IN THE UNITED STATES DISTRICT COURT FOR THE EASTERN DISTRICT OF PENNSYLVANIA
DAWN KENNEDY, CIVIL ACTION
Plaintiff, NO. 20-0395-KSM v.
CITY OF PHILADELPHIA,
Defendant.
MEMORANDUM
MARSTON, J. March 1, 2022
Plaintiff Dawn Kennedy claims that Defendant the City of Philadelphia’s Police Department’s practice of using a hair follicle test to determine drug use of its officers disparately impacts African American officers in violation of Title VII of the Civil Rights Act of 1964. (Doc. No. 1 ¶ 24.) Kennedy claims the hair follicle test exposes African American officers to a heightened risk of a false positive on the drug test because their hair is coarser, contains more melanin, and is more likely to have been treated with “ethnic hair care products.” (Id.) Kennedy also brings equal protection and municipal liability claims under 42 U.S.C. § 1983. (Id. at 32.) Presently before the Court is the City’s motion for summary judgment. (Doc. No. 34.) For the reasons below, the City’s motion is granted. I. BACKGROUND Viewing the evidence in the light most favorable to Kennedy, the relevant facts are as follows. 1 A. Factual Background 1. Kennedy’s Employment, Drug Testing, and Eventual Resignation Kennedy, an African American woman, began working as an officer with the Philadelphia Police Department in 2001.1 (Doc. No. 34-2 at 2.) Throughout Kennedy’s tenure, she was drug tested, at random, at least once a year. (Doc. No. 35-3 at 9.) She typically received
urine tests but occasionally received hair tests. (Id.) On March 20, 2019, Kennedy was randomly selected for urine and hair drug testing. (Doc. No. 34-2 at 10.) She provided a urine sample, and the Internal Affairs officers administering the test collected a hair sample from the nape of Kennedy’s neck. (Id.; Doc. No. 35-3 at 10.) In the few weeks leading up to the drug test, Kennedy had been saturating her hair in that area with Wild Growth, a hair growth oil. (Doc. No. 35-3 at 10.) Kennedy’s urine test came back negative, but her hair test was positive for marijuana. (See Doc. No. 34-5 at 9 (Kennedy’s lab results revealing that Kennedy’s hair contained 0.28 pg/mg of the THC2 metabolite, well above 0.10 pg/mg, the threshold for a positive marijuana test).)3 Kennedy’s
hair test came back negative for all other drugs tested. (Id.) Upon receiving the test results, two Internal Affairs officers met with Kennedy while she was off duty, informed her that she had tested positive for marijuana, and confiscated her Police Department-issued weapon. (Doc. No. 35-3 at 12.) Kennedy was shocked and maintains that
1 Kennedy resigned from the Police Department for personal reasons in 2011 and was reinstated after roughly six months away from the force. (Doc. No. 34-3 at 9.) 2 Tetrahydrocannabinol, or THC, is the principal psychoactive component of marijuana. (Doc. No. 34-13 at 5.) 3 The Police Department’s Medical Review Officer determined that Kennedy’s medical history and prescription drug use did not explain the levels of the THC metabolite present in Kennedy’s hair. (Doc. No. 34-1 ¶ 17; 35-2 ¶ 17.) she has never smoked, ingested, or otherwise used marijuana. (Id.) Although she has never used marijuana, as a police officer, Kennedy came into contact with marijuana “[p]ractically every night” she was on duty. (Id.) She explained that she “arrest[s] people that are either selling or smoking it” and that she was “pretty much exposed to marijuana [every night].” (Id.) For instance, when she was working in the intake room at the police station, if officers found
marijuana on an arrestee’s person, “they would leave it on the counter while they were filling out paperwork,” so she “was constantly around marijuana at work.” (Id.) Even though she was frequently around marijuana, Kennedy agrees that she rarely—if ever—actually touched marijuana because “it was always in a bag.” (Id. at 13.) Despite her negative urine test and contention that she had never used marijuana, Kennedy was under the impression that she had to resign or risk termination (which would cause her to lose her pension). (Doc. No. 35-3 at 17–18.) She resigned on March 28, 2019. (Doc. No. 34-2 at 2.) The same day she resigned, Kennedy had Lab Corp, a private laboratory testing company not affiliated with the Police Department, conduct another hair test.4 (Doc. No. 35-3 at 12; Doc.
