the complaint,” not that the filing of the complaint necessarily caused this increase. Id. at 13-14. For these reasons, the Court strikes these proffered opinions.
4. The “Chilling Effect” Opinion
Lastly, Texas Roadhouse moves to exclude Crawford’s opinion, offered in rebuttal to Saad’s report, that his results are “consistent with post-opening chilling of applications from people who were 40 or older.” D. 586-5 ¶ 61; D. 585 at 23. First, Texas Roadhouse argues this disclosure was untimely under Fed. R. Civ. P. 26(a)(2)(B)(i)-(iii). D. 585 at 23-24. Pursuant to Fed. R. Civ. P. 26(a)(2)(D)(ii), rebuttal disclosures made by experts must be “intended solely to contradict or rebut evidence on the same subject matter identified by another party” and must be disclosed within 30 days after the other party’s disclosure. Fed. R. Civ. P. 26(a)(2)(D)(ii); see In re High-Tech Employee Antitrust Litig., No. 11-cv-02509-LHK, 2014 WL 1351040, at *3 (N.D. Cal. Apr. 4, 2014). Pursuant to Rule 37, a party that fails to adhere to the Rule 26(a) requirements “is not allowed to use that [untimely] information ... to supply evidence on a motion, at a hearing, or at a trial, unless the failure was substantially justified or is harmless.” Fed. R. Civ. P. 37(c)(1).
Here, Texas Roadhouse argues that Crawford’s analysis of the “chilling effect” was untimely because it was not disclosed as an opinion in his primary expert report and did not contradict or rebut anything raised by its own expert, Saad. D. 585 at 23-24. Indisputably, Crawford’s opening expert reports include no analysis or opinion on “chilling effect.” See D. 586-2. The Court, however, does not agree that the introduction of this opinion was not appropriately disclosed in Crawford’s rebuttal to Saad’s report. Saad, in relevant part, criticized Crawford for using census data as a benchmark where the paper application benchmark was sufficiently representative of the percentage of PAG applicants. D. 595-2 ¶ 52. Saad does so by analyzing the paper applications by store-month instead of store-year and includes a variable into his calculations that takes into account the fact that the periods leading up to store openings — as opposed to post-opening periods — tend to produce a substantially larger number of applicants. Iff ¶ 54. To rebut this argument, Crawford noted that the paper applications were' incomplete and unrepresentative of the older applicants and points out several flaws in Saad’s calculations. D. 586-5 ¶¶ 54-60. In doing so, he further opined that the statistical results of pre-opening and post-opening PAG applications were “consistent with post-opening chilling of applications.” Id. ¶ 61. This contention is responsive to Saad’s opinion that the paper application benchmark was a more complete and accurate measure of the number of 40 + applicants. The “chilling effect” opinion that Crawford offers in rebuttal is responsive to this critique. Accordingly, its disclosure was not untimely under Fed. R. Civ. P. 26 and the Court will not preclude its admission on that basis.
Texas Roadhouse additionally seeks to exclude this opinion on the grounds that Crawford’s testimony as to the “chilling” effect will not assist the trier of fact because it is speculative. D. 585 at 21-23. Similar to his proffered testimony on the 2010 determination letter and the 2011 litigation, Crawford provides no opinion or basis as to the causal effect of post-opening “chilling.” In his deposition, Crawford explains that he has “no empirical evidence that demonstrates a causal relationship” and makes no professional opinion as to the causal relationship. D. 586-1 at 16-17. Instead, he posits only that the numbers are consistent with the hypothesis of “chilling” as the cause and recognizes that a decline in applications from those 40 and older could have resulted from other external economic factors. Id. at 16-17, 20. For these reasons, the Court strikes Crawford’s proffered opinion as to the “chilling” effect.
