El v. Southeastern Pennsylvania Transportation Authority ("SEPTA")

AMBRO, Circuit Judge. This appeal arises out of a Title VII action alleging employment discrimination based on race. Plaintiff Douglas El claims that the Southeastern Pennsylvania Transportation Authority (“SEPTA”) unnecessarily disqualifies applicants because of prior criminal convictions — a policy that he argues has a disparate impact on minority applicants because they are more likely than white applicants to have convictions on their records. 1 The Court granted summary judgment, however, in favor of SEPTA, concluding that it had borne the burden of proving that its policy is consistent with business necessity. Though we have reservations about such a policy in the abstract, we affirm here because El did not present any evidence to rebut SEPTA’s expert testimonyI. Factual Background and Procedural History In January 2000, King Paratransit Services, Inc. (“King”) conditionally hired El to drive paratransit buses. The position involves providing door-to-door and curb-to-curb transportation service for people with mental and physical disabilities. King subcontracted with SEPTA to provide paratransit services on SEPTA’s behalf. King’s subcontract with SEPTA disallowed hiring anyone with, among other things, a violent criminal conviction. Accordingly, among the conditions stipulated in El’s offer was successful completion of a criminal background check. Within the first few weeks of El’s employment, King discovered that El had a 40-year-old conviction for second-degree murder. 2 Following the terms of King’s subcontract with SEPTA and El’s employment offer, King terminated his employment. According to King personnel, the murder conviction was their sole reason. As the background check revealed, El was convicted of second-degree murder in 1960. According to his testimony, the *236 murder took place in the context of a gang-related fight in which the victim was shot and died. El was 15 years old at the time, and the victim was 16. El claims not to have been the triggerman, and, indeed, he was not the only person convicted of the murder, but no objective report of the circumstances appears in the record before us. Following his conviction, El served three-and-a-half years for his crime. This now 47-year-old conviction is El’s only violent offense. According to the contract in place between King and SEPTA in 2000, King was required to ensure that anyone in SEPTA service as a driver or attendant have: e. no record of driving under [the] influence (DUI) of alcohol or drugs, and no record of any felony or misdemeanor conviction for any crime of moral turpitude or of violence against any person(s); f. have no record of any conviction within the last seven (7) years for any other felony or any other misdemeanor in any category referenced below (see section F.2.10.C) [listing specific offenses], and not be on probation or parole for any such crime, no matter how long ago the conviction for such crime may be. App. at 429. The parties dispute whether this provision accurately states the hiring policy that was applied to El. SEPTA contends that it does. El, on the other hand, argues that King and SEPTA applied a much broader exclusion taken from language in another part of the contract that seems to disallow hiring anyone with a criminal conviction of any kind. Specifically, El argues that King applied a nearby provision in the contract stating that “[t]he Contractor [King] shall ... reject/bar any applicant or current employee from SEPTA-related work whose record includes ... any conviction for any felony and/or misdemean- or.” App. at 430. The District Court found that King applied the narrower policy. King personnel testified that they applied the narrower policy to El and to all of its SEPTA-related applicants. Moreover, personnel from other SEPTA subcontractors testified that they applied the narrower policy in similar contracts, and SEPTA personnel testified that the narrower policy was the one that SEPTA intended for them to apply and the one that they referred to when asked for assistance with contract interpretation. SEPTA’s transactional lawyers may have been less than precise in writing an internally inconsistent contract, but all of the record evidence shows that one particular interpretation of that inconsistency prevailed, 3 and so we cannot conclude that the issue is genuinely disputed. Thus, we decide this case on the basis of the narrower hiring policy quoted above. After his employment was terminated, El filed a complaint with the Equal Employment Opportunity Commission (“EEOC”) in which he alleged that SEPTA’S hiring policy violated Title VII of the Civil Rights Act of 1964 4 by discriminating on the basis of race. Specifically, he argued that the policy has a disparate im *237 pact: because African Americans and Hispanics are more likely to have a criminal record, they are more likely to run afoul of the policy. After investigating his complaint, the EEOC found in El’s favor. The agency was, however, unable to resolve the dispute, and the Civil Rights Division of the Department of Justice declined to pursue the matter. El elected to pursue this claim himself in District Court as a class action. The District Court decided not to determine immediately whether to certify the proposed class. Rather, it allowed full discovery leading up to a period in which parties could file dispositive motions. After completing discovery, SEPTA moved for summary judgment, arguing that (1) it was not El’s employer for Title VII purposes, (2) El had not submitted sufficient evidence that SEPTA’s policy had a disparate impact on racial minorities, (3) it had submitted sufficient evidence to prove that its policy was justified by business necessity, and (4) El had not submitted sufficient evidence of an alternative policy that would accomplish SEPTA’s legitimate goal of public safety. The District Court denied the motion on the first two grounds, but granted it on the second two, thus effectively ending the litigation in SEPTA’s favor. This appeal follows. 