No. 36-8 at 2.) Lab Corp technicians collected hair from the center of Kennedy’s head.5 (Id. at 15.) Kennedy uses a different hair oil on this portion of her hair than she uses on the hair at the nape of her neck. (Doc. No. 35-3 at 23.) This test came back negative for every drug tested, including marijuana. (Doc. No. 36-8 at 2.)
4 This hair test was conducted nine days after Internal Affairs conducted the initial test. (Doc. No. 35-3 at 12.) 5 Kennedy’s hair was not dyed at the time of the initial test, but she dyed the tips of her hair blonde at some point in the two weeks between receiving the initial test and the follow-up test. (Doc. No. 35-3 at 15–16.) Her roots remained undyed. (Id.) Kennedy underwent the follow-up test at her own expense, but, other than through this lawsuit, she has never informed anyone at the Police Department that the follow-up test came back negative for marijuana. (Doc. No. 35-3 at 17, 20.) 2. The Police Department’s Drug Testing Policy Philadelphia Police Department Directive 6.5 sets forth the Department’s drug testing
policy. (Doc. No. 34-9.) The policy provides that officers shall be randomly drug tested by urinalysis and/or hair testing. (Id. at 3, 11.) Typically, 90% of the tests administered are urine tests, and the remaining 10% are hair tests. (Doc. No. 36-14 at 4.) The policy sets “cut-off levels” for what constitutes a positive test and provides that an officer will be dismissed if her urine or hair test indicates a positive result. (Id. at 12, 25.) If an officer receives a positive hair test but disputes the result, she may undergo a “reconfirmation test”: within ten days of receiving a positive result, an officer wishing to reconfirm the result must provide a hair sample to her commanding officer (who will transmit the sample to the Police Department’s contracted laboratory) and bear the costs of the reconfirmation testing. (Id. at 14.) The policy also sets
forth a procedure for instances when officers are accidentally exposed to a controlled substance: if an officer indirectly or accidentally “breathed, ingested . . . or otherwise internalized illegal controlled substances,” the officer must “immediately submit a memorandum detailing the incident to their Commanding Officer.” (Id. at 16.) The Police Department implemented the drug testing policy because the Department “has a paramount interest in protecting and serving the public by ensuring that its officers are fit to perform their duties” and because drug use “has an adverse effect upon a police officer’s ability to execute their duties.” (Id. at 3.) From January 1, 2014 through December 31, 2019, the Police Department administered 23,898 drug tests to its officers. (Doc. No. 34-1 ¶ 24; Doc. No. 35-2 ¶ 24.) A racial breakdown of the officers tested is provided in the chart below. Race Officers Tested Caucasian 13,229 African American 7,992 Latino 2,068 Indian 31 Other 97
(Id.) Of those tested, five officers tested positive for marijuana through urine tests—four African American officers and one Caucasian officer. (Doc. No. 34-1 ¶ 25; Doc. No. 35-2 ¶ 25.) Over the same period, the Police Department conducted 2,395 combination hair and urine tests.6 (Id.) Four officers tested positive for some drug though hair testing—two African American officers and two Caucasian officers. (Id.) This means that 0.01% of Caucasian officers and 0.02% of African American officers tested positive for marijuana though hair testing. (Doc. No. 34-1 ¶ 26; Doc. No. 35-2 ¶ 26.) 3. The Risk of False Positives in Hair Tests Kennedy’s expert, Dr. David A. Kidwell, opines that African Americans are more likely to receive false positives in hair tests than are individuals of other races. (Doc. No. 35-4 at 15.) He explains that individuals who are “passively exposed” to THC through their environment may test positive for certain drugs even if they have not actively used them. (Id. at 3–4.) He explains that African hair is more likely to absorb drugs from the environment because dark hair has more melanin, and certain drugs, including cocaine, amphetamines, and opiates, bind to melanin. (Id. at 4.) Critically, however, Dr. Kidwell acknowledges that THC “[is] not considered to bind to melanin.” (Id. at 5.) He notes that THC may bind to sebaceous secretions
6 The remaining tests (of which there were approximately 21,000) were urine tests with no hair testing component. but does not indicate whether individuals with African hair are more likely to have increased sebaceous secretions. (Id.) Dr. Kidwell also states that hair oils, such as the oils Officer Kennedy used, “allow concentration and binding of THC from the environment” and further explains that such exposure is “greatly enhanced” by cosmetic hair treatments, such as straightening treatments, “that African Americans use.” (Id. at 15.)