B. EEOC’s Motion to Strike Portions of Expert Report and Testimony of Saad
1. Use of PUMS Data
Instead of using EEO tabulation census data as Crawford does, Texas Roadhouse’s expert, Saad, relies upon American Community Survey Public Use Microdata Sample (“PUMS”) data as his external labor market benchmark. D. 595-2 ¶ 68. Specifically, Saad used zip codes to determine the Public Use Microdata Sample Area (“PUMA”) for each store location, calculated availability rates for each restaurant based upon the PUMAs and compared that against the representation of PAG individuals in FOH positions in the same Texas Roadhouse location. Id. ¶¶ 68-69.
The EEOC now moves to strike Saad’s use of the PUMS data and all resulting opinions on the basis that the PUMS data is unreliable. D. 594 at 17-19. The Court disagrees. In his reply report, Crawford pointed out the flaws in Saad’s original PUMS analysis and calculated the corrected outcomes that derive from accounting for non-unique PUMAs, D. 586-4 ¶¶ 65-67, and Saad admitted this error, D. 623-8 at 2-5. To the extent that there were errors before or perhaps remaining drawbacks to Saad’s analysis, use of an imperfect proxy, as it was for Crawford’s analysis, is not grounds for exclusion. Flebotte, 2000 WL 35539238, at *4; McMillan, 140 F.3d at 302. Any perceived shortcomings associated with using the PUMS data goes to the weight, not the admissibility, of the evidence. Freeman v. Package Mach. Co., 865 F.2d 1331, 1338 (1st Cir. 1988). In addition, the parties’ differing opinions as to which party the corrected PUMS data supports, D. 594 at 16; D. 621 at 8-10, can again be addressed in the course of direct and cross-examinations of both Saad and Crawford and, ultimately, will be resolved by the jury. See Milward, 639 F.3d at 15. The Court denies the EEOC’s motion on this ground.
2. Statistically Significant Thresholds
EEOC also objects to Saad’s opinion on statistically significant thresholds for variation in store-level hiring shortfalls because it amounts to testimony about the legal standard, i.e., whether EEOC’s conduct amounts to a pattern or practice of age discrimination. D. 594 at 19-21. While it is true that Fed. R. Evid. 704 bars “opinions which would merely tell the jury what result to reach” and “opinions phrased in terms of inadequately explored legal criteria,” Chow v. Zimny, No. 10-cv-10572-GAO, 2014 WL 4964408, at *1 (D. Mass. Sept. 30, 2014) (internal quotation marks and citation omitted) and Fed. R. Evid. 702 provides that an expert “may not ‘assist’ the jury by expounding upon the law ... because to do so would intrude upon the province of the trial judge,” Ji v. Bose Corp., 538 F.Supp.2d 354, 357-58 (D. Mass. 2008) (citing Nieves-Villanueva v. Soto-Rivera, 133 F.3d 92, 100 (1st Cir. 1997)), an expert may still offer an opinion that bears upon the factual determination that the jury will have to make about whether the EEOC has shown that Texas Roadhouse engaged in a pattern or practice of discrimination. Saad’s opinion is that “from a statistical perspective” he does not believe the numbers support a pattern or practice of age discrimination. D. 595-2 ¶ 8; see also id. ¶¶ 5, 18, 24-26, 34, 46, 120, 142. In the context of proper (and, perhaps, contemporaneous) jury instructions to the jurors that they get to decide what weight, if any, they give to any testimony, including opinions offered by the parties’ competing experts and that the issue of whether the EEOC has met its burden of proof is for the jury to decide, Saad’s proffered opinion falls within the bounds of Fed R. Evid. 702 and 704.
The Court also will not strike Saad’s opinion regarding the variation in store-by-store data on the grounds, as the EEOC contends, D. 594 at 20, that it is unreliable because Saad admitted he had no basis to opine on how much variation in PAG hiring is too much to prevent a finding of pattern-or-practice discrimination. In his expert report, Saad opined that “[i]f one examines Dr. Crawford’s backup materials, one finds wide variation in the nature of the hiring patterns across the hundreds of [Texas Roadhouse] stores” and that if Crawford had “examined his own store level statistical findings, he would have seen wide variations in hiring outcomes between stores” which “are inconsistent with the statistical notion of a common ‘pattern or practice.’ ” D. 595-2 ¶ 8. That there were variations in store to store shortfalls and Saad testified that he did not know how much store to store variation was too much to preclude a legal finding of pattern or practice discrimination, D. 595-3 at 36, does not undermine the reliability of his opinion where he explained that the variations present in this data “are inconsistent from a statistical perspective with the hypothesis that the standard operating procedure across [Texas Roadhouse] store locations is to treat applicants aged 40 and above (40 + ) adversely ...D. 595-2 ¶ 7.