5 II. Standard of Review The standard for awarding summary judgment is well-worn: it is fitting when “there is no genuine issue as to any material fact and ... the moving party is entitled to judgment as a matter of law.” Fed.R.Civ.P. 56(c). The moving party “bears the initial responsibility of informing the district court of the basis for its motion, and identifying those portions of the pleadings, depositions, answers to interrogatories, and admissions on file, together with the affidavits, if any, which it believes demonstrate the absence of a genuine issue of material fact.” Celotex Corp. v. Catrett, 477 U.S. 317, 323, 106 S.Ct. 2548, 91 L.Ed.2d 265 (1986) (internal quotation marks omitted). Because SEPTA sought summary judgment on its affirmative defense of business necessity, it would bear the burden of proof 6 at trial and therefore must show that it has produced enough evidence to support the findings of fact necessary to win. Marzano v. Computer Sci. Corp., Inc., 91 F.3d 497, 502 (3d Cir.1996); Sorba v. Penn. Drilling Co., Inc., 821 F.2d 200, 202-03 (3d Cir.1987). When a witness’s credibility is critical to supporting the necessary findings of fact, the District Court must consider whether there are sufficient grounds for impeachment that would place the facts to which he testifies in legitimate dispute. See Horowitz v. Fed. Kemper Life Assur. Co., 57 F.3d 300, 302 n. 1 (3d Cir.1995) (“Summary judgment is inappropriate when a case will turn on credibility determinations.”) (citing Anderson v. Liberty Lobby, Inc., 477 U.S. *238 242, 255, 106 S.Ct. 2505, 91 L.Ed.2d 202 (1986)). In considering the evidence, the court should draw all reasonable inferences against the moving party. Anderson, 477 U.S. at 255, 106 S.Ct. 2505 (“The evidence of the non-movant is to be believed, and all justifiable inferences are to be drawn in his favor.”) (citing Adickes v. S.H. Kress & Co., 398 U.S. 144, 158-59, 90 S.Ct. 1598, 26 L.Ed.2d 142 (1970)); see also Atkinson v. LaFayette Coll., 460 F.3d 447, 451 (3d Cir.2006). If the moving party successfully points to evidence of all of the facts needed to decide the ease on the law short of trial, the non-moving party can defeat summary judgment if it nonetheless produces or points to evidence in the record that creates a genuine issue of material fact. Josey v. John R. Hollingsworth Corp., 996 F.2d 632, 637 (3d Cir.1993). The non-moving party cannot rest on mere pleadings or allegations; rather it must point to actual evidence in the record on which a jury could decide an issue of fact its way. Berckeley Inv. Group, Ltd. v. Colkitt, 455 F.3d 195, 201 (3d Cir.2006) (“In this respect, summary judgment is essentially ‘put up or shut up’ time for the non-moving party: the non-moving party must rebut the motion with facts in the record and cannot rest solely on assertions made in the pleadings, legal memoranda, or oral argument.”). Put another way, it is inappropriate to grant summary judgment in favor of a moving party who bears the burden of proof at trial unless a reasonable juror 7 would be compelled to find its way on the facts needed to rule in its favor on the law. See Matsushita Elec. Indus. Co., Ltd. v. Zenith Radio Corp., 475 U.S. 574, 587, 106 S.Ct. 1348, 89 L.Ed.2d 538 (1986) (holding that summary judgment is appropriate “[w]here the record taken as a whole could not lead a rational trier of fact to find for the non-moving party....”); Hill v. City of Scranton, 411 F.3d 118, 127 n. 11 (3d Cir.2005) (“Of course, because the defendant bears the burdens of proof and persuasion on the third prong[,] ... to prevail at summary judgment on this prong the defendant must present evidence of such quality that no reasonable juror could conclude that the protected activity was the but-for cause of the termination.”); see also Turner v. Hershey Chocolate U.S., 440 F.3d 604, 612 (3d Cir.2006) (“Were we to uphold the District Court’s conclusion and grant of summary judgment we would need to conclude that reasonable jurors could not but find that rotating among all three tables is an essential function of the shaker table inspector position at Hershey.”) (emphasis added). After all, the burden of proof includes the obligation to persuade the factfinder that one’s propositions of fact are indeed true. Black's Law Dictionary 190 (7th ed.1999). Thus, if there is a chance that a reasonable factfin-der would not accept a moving party’s necessary propositions of fact, pre-trial judgment cannot be granted. Specious objections will not, of course, defeat a motion for summary judgment, but real questions about credibility, gaps in the evidence, and doubts as to the sufficiency of the movant’s proof, will. III. Discussion A. The Business Necessity Defense 1. Contours of the Defense The Supreme Court first recognized that Title VII plaintiffs can make out *239 a viable employment discrimination claim without alleging or proving discriminatory intent in Griggs v. Duke Power, 401 U.S. 424, 91 S.Ct 849, 28 L.Ed.2d 158 (1971). The Court held that plaintiffs can succeed by showing that the challenged employment policy has a discriminatory effect that is not justified by the needs of the defendant’s business. The Court announced that these “disparate impact” cases should proceed in two steps: (1) the plaintiff must prove that the challenged policy discriminates against members of a protected class, and then (2) the defendant can overcome the showing of disparate impact by proving a “manifest relationship” between the policy and job performance. This second step came to be known as the “business necessity” defense, 8 and it serves as an employer’s only means of defeating a Title VII claim when its employment policy has a discriminatory effect. 9 The Supreme Court further developed the business necessity defense over a series of cases. In Griggs, it dealt with aptitude tests administered by an employer in making hiring decisions. The Court held that discriminatory employment tests must “bear a demonstrable relationship to successful performance of the jobs for which it was used.” Griggs, 401 U.S. at 431, 91 S.Ct. 849. It further held that “any given requirement must have a manifest relationship to the employment in question.” Id. at 432, 91 S.Ct. 849. In Albemarle Paper Co. v. Moody, 422 U.S. 405, 95 S.Ct. 2362, 45 L.Ed.2d 280 (1975), it elaborated on the use of discriminatory tests by adopting the EEOC’s determination that test results must predict or correlate with “important elements of work behavior which comprise or are relevant to the job or jobs for which candidates are being evaluated.” Id. at 431, 95 S.Ct. 2362 (quoting 29 C.F.R. § 1607.4(c)). In Dothard v. Rawlinson, 433 U.S. 321, 97 S.Ct. 2720, 53 L.Ed.2d 786 (1977), the Court rejected height and weight criteria for hiring prison guards, holding that discriminatory requirements must “be shown to be necessary to safe and efficient job performance.” Id. at 331 n. 14, 97 S.Ct. 2720. The employer in that case argued that strength was an essential quality and that the