The City’s expert, Dr. Leo Kadehjian, disputes that African hair is more likely to passively absorb THC from the environment. (Doc. No. 34-8 at 18.) He explains that basic drugs, such as cocaine or methamphetamine, preferentially bind with melanin but acidic drugs, such as THC, have not been shown to preferentially bind with melanin. (Id.) But even if THC were more likely to bind to hair follicles that contain more melanin, the Police Department tests only for the THC metabolite, THC carboxylic acid. (Id. at 6.) Dr. Kadehjian explains that the THC metabolite “does not exist in the environment, does not exist in cannabis plant material or THC-containing products, [and] does not exist in cannabis smoke.” (Id.) The THC metabolite is only formed “upon passage of THC through the body and liver metabolism,” so the presence of
THC metabolite in hair “is recognized as definitive for ingestion of THC.” (Id.) Dr. Kidwell disagrees and contends that THC can metabolize on a hair follicle if exposed to oxygen and light. (Doc. No. 35-4 at 8.) This, Dr. Kidwell suggests, undermines Dr. Kadehjian’s conclusion that the presence of the THC metabolite necessarily indicates ingestion. (Id.) B. Procedural History On October 2, 2019, Kennedy brought suit against the City in the Philadelphia Court of Common Pleas, alleging that the Police Department’s policy of conducting hair follicle tests has a disparate impact on African Americans. (Doc. No. 1 at 24.) The City removed the case to this Court. (Id. at 6–7.) Following a period of discovery, the City moved for summary judgment on all claims. (Doc. No. 34.) First, the City argues that Kennedy fails to establish a prima facie Title VII disparate impact claim because she has not produced any evidence showing that the Police Department’s use of hair drug tests disparately impacts African American officers. (Id. at 6.) Second, the City argues that, even if Kennedy has presented a prima facie disparate impact,
her claim still fails because the City’s decision to utilize hair testing serves a legitimate business goal. (Id. at 9.) Third, the City argues that Kennedy’s equal protection claim fails because Section 1983 requires intentional, rather than unintentional, discrimination. (Id. at 17.) Finally, the City argues that the Section 1983 municipal liability claim fails because she has not established an underlying constitutional violation. (Id. at 18.) Kennedy opposes the motion. (Doc. No. 35.) She argues that her Title VII disparate impact claim survives because she has shown that hair testing for marijuana has a discriminatory effect on African Americans (id. at 6) and that an alternate policy exists that would serve the Police Department’s legitimate goals without a disparate impact (id. at 18). Kennedy argues that
her equal protection claim survives because the Police Department’s use of hair testing despite their purported knowledge of its potential racial biases rises to the level of intentional discrimination. (Id. at 21.) Last, she argues that she has established municipal liability because any system that disparately impacts African Americans must be the product of the City’s decisionmakers’ deliberate indifference. (Id. at 20.) II. LEGAL STANDARD Summary judgment is appropriate when the “pleadings, depositions, answers to interrogatories, and admissions on file, together with the affidavits, if any, show that there is no genuine issue as to any material fact and that the movant is entitled to judgment as a matter of law.” Fed. R. Civ. P. 56(c). “[T]he mere existence of some alleged factual dispute between the parties will not defeat an otherwise properly supported motion for summary judgment; the requirement is that there be no genuine issue of material fact.” Anderson v. Liberty Lobby, Inc., 477 U.S. 242, 247–48 (1986). A dispute is genuine if the “evidence is such that a reasonable jury could return a verdict for the nonmoving party,” and a fact is material if it “might affect the
outcome of the suit under the governing law.” Id. at 248. “[A]t the summary judgment stage the judge’s function is not himself to weigh the evidence and determine the truth of the matter but to determine whether there is a genuine issue for trial.” Id. at 249. “[T]he inferences to be drawn from the underlying facts must be viewed in the light most favorable to the party opposing the motion.” Matsushita Elec. Indus. Co. v. Zenith Radio Corp., 475 U.S. 574, 587 (1986) (quotation marks and alterations omitted). III. ANALYSIS A. Title VII Kennedy argues that the Police Department’s practice of conducting hair follicle tests