Finally, the EEOC maintains that Saad’s testimony that the shortfalls identified by Crawford are insufficient statistical thresholds for establishing pattern or practice violations is unreliable because it is based upon speculation and Saad’s subjective belief. D. 594 at 21; see, e.g., D. 595-2 ¶¶ 19, 22, 25-26. This is not a sufficient basis for excluding this portion of Saad’s testimony. As an example, in his expert report, Saad writes “[o]ver all stores, jobs and years analyzed by Dr. Crawford a statistically significant shortfall at two standard deviations is found only 26% of the time. A statistically insignificant shortfall is found 74% of the time.” D. 595-2 ¶ 19. In his deposition, when asked when he would exclude the word “only,” Saad testified that was not an expert driven issue. D. 595-3 at 39. Whether Saad, as an economist, has explained the significance of the shortfall to his ultimate conclusion about nationwide pattern or practice and has done so upon the basis of statistical analysis and identifiable data, the jury may choose to accept or reject that conclusion after hearing his testimony upon both direct and cross examination.
3. Duration Dependence Theory and the Cox Proportional Hazards Model
Finally, the EEOC moves to strike Saad’s analysis stemming from “duration dependence.” D. 594 at 6-17. Saad opined that Crawford’s age discrimination analysis was flawed because he assumed all applicants were similarly situated — i.e., same productivity, same motivation, etc.— and that FOH applicants to Texas Roadhouse under 40 and over 40 are equal substitutes. D. 595-2 ¶¶ 84-89. Saad posits that for Crawford’s statistical analysis to be accurate, it must incorporate duration dependence, a theory that accounts for changes in average applicant pools given the changes that workers undergo as they progress through their careers. Id. ¶¶ 86-96. In Saad’s estimation, Crawford did not account for the human capital that some older workers acquire over time and thus are no longer interested in, or available for, entry-level positions; on the other hand, a pool of younger workers will .include both those with high potential who may later develop skills to rise in rank in the job market and lower performing young workers, all of whom are applying for entry-level positions. Id. ¶ 87. Thus, Saad opines that those in the PAG applicant pool applying for FOH positions at Texas Roadhouse may not be equal substitutes for those in the younger applicant pool in terms of potential and that this must be accounted for in the statistical analysis. Id. ¶¶ 86-96. To apply this theory, Saad examined data from the National Longitudinal Study of Youth (“NLS”) for both those in the study since 1979 and those in the study since 1997, id. ¶¶ 100, 100 n. 71, 101-03, and incorporated the unemployment spell data derived from those studies into a Cox proportional hazards statistical model. Id. ¶¶ 103-04. Saad then took the hazard ratio — representing the likelihood that an unemployed PAG individual is to become employed at a new job at any given time as an individual under 40 — and used it to modify Crawford’s analysis to account for the concept of duration dependence. Id. ¶¶ 104-07, 114, 118.