height and weight criteria served as a proxy for strength. The Court rejected this argument, holding that while strength may have been an essential quality, the employer had not specified the amount of strength necessary or demonstrated any correlation between these height and weight criteria and the necessary amount thereof. Id. at 331-32, 97 S.Ct. 2720. In Connecticut v. Teal, 457 U.S. 440, 102 S.Ct. 2525, 73 L.Ed.2d 130 (1982), it held that an employer may not justify using a discriminatory test for determining promotion eligibility by also using an affirmative action system after the fact to achieve an appropriate racial balance. As the Court noted, Title VII operates not primarily to the benefit of racial or minority groups, but to ensure that individual applicants receive the consideration they are *240 due and are not screened out by arbitrary policies or devices. Id. at 453-54, 102 S.Ct. 2525. For our purposes, two aspects of these cases are noteworthy. First, the Court refused to accept bare or “commonsense”-based assertions of business necessity and instead required some level of empirical proof that challenged hiring criteria accurately predicted job performance. Dothard is particularly noteworthy because the Court rejected an employer’s common-sense argument that prison guards must be relatively strong to justify criteria that roughly measured strength. The lesson is that employers cannot rely on rough-cut measures of employment-related qualities; rather they must tailor their criteria to measure those qualities accurately and directly for each applicant. Second, the Court did not allow employers to rely on “more is better”-style reasoning to justify their policies. In Griggs, Albemarle, and Dothard, the employers argued that the challenged criteria were justified by the fact that one would naturally prefer smarter or stronger employees to less intelligent or weaker ones, and so it was of no moment that the criteria might be set a bit higher than strictly necessary. The Court held, however, that some abstract notion that more of a given quality is better is insufficient to justify a discriminatory policy under Title VII; rather, the employer must present real evidence that the challenged criteria “ ‘measure[s] the person for the job and not the person in the abstract.’ ” Dothard, 433 U.S. at 332, 97 S.Ct. 2720 (quoting Griggs, 401 U.S. at 436, 91 S.Ct. 849). The Supreme Court has never dealt directly with criminal record policies, though it has done so tangentially with criminal behavior in two cases. In McDonnell Douglas Corp. v. Green, 411 U.S. 792, 93 S.Ct. 1817, 36 L.Ed.2d 668 (1973), the Court sustained an employer’s refusal to rehire a former employee on the ground that the employee had participated in various disruptive, illegal protests in front of the employer’s premises. Id. at 794-95, 804, 93 S.Ct. 1817. Specifically, it held that the employer’s fear that this employee would continue to be disruptive in violation of the law was a legitimate business reason for the refusal. Id. at 804, 93 S.Ct. 1817. In New York City Transit Authority v. Beazer, 440 U.S. 568, 99 S.Ct. 1355, 59 L.Ed.2d 587 (1979), the Court held that it was permissible under Title VII to refuse to hire anyone using methadone to treat their addiction to illegal drugs for “safety sensitive” positions on a city transit system because such a policy serves the “legitimate employment goals of safety and efficiency.” Id. at 587 n. 31, 99 S.Ct. 1355. Although these two Supreme Court cases deal with illegal activity on the part of the applicant, neither one squarely addresses the issue of prior convictions. In McDonnell Douglas, the employer had other specific reasons for fearing disruption from the applicant than the mere existence of a criminal record. In Beazer, the Court addressed the suitability of hiring people actively using methadone to recover from addiction to illegal drugs, not the suitability of people with records of past criminal behavior. Moreover, the business necessity defense was not the focus of either case, and so the Court did not articulate the contours of the defense with any specificity. In 1989 the Supreme Court expanded the business necessity defense in Wards Cove Packing Co., Inc. v. Atonio, 490 U.S. 642, 109 S.Ct. 2115, 104 L.Ed.2d 733 (1989). There it held that a challenged discriminatory employment practice need not be necessary in the sense of “essential” or “indispensable” to pass muster under Title VII; rather, the practice must mere *241 ly “serve[ ], in a significant way, the legitimate employment goals of the employee.” Wards Cove, 490 U.S. at 659, 109 S.Ct. 2115. Even more significant was that it shifted the burden of proof from the employer to the employee. Id. Recognizing this holding as a departure from Griggs, Congress responded with the Civil Rights Act of 1991 (the “Act”), which placed back on the employer the burden of proof. 42 U.S.C. § 2000e-2(k). The Act also abrogated the Wards Cove definition of business necessity. Civil Rights Act of 1991, § 3(2), Pub.L. No. 102-166, 105 Stat. 1071 (1991) (stating that a purpose of the Act is to codify the concept of business necessity as it existed prior to Wards Cove). Congress noted both in the purpose section of the Act and in an authoritative interpretive memorandum that “[t]he terms ‘business necessity’ and ‘job related’ [as used in the Act] are intended to reflect the concepts enunciated by the Supreme Court in Griggs v. Duke Power Co., 401 U.S. 424, 91 S.Ct. 849, 28 L.Ed.2d 158 (1971), and in the other Supreme Court decisions prior to Wards Cove Packing Co., Inc. v. Atonio, 490 U.S. 642, 109 S.Ct. 2115, 104 L.Ed.2d 733 (1989).” 