The EEOC challenges duration dependence and its application via the Cox proportional hazards model as “junk science.” D. 594 at 7; see D. 586-4 ¶ 10(t)-(v). The EEOC argues that Saad’s use of the duration dependence theory is inadmissible because it has never before been applied in the hiring context and Saad fails to cite peer-reviewed publications to support his application here. Id. at 9. The duration dependence theory, however, is not new. See, e.g., Barnes v. The Hershey Co., No. 12-cv-01334-CRB, 2016 WL 192310, at *9-10 (N.D. Cal. Jan. 15, 2016) (considering the duration dependence statistical theory at summary judgment in an age-based employee termination case); see also Paxton v. Lanvin-Charles of the Ritz, Inc., No. 77-cv-28-LBS, 1978 WL 13903, at *4 (S.D.N.Y. July 24, 1978), aff. sub nom. Paxton v. Lanvin-Charles of the Ritz, Inc., 594 F.2d 852 (2d Cir. 1978) (explaining that there is a “natural tendency of the newly hired employee to be young” because older workers move out of the workforce' as younger ones constantly move in). Although application of this theory to the hiring context may be new, D. 594 at 10, D. 586-4 ¶ 10(t), neither duration dependence as a theory nor the Cox proportional hazard model as a method is novel. See Barnes, 2016 WL 192310, at *9-10 (analyzing duration dependence theory in age-based discriminatory termination case); D. 623-9; D. 623-10. Even if applying duration dependence via the proportional hazard model were novel here and applied for the purpose of this litigation, this remains a single factor in the Fed. R. Evid. 702 analysis where Saad has otherwise met that rule’s criteria. See Granfield v. CSX Transp., Inc., 597 F.3d 474, 486 (1st Cir. 2010) (explaining that when assessing an expert’s admissibility, one factor is whether the testimony was created for the purpose of the litigation).
While the Frye standard of general acceptability is no longer the touchstone of admissibility of expert opinion under Fed. R. Evid. 702 post-Daubert, whether a methodology has been peer reviewed remains one factor for the Court to consider when addressing challenges to the admissibility of expert testimony. See Milward, 639 F.3d at 14, 22. To support exclusion, the EEOC relies upon Wyche v. Marine Midland Bank, No. 94-cv-4022-DC, 1997 WL 109564, at *1 (S.D.N.Y. Mar. 11; 1997) to argue that the Cox proportional hazard model is not generally accepted in the scientific community. That case, however, does not warrant exclusion of the opinion here. First, the Wyche court focused on whether “jurors may find the model difficult to understand” and thus that the expert opinion should be excluded on the grounds that it would not assist the trier of fact. Id. at *1. No such concern exists here and any complexities of applying the model to age discrimination cases can be explained during examination and is otherwise adequately explained in Saad’s expert report. To the extent that the Wyche court also excluded the proportional hazards model on the basis that the proponent “ha[d] not demonstrated that the [model] has generally been accepted in the scientific community,” id. at *1, the same cannot be said here. As explained, the record indicates that the general use of the Cox proportional hazards model is not novel and has been the subject of publication and peer comment. See, e.g., D. 623-9 (investigating time-dependent effects in a proportional hazards regression in a 1990 journal article and explaining generally that the model “has proved an exceedingly useful method of analyzing survival or failure data”); D. 623-10 (providing a 1972 journal article on Cox’s regression models and peer discussion of the same). Moreover, Texas Roadhouse has demonstrated that at least some commentators posit that the proportional hazards model is appropriate to apply in age discrimination cases. See Michael 0. Finkelstein & Bruce Levin, Proportional Hazard Models for Age Discrimination Cases, 34 Jurimetrics J. 153, 157 (1994) (explaining that variations of the proportional hazards models has been used in other notable cases and “are also appropriate in the context of age discrimination”). This demonstrates at least some acceptance of the use of this model in age discrimination cases and some level of reliability even if, as Wyche explains, there may not be overall general acceptance in the community. Wyche, 1997 WL 109564, at *1. For these reasons, Wyche does not compel exclusion of Saad’s application of duration dependence via the proportional hazards model in this case.
The EEOC next asserts that even if Saad’s methodology was sound, he did not apply it in a reliable manner because he used NLS study data to determine what the hazard ratio calculation should be and that data is improperly applied here. D. 594 at 12. Specifically, the EEOC contends that Saad’s reliance on the NLS data to compute the hazard ratio calculation is flawed because the NLS longitudinal data is based upon study participants who were between the ages of 21 and 33 and 41 and 55 for the 2006-2013 time period and thus did not serve as a proper calculation for all applicants under'40 and all applicants over 40 as it missed those under 21, between 33 and 41, and those over 55. Id. Again, this alone — the use of an imperfect proxy to estimate the expected productivity of under 40 and over 40 workers — does not make Saad’s analysis inadmissible but instead goes to the weight of his findings. Flebotte, 2000 WL 35539238, at *4; McMillan, 140 F.3d at 302; Freeman, 865 F.2d at 1338. The EEOC also argues that this manner of calculating the hazard ratio is ill-fitting here because Saad calculates the hazard ratio based upon the lengths of unemployment experienced by older workers in the NLS study without assessing whether Texas Roadhouse applicants are employed or unemployed when they apply for a FOH position. D. 594 at 13. This is not unlike the EEOC’s own expert’s calculations based upon census data which considered the age of individuals already employed in FOH corollary positions and not those applying for such positions.