137 Cong. Rec. S15, 273-01 (daily ed. Oct. 25, 1991) (statement of Sen. Danforth); see also Civil Rights Act of 1991, §§ 3(2) & 105(b) (adopting the Griggs definition of business necessity and stating that only the interpretative memorandum quoted may be used in construing the Act). Thus, the text and legislative history lead directly to the conclusion that Congress intended to codify the Griggs definition of business necessity, as clarified and developed in the Supreme Court’s pre-Wards Cove jurisprudence. Unfortunately, as numerous courts and commentators have noted, Griggs and its progeny did not provide a precise definition of business necessity. See, e.g., Lanning v. Southeastern Pa. Transp. Auth., 181 F.3d 478, 488 (3d Cir.1999) (Lanning I) (noting that the Act was so unclear that both proponents and opponents of a strict business necessity standard claimed victory); Susan S. Grover, The Business Necessity Defense in Disparate Impact Discrimination Cases, 30 Ga. L.Rev. 387, 391-93 (1996); Andrew C. Spiropoulos, Defining the Business Necessity Defense to the Disparate Impact Cause of Action: Finding the Golden Mean, 74 N.C.L.Rev. 1479, 1520 (1996). Normally, we would look to additional legislative history to determine if it clarifies what Congress meant by business necessity. However, Congress stipulated that courts may not consider any document other than the interpretive memorandum quoted above as the Act’s legislative history. Civil Rights Act of 1991, § 105(b) (stating that nothing other than a specified interpretive memorandum should be considered legislative history and thereby used to construe the Act). In Lanning I and II, we heeded Congress’s instruction and looked no further than the memorandum. Lanning v. Southeastern Pa. Transp. Auth., 308 F.3d 286, 289 (3d Cir.2002) (Lanning II); Lanning I, 181 F.3d at 488. While some may be skeptical of Congress’s power to instruct courts what legislative history they may take into account when interpreting a statute, we need not consider anything beyond the interpretive memorandum because doing so would be unhelpful in this case. 10 Members of Con *242 gress simply could not agree on a precise definition of business necessity; all they could agree upon was overruling Wards Cove and reinstating the Supreme Court’s somewhat conflicting post-Griggs and pre- Wards Cove jurisprudence. Thus, our task is to be as faithful to that intent as possible. Attempting to implement the Griggs standard, we have held that hiring criteria must effectively measure the “minimum qualifications for successful performance of the job in question.” See Lanning I, 181 F.3d at 481. This holding reflects the Griggs/Albemarle/Dothard rejection of criteria that are overbroad or merely general, unsophisticated measures of a legitimate job-related quality. It is also consistent with the fact that Congress continues to call the test “business necessity,” not “business convenience” or some other weaker term. However, hiring policies need not be perfectly tailored to be consistent with business necessity. As we held in Lan-ning II, employers need not set the bar so low that they consider an applicant with some, but unreasonably low, probability of successful performance. Lanning II, 308 F.3d at 292 (“It would clearly be unreasonable to require SEPTA applicants to score so highly on the run test that their predicted rate of [job] success be 100%. It is perfectly reasonable, however, to demand a chance of success that is better than 5% to 20%.”). After all, the Supreme Court has held that Title VII never forces an employer to accept an unqualified — or even less qualified — applicant in the name of non-discrimination. Griggs, 401 U.S. at 436, 91 S.Ct. 849 (“Congress has not commanded that the less qualified be preferred over the better qualified simply because of minority origins.”). Moreover, the Supreme Court has noted that bright-line criteria, such as aptitude tests, are legitimate and useful hiring tools so long as they accurately measure a person’s qualifications. Id. Putting these standards together, then, we require that employers show that a discriminatory hiring policy accurately — but not perfectly — ascertains an applicant’s ability to perform successfully the job in question. In addition, Title VII allows the employer to hire the applicant most likely to perform the job successfully over others less likely to do so. 2. Applying the Defense to Criminal Conviction Policies Prior decisions on business necessity do not directly control here. The standards set out in Griggs and its progeny (including the standards noted by our Court in Lanning I and II) do not parallel the facts of this case. In the cases cited above, the hiring policies at issue were tests designed or used — at least allegedly — to measure an employee’s ability to perform the relevant jobs. Here, however, the hiring policy has nothing to do with the applicant’s ability to drive a paratransit bus; rather, it seeks to exclude applicants who, while able to drive a bus, pose too much of a risk of potential *243 harm to the passengers to be trusted with the job. Thus, our standard of “minimum qualifications necessary for successful performance of the job in question” is appropriate in test-score cases, but awkward here because “successful performance of the job” in the usual sense is not at issue. See Lanning I, 181 F.3d at 482. SEPTA could argue that successful performance of the job includes not attacking a passenger and, therefore, that the standard is still appropriate. However, the standard is worded to address ability, not risk. Yet, the issue before us is the risk that the employee will harm a passenger, and the phrase “minimum qualification” simply does not fit, as it is hard to articulate the minimum qualification for posing a low risk of attacking someone. The only reported appellate level case to address squarely the issue of exclusions from eligibility on the basis of prior convictions is Green v. Missouri Pac. R.R. Co., 523 F.2d 1290 (8th Cir.1975). There the employer refused to hire anyone for any position who had been convicted of any offense other than a minor traffic violation. Id. at 1292. Green had applied for an office job, and he was not considered because of a previous conviction for refusing to answer the draft (after failing to qualify as a conscientious objector). Id. at 1292-93. The Court held that the employer’s policy was too broad to be justified by business necessity. Id. at 1298-99. Green, however, presented materially different facts than those before us in two respects. First, the job in Green was an office job at a corporate headquarters; it did not require the employee to be alone with and in close proximity to vulnerable members of society. The public safety concern is of more moment in our case. Second, the hiring policy in Green prevented hiring a person with any criminal conviction, “no matter how remote, insubstantial, or unrelated to [the] applicant’s personal qualifications as an employee.” Id. at 1296 (quoting McDonnell Douglas, 411 U.S. at 806, 93 S.Ct. 1817). Here, SEPTA’s policy only prevents consideration of people with certain types of convictions — those that it argues have the highest and most unpredictable rates of recidivism and thus present the greatest danger to its passengers. In this context, Green was an easier case insofar as the Supreme Court has held firmly that an employer with an extremely broad exclusionary policy that fails to offer any empirical justification for it is unable to make out a successful business necessity defense, Dothard, 433 U.S. at 334, 97 S.Ct. 2720, whereas SEPTA has a narrower policy for a position in which criminal convictions are more job-related. The EEOC has spoken to the issue in its Compliance Manual, which states that an applicant may be disqualified from a job on the basis of a previous conviction only if the employer takes into account: 1. The nature and gravity of the offense or offenses; 2. The time that has passed since the conviction and/or completion of the sentence; and 3. The nature of the job held or sought. Equal Empl. Opp. Comm. Compliance Manual § 605 App. The EEOC clarifies that “nature and gravity of the offense” means for employers to consider the circumstances of that offense. Id. The EEOC’s Guidelines, however, do not speak to whether an employer can take these factors into account when crafting a bright-line policy, nor do they speak to whether an employer justifiably can decide that certain offenses are serious enough to warrant a lifetime ban. SEPTA’s policy arguably takes into account the sensitive nature of the job and sorts applicants by type of offense. For some offenses, it *244 considers the time since the conviction; for others, it does not. In addition, it does not appear that the EEOC’s Guidelines are entitled to great deference. While some early cases so held in interpreting Title VII, Griggs, 401 U.S. at 434, 91 S.Ct. 849, more recent cases have held that the EEOC is entitled only to Skidmore deference. EEOC v. Arabian American Oil Co., 499 U.S. 244, 257, 111 S.Ct. 1227, 113 L.Ed.2d 274 (1991) (citing Skidmore v. Swift & Co., 323 U.S. 134, 140, 65 S.Ct. 161, 89 L.Ed. 124 (1944)) (superseded by statute on unrelated grounds). Under that standard, the EEOC gets deference in accordance with the thoroughness of its research and the persuasiveness of its reasoning. Id. Here, the EEOC’s policy was rewritten to bring it in line with the Green case, but the policy document itself does not substantively analyze the statute. See Equal Empl. Opp. Comm. Compliance Manual § 605 App. 11 Considering the dearth of authority directly on point, we believe that our standards from Lanning I and II — namely that discriminatory hiring policies accurately but not perfectly distinguish between applicants’ ability to perform successfully the job in question — can be adapted to fit the context of criminal conviction policies. In a broad sense, hiring policies, such as the one at issue here, ultimately concern the management of risk. In Lanning I & II, we dealt with how employers manage the risk that applicants will be unable to perform the job in question. See Lanning II, 308 F.3d at 287-88; Lanning I, 181 F.3d at 482. Here we deal with the risk that an applicant will endanger the employer’s patrons. In both cases, it is impossible to measure the risk perfectly, 12 and in both cases Title VII *245 does not ask the impossible. It does, however, as in the case of performance-related policies, require that the policy under review accurately distinguish between applicants that pose an unacceptable level of risk and those that do not. 13 El urges us to go further and hold that Title VII prohibits any bright-line policy with regard to criminal convictions; he argues, rather, that Title VII requires that each applicant’s circumstances be considered individually without reference to any bright-line rules. We decline to go so far. If a bright-line policy can distinguish between individual applicants that do and do not pose an unacceptable level of risk, then such a policy is consistent with business necessity. Whether a policy can do so is most often a question of fact that the district courts — and juries — must resolve in specific cases. 14 3. Could a Reasonable Juror Find that SEPTA’s Policy Is Not Consistent with Business Necessity? In arguing that its policy is consistent with business necessity, SEPTA claims that it has presented evidence such that a reasonable juror must find that: (1) the job of a paratransit driver requires that the driver be in very close contact with passengers, (2) the job requires that the driver often be alone with passengers, (3) paratransit passengers are vulnerable because they typically have physical and/or mental disabilities, (4) disabled people are disproportionately targeted by sexual and violent criminals, (5) violent criminals reci-divate at a high rate, (6) it is impossible to predict with a reasonable degree of accuracy which criminals will recidivate, (7) someone with a conviction for a violent crime is more likely than someone without one to commit a future violent crime irrespective of how remote in time the conviction is, and (8) SEPTA’s policy is the most accurate way to screen out applicants who present an unacceptable risk. As an initial matter, we agree with SEPTA that these facts, if proved, would be sufficient to show that its policy is consistent with business necessity, at least as it applies to a person with a violent criminal conviction like El. 15 If someone with a violent conviction presents a materially higher risk than someone without one, no matter which other factors an employer considers, then SEPTA is justified in not considering people with those convic *246 tions. 16 For example, SEPTA may be able to show that a policy excluding all violent offenders is justified by business necessity because other factors — such as age at conviction, the number of violent convictions, and/or the remoteness of that conviction — are unreliable or otherwise fail to reduce the risk to an acceptable level. In support of its summary judgment motion, SEPTA submitted the reports of three experts. 17 All three rely heavily on data from the Department of Justice that tracked recidivism of prisoners within three years of them release from prison. Indeed, those data show relatively high rates of recidivism in those first three years. But what about someone who has been released from prison and violence-free for 40 years? The DOJ statistics do not demonstrate that someone in this position — or anything like it — is likely to reci-divate. 