The EEOC’s remaining contentions— that Saad treats the lower odds of an applicant in the PAG being hired as an inevitable consequence or the fact that he concentrates on the single variable of low productivity, id. at 13-14 — do not render his analyses as inadmissible but instead are “cloaked as objections ... under Rule 702, [but] are actually objections about the weight of the evidence.” Granfield v. CSX Transp., Inc., 597 F.3d 474, 487 (1st Cir. 2010). This is also true of the EEOC’s contentions that Saad is inconsistent in arguing that those who are over 40 when hired earn higher wages, stay with Texas Roadhouse longer and are more likely to be promoted than their younger counterparts and then simultaneously applying the duration dependence theory on the basis that those in the PAG who seek entry-level jobs are less likely to be productive and successful workers. D. 594 at 16. Any such limitations of his analysis are concerns to be raised on cross-examination and are a matter for the jury to consider and weigh. Milward, 639 F.3d at 22 (explaining that “[w]hen the factual underpinning of an expert’s opinion is weak, [that] is a matter affecting the weight and credibility” of that expert’s opinion) (quoting United States v. Vargas, 471 F.3d 255, 264 (1st Cir. 2006)).
For the foregoing reasons, the Court denies the EEOC’s motion to strike portions of Saad’s proffered expert opinion.
C. The EEOC’s Motion to Strike Report and Testimony of Dunleavy
Texas Roadhouse retained Dunleavy to answer two separate but related questions: whether the Texas Roadhouse interviewing and hiring materials reflected the characteristics of behavioral interview questions found in industrial and organizational psychology literature and whether Texas Roadhouse’s “legendary traits” reflected the worker characteristics of jobs akin to the Texas Roadhouse FOH positions. D. 602-1 at 6; D. 602-2 at 9. The EEOC now moves to strike Dunleavy’s expert opinion on the grounds that (1) his analysis of a limited universe of hiring materials at Texas Roadhouse are not relevant to this case; (2) his report and testimony are unreliable because Dunleavy did not conduct a job analysis and his analysis is otherwise not based upon scientific principles and methods; and (3) his opinion will not assist the trier of fact in understanding the evidence or determining any disputed issue. D. 600.
Here, the EEOC first fails to demonstrate how Dunleavy’s report and testimony is not relevant to the material issues. The EEOC correctly points out that Dunleavy did not express an opinion as to whether age was a factor in Texas Roadhouse’s hiring, D. 602-2 at 8, and that Dunleavy focused only on one set of hiring materials produced by Texas Roadhouse that were made available for discretionary use in the hiring of FOH jobs, D. 602-1 at 3-4; D. 602-2 at 6-7,' 10. Moreover, the EEOC properly notes that there is no basis to conclude that the interview materials Dunleavy reviewed were actually used in store-level hiring decisions because Managing Partners used these interview questions at their discretion and could ask other questions by choice. D. 602-2 at 18-21; 60-61.
The EEOC, however, bases its claims against Texas Roadhouse in part on the fact that “Texas Roadhouse exercises centralized control over a detailed training regime” in which “managers are trained on how and whom to hire” and where “hiring for [Texas Roadhouse’s] image means hiring young.” D. 615 at 9. That is, centralized control is part of what it wants the jury to believe led to Texas Roadhouse’s nationwide pattern and practice of age discrimination. Therefore, the hiring materials that Texas Roadhouse distributed, either in whole or in part, bears upon the nationwide message it was sending to its