18 One of SEPTA’s experts was Dr. Alfred Blumstein, a noted authority on recidivism. He stated: It is also the case that an individual’s propensity to commit a future violent crime decreases as that individual’s crime-free duration increases. That is, an individual with a prior violent conviction who has been crime-free in the community for twenty years is less likely to commit a future crime than one who has been crime-free in the community for only ten years. But neither of these individuals can be judged to be less or equally likely to commit a future violent act than comparable individuals who have no prior violent history. It is possible that those differences might be small, but making such predictions of comparable low-probability events is extremely difficult, and the criminological discipline provides no good basis for making such predictions with any assurance that they will be correct. App. at 953 (citation to DOJ statistics omitted) (emphasis added). This statement bridges, as best it can, the gap between the three-year statistics and El’s 40 year-old conviction. Because Dr. Blumstein is a duly qualified professional criminologist and because nothing in the record rebuts his statement, we must take him at his word that former violent criminals who have been crime free for many years are at least somewhat more likely than members of the general population to commit a future violent act. He notes that the difference between the probability that someone with a remote conviction and someone with no conviction *247 will commit a future violent crime “might be small,” but given the marked sensitivity of the paratransit position at issue, a small but extant difference is sufficient. It is also noteworthy that Dr. Blumstein reports that the criminological discipline is incapable of distinguishing accurately between violent criminals who are and are not likely to commit future violent crimes. In other words, he believes that SEPTA’s policy distinguishes as accurately as the criminological discipline allows. Again, because we see nothing in the record rebutting this statement, we must take Dr. Blumstein at his word. SEPTA also submitted the report of Dr. Dick Sobsey, an education psychologist. Dr. Sobsey reported that disabled people are proportionately more likely than others to be the victims of violent or sexual crimes. He further reported that employees of transportation providers commit a disproportionate share of those crimes against disabled people. Like Dr. Blum-stein, Dr. Sobsey claims that the strength of violent criminal activity as a predictor of future criminal activity “moderates over time but remains regardless of how much time passes.” App. at 920. Dr. Sobsey’s report, therefore, provides evidence for SEPTA’s argument that paratransit positions are extraordinarily sensitive, and that screening out individuals with violent convictions — no matter how remote — is appropriate. Thus, on this record, we have little choice but to conclude that a reasonable juror would necessarily find that SEPTA’s policy is consistent with business necessity. This is not to say that we are convinced that SEPTA’s expert reports are ironclad in the abstract. But El chose neither to hire an expert to rebut SEPTA’s experts on the issue of business necessity nor even to depose SEPTA’s experts. These choices are fatal to his claim, for a party opposing summary judgment “cannot rest solely on assertions made in the pleadings, legal memoranda, or oral argument.” Berckeley Inv. Group, Ltd., 455 F.3d at 201. Here, there is nothing in the record that raises any reasonable credibility question about SEPTA’s expert evidence, re-buttable as it may be. Thus, we must conclude that the reasonable juror would believe those experts. Had El produced evidence rebutting SEPTA’s experts, this would be a different case. Had he, for example, hired an expert who testified that there is time at which a former criminal is no longer any more likely to recidivate than the average person, then there would be a factual question for the jury to resolve. Similarly, had El deposed SEPTA’s experts and thereby produced legitimate reasons to doubt their credibility, there would be a factual question for the jury to resolve. • Here, however, he did neither, and he suffers pre-trial judgment for it. Despite not deposing SEPTA’s experts or hiring experts of his own, El did produce evidence in the form of testimony from SEPTA personnel through which he attempts to “rebut the motion with facts in the record,” as Berckeley requires. Id. All he brought out, however, was evidence that raises questions about SEPTA’s care in formulating its hiring policy. In response to El’s interrogatories, SEPTA named 11 employees and former employees that could speak to the business necessity of SEPTA’s policy. Of those 11, El deposed eight of them. (The other three, El claims, had moved away and could not be found. SEPTA does not dispute this characterization, nor does it assert that these three would have provided materially different testimony.) Reading through those depositions, it is striking that not one of the witnesses that SEPTA named was able to explain — beyond a general concern for passenger safety — why this par *248 ticular policy was chosen from among myriad possibilities. See Dothard, 433 U.S. at 331, 97 S.Ct. 2720. Even Vincent Walsh, the drafter of the policy, could provide little insight into how the policy was written, on what research or information it was based, or why it was structured as it was. This inability is particularly striking given that the policy SEPTA claims it applied makes distinctions among crimes, setting apart some crimes for a lifetime ban from SEPTA employment and applying a seven-year ban to others. If the policy were developed with anything approaching the level of care that Griggs, Albemarle, and Dothard seem to contemplate, then we would expect that someone at SEPTA would be able to explain how it decided which crimes to place into each category, how the seven-year number was selected, and why SEPTA thought a lifetime ban was appropriate for a crime like simple assault. Almost all of El’s relevant questions about the policy were met with silence from SEPTA personnel, suggesting the reasonable inference that SEPTA has no real basis for asserting that its policy accurately distinguishes between applicants that do and do not present an unacceptable level of risk. Title VII, however, does not measure care in formulating hiring policies; rather, it requires that an employer be able to show that its policy is consistent with business necessity when challenged. Granted, the two will typically go hand-in-hand. Here, however, for all of SEPTA’s apparent loose manner in formulating and defending its policy, it produced credible expert testimony that its policy accurately screened out applicants too likely to commit acts of violence against paratransit passengers. El’s evidence (that SEPTA took little care in formulating its hiring policy), through troubling, does not directly answer SEPTA’s (that the policy, however little care went into formulating it, is accurate because those who have committed a violent crime, no matter how long ago, are more likely than the members of the general population to commit a future violent act). Therefore, El does not defeat summary judgment. El also submits the EEOC’s conclusion that SEPTA was unable to establish the suitability of its policy, and that El’s youth at the time of his conviction and the length of time since that conviction, indicate that he would not pose a threat to SEPTA’s passengers. Admissibility questions aside, 19 the EEOC determination is terse and simply asserts the relevance of El’s youth and the remoteness of his conviction without explanation, analysis, or authority. It provides nothing of substance on which the jury could rely, and so its rebuttal of SEPTA’s experts can create no more than a “scintilla” of support for El’s position. Anderson, 477 U.S. at 252, 106 S.Ct. 2505. Thus, it is insufficient to create an issue of material fact. Taking all of the record evidence into account, there is no substantive evidence on which a reasonable juror could find that SEPTA’s policy is inconsistent with business necessity. Summary judgment in SEPTA’s favor was, therefore, appropriate. B. The Alternative Policy Issue Also on appeal is the District Court’s grant of summary judgment in favor of SEPTA on the alternative policy issue. Under the Civil Rights Act, a Title *249 VII plaintiff can prevail despite an employer’s successful assertion of business necessity if the plaintiff points out an “alternative employment practice” that (1) serves the employer’s legitimate goals as effectively as the challenged practice, and (2) results in less of a disparate impact. 42 U.S.C. § 2000e-2(k)(l)(A)(ii) (codifying the pre-Wards Cove standard for showing the propriety of an alternative employment practice); Albemarle, 422 U.S. at 425, 95 S.Ct. 2362 (requiring that the alternative practice be as effective as the challenged practice and not have “a similarly undesirable racial effect.”); see also Watson v. Fort Worth Bank, 487 U.S. 977, 997-78, 108 S.Ct. 2777, 101 L.Ed.2d 827 (1988) (O’Connor, J.) (plurality) (stating that the alternative practice must be as effective as the challenged practice). El bears the burdens of proof and persuasion here. 42 U.S.C. § 2000e-2(k)(l)(A)(ii). As the Seventh Circuit Court of Appeals has held, to prevail on this issue a Title VII plaintiff must come forward with evidence that his proposed policy would have less of a disparate impact. Allen v. City of Chicago, 351 F.3d 306, 315 (7th Cir.2003). The District Court found no evidence in the record indicating that any alternative policy would have less of a disparate impact. Having reviewed the record, we agree. SEPTA is thus entitled to summary judgment on this issue. IV. Conclusion Because no reasonable juror on this record could find that SEPTA’s hiring policy is inconsistent with business necessity, we affirm the District Court’s grant of summary judgment on that issue. We also affirm the Court’s grant of summary judgment on the alternative policy issue. Footnotes [1] We do not comment on whether the policy actually has a disparate impact, as the District Court ruled that the issue was not resolvable at the summary judgment stage. See infra, at 236-37. [2] El actually had disclosed the conviction on his application, but King personnel apparently did not notice it until they examined the criminal background report. [3] It may be that other SEPTA subcontractors applied other hiring policies around this time. Because El does not — at least at this time in the litigation — represent a class, the only hiring policy properly at issue is the one applied to him, and all of the evidence indicates that it was the narrow policy that King used in deciding to terminate his employment. [4] Title VII broadly prohibits employers from discriminating against applicants and employees on the basis of race, color, religion, sex, or national origin. See generally 42 U.S.C. § 2000e-2. [5] The District Court had jurisdiction under 28 U.S.C. § 1331; we have jurisdiction under 28 U.S.C. § 1291. [6] Many of the cases we cite use the terms "burden of proof" and “burden of persuasion” interchangeably. Yet the two concepts are not identical. The burden of proof comprises the burdens of production and persuasion. McCann v. Newman Irrevocable Trust, 458 F.3d 281, 287 (3d Cir.2006). The former is the obligation to come forward with evidence of a litigant’s necessary propositions of fact. It often matters most before trial because plaintiffs who have not come forward with hard evidence to support their necessary allegations cannot survive a summary judgment motion by the defense. The burden of persuasion, on the other hand, is the obligation to convince the factfinder at trial that a litigant's necessary propositions of fact are indeed true. 2IB Charles Alan Wright & Kenneth W. Graham, Jr., Fed.. Prac. & Proc. § 5122 (3d ed.2005); Black’s Law Dictionary 190 (7th ed.1999). [7] Our use of the term "reasonable juror” here is purposeful, for the Supreme Court has held that "some metaphysical doubt as to the material facts” will not defeat a motion for summary judgment. Matsushita Elec. Indus. Co., 475 U.S. at 586, 106 S.Ct. 1348. A reasonable juror will be compelled to find for the moving party unless there are reasonable— not fanciful or illusory — concerns with the moving party’s evidence. [8] The name derives from the Griggs opinion: The [Civil Rights] Act proscribes not only overt discrimination but also practices that are fair in form, but discriminatory in operation. The touchstone is business necessity. If an employment practice which operates to exclude Negroes cannot be shown to be related to job performance, the practice is prohibited. 401 U.S. at 431, 91 S.Ct. 849 (emphasis added). [9] As we detail in Part III.B, infra, the successful assertion of the business necessity defense is not an ironclad shield; rather, the plaintiff can overcome it by showing that an alternative policy exists that would serve the employer’s legitimate goals as well as the challenged policy with less of a discriminatory effect. [10] Even if we did review additional legislative history, it would not clarify the issue. In floor debate, then-Minority Leader Senator Robert Dole stated that the Act's definition of business necessity is less strict than those articulated in the initial versions of the Senate bill that eventually became the Act and the parallel House bill. 137 Cong. Rec. S 15, 472-01 (daily ed. Oct. 30, 1991) (statement of Sen. Dole). In the initial version of the *242 Senate bill, business necessity was defined as "bear[ing] a manifest relationship” to "the performance of actual work activities required by the employer for a job or class of jobs” or "any behavior that is important to the job, but may not comprise actual work activities.” S. 1745, 102d Cong. § 7 (Sep. 26, 1991). In the initial version of the House bill, business necessity was defined as "bear[ing] a significant relationship to successful performance of the job.” H.R. 1, 102d Cong. § 3 (Jan. 6, 1991). A confusing aspect of the House bill is that, while it purports to overrule Wards Cove, the language is strikingly similar to that used in the case. Compare id. with Wards Cove, 490 U.S. at 659, 109 S.Ct. 2115. [11] Two district courts have published decisions on prior conviction policies. In EEOC v. Carolina Freight Carriers Corp., 723 F.Supp. 734 (S.D.Fla.1989), the employer refused to hire any applicant for a truck driver position who had been convicted of any felony or misdemeanor related to theft. Id. at 737-38. The employer stated that its drivers were in positions of trust because the opportunity to steal cargo was so great and it could not risk hiring anyone with a theft-related or other serious criminal conviction. Id. at 738. The District Court for the Southern District of Florida held that the employer demonstrated the requisite business necessity, specifically criticizing the holding in Green and asserting that making out a business necessity defense does not require strict proof that a conviction policy is effective. Id. at 752-53. Indeed, the employer seems not to have submitted any recidivism statistics or any other indicia of the effectiveness of its policy but for its own statements that it considered the policy effective. Id. at 754. Besides this shortcoming, Carolina Freight was decided under the Wards Cove definition of business necessity. Before Green, a District Court in Louisiana dealt with a case in which an African American bellman was fired because of a criminal record involving serious property-related crimes. Richardson v. Hotel Corp. of America, 332 F.Supp. 519 (E.D.La.1971), aff'd per curiam at 468 F.2d 915 (5th Cir.1972) (affirming the District Court without opinion). Specifically, .the hotel refused to hire anyone as a bellman who had been convicted of a serious crime (left undefined in the opinion). Id. at 521. The District Court found that the policy was warranted by business necessity because of the sensitive nature of a bellman's job, particularly a bellman’s easy access to guests’ rooms. Id. As the Richardson case was decided shortly after Griggs, the District Court applied the Griggs "manifest relationship” standard. Id. (quoting Griggs, 401 U.S. at 432, 91 S.Ct. 849). As in Carolina Freight, the District Court required no sophisticated showing that the policy was narrowly-tailored, and the opinion does not cite recidivism statistics. [12] As SEPTA discovered in the tragic case of paratransit driver David deSouza, even applicants with clean criminal records sometimes endanger passengers. At the time of his hire by King, deSouza had no prior criminal convictions. Nevertheless, he attacked and raped a passenger while serving as a SEPTA driver. [13] It may seem odd to speak of an acceptable level of risk in this context, given the horrors that drivers can inflict on disabled passengers, but, as the deSouza case demonstrates, some level of risk is inevitable, see supra note 12. SEPTA may minimize that risk to the extent reasonably possible, but whatever criteria it uses must distinguish with sufficient accuracy between those who pose that minimal level of risk and those who pose a higher level. [14] In this case, we have no occasion to hold that bright-line policies in the criminal conviction context are per se invalid. Indeed, we have upheld policies in other Title VII contexts that effectively bar an applicant from employment on the basis of a single, bright-line test result, but whatever criteria it uses must distinguish with sufficient accuracy between those who pose that minimal level of risk and those who do not. Lanning II, 308 F.3d at 291-92 (affirming SEPTA's use of a bright-line aerobic capacity test to bar applicants from employment as transit police officers). [15] It is worth noting that SEPTA also perpetually bans from paratransit employment people with convictions for crimes of moral turpitude. Because the evidence submitted has focused on violent convictions like El's, we have no occasion to determine whether SEPTA's policy may be justified as to those convicted of non-violent crimes of moral turpitude. [16] SEPTA too heavily emphasizes the sixth alleged fact: that it is impossible to predict which criminal will recidivate. This fact, if proved, is of little use because it is also impossible to predict which wow-criminal will commit a crime. What matters is the risk that the individual presents, taking into account whatever aspects of the person’s criminal history are relevant. Thus, if screening out applicants with very old violent criminal convictions accurately distinguishes between those who present an unacceptable risk, then reliance on this factor is appropriate; if the criterion is inaccurate or overbroad in the case of very old convictions, then it is inappropriate for Title VII purposes. [17] El withdrew all objections to the experts’ qualifications for purposes of SEPTA's motion for summary judgment, so we, like the District Court, assume that the experts are duly qualified to offer admissible evidence. [18] SEPTA's report from Dr. David Griffin, a statistician who principally testified on whether its policy has a disparate impact, relies entirely on these statistics. Moreover, Dr. Griffin claimed no particular expertise in criminology or any relevant discipline other than statistics. As we already address the limitations of these statistics in discussing the other two experts, we do not address further Dr. Griffin's statements. [19] EEOC determinations are relevant substantive evidence in Title VII cases. Like all relevant evidence, they are excludable under Federal Rule of Evidence 403 if substantially more prejudicial than probative. Coleman v. Home Depot, Inc., 306 F.3d 1333, 1344-45 (3d Cir.2002). The District Judge did not rule on the admissibility of the determination in this case.