disparately impacts African American officers in violation of Title VII of the Civil Rights Act of 1964. (Doc. No. 1 ¶ 24.) Under Title VII, it is unlawful for an employer to “fail or refuse to hire or discharge any individual, or otherwise to discriminate against any individual . . . because of such individual’s race, color, religion, sex, or national origin.” 42 U.S.C. § 2000e-2(a)(1). Title VII bars practices of intentional discrimination and practices that “are not intended to discriminate but in fact have a disproportionately adverse effect,” or “disparate impact,” on minorities. Ricci v. DeStefano, 557 U.S. 557, 577 (2009). “Disparate-impact litigation proceeds in three steps.” Nat’l Ass’n for the Advancement of Colored People v. N. Hudson Regional Fire & Rescue, 665 F.3d 464, 476 (3d Cir. 2011). First, a plaintiff must establish a prima facie case by demonstrating that the application of the challenged practice has caused a “significant” pattern of discrimination. Id. (citing Newark Branch, Nat’l Ass’n for the Advancement of Colored People v. City of Bayonne, 134 F.3d 113, 121 (3d Cir. 1998)). Once the plaintiff has established her prima facie case, the burden shifts to the employer to demonstrate that the challenged practice is “consistent with business necessity.” Id. (citing 24
U.S.C. § 2000e-2(k)(1)(A)(i)). Finally, the burden shifts back to the plaintiff to show that alternative practices would viably address the business necessity with “less discriminatory effects.” Id. (citing 24 U.S.C. § 2000e-2(k)(1)(A)(ii)). The Court now considers Kennedy’s claim pursuant to this framework. 1. Prima Facie Case To establish a prima facie case of disparate impact, a plaintiff must “‘(1) identify a specific employment policy or practice of the employer and (2) proffer evidence, typically statistical evidence, (3) of a kind and degree sufficient to show that the practice in question has caused’ an adverse impact, such as exclusion of applicants for jobs or promotions, ‘(4) because
of their membership in a protection group.’” Green v. City of Philadelphia, Case No. 2:19-cv- 03156-JDW, 2020 WL 7695956, at *3 (E.D. Pa. Dec. 28, 2020) (quoting Stagi v. Nat’l R.R. Passenger Corp., 391 F. App’x 133, 136 (3d Cir. 2010)). “An adverse effect on a single employee, or even a few employees, is not sufficient to establish disparate impact.” Massarsky v. Gen. Motors Corp., 706 F.2d 111, 121 (3d Cir. 1983). Statistical Significance To show that the challenged practice caused a disparate impact, the plaintiff must “prove a significant statistical disparity and . . . demonstrate that the disparity [complained] of is the result of one or more of the employment practices [being] attack[ed].” N. Hudson Regional Fire & Rescue, 665 F.3d at 476 (emphases added). “The most widely used means of showing that an observed disparity in outcomes is sufficiently substantial . . . is to show that the disparity is sufficiently large that it is highly unlikely to have occurred at random.” Stagi, 391 F. App’x 137. This is typically shown by statistical significance. Id. Statistical significance can be measured by an “odds ratio.” In re Zoloft (Sertraline
Hydrocloride) Prods. Liab. Litig., MDL No. 2342, 2015 WL 7776911, at *2 (E.D. Pa. Dec. 2, 2015). For instance, the odds ratio in this case could be calculated by dividing the frequency with which African American officers receive false positive results from hair tests by the frequency with which non-African American officers receive false positive results from hair tests. Where the odds of testing positive are the same for African American officers as for other officers, the odds ratio is one. Id. An odds ratio below one would indicate that African American officers are less likely to receive false positives than their non-African American counterparts, and an odds ratio above one would indicate that African American officers are more likely to receive false positives than their non-African American counterparts. Cf. Soldo v.
Sandoz Pharms. Corp., 244 F. Supp. 2d 434, 535 (W.D. Pa. 2003). The City argues that Kennedy fails to establish a prima facie case of disparate impact because she has not put forth “any evidence establishing a significant pattern of disparate treatment against African American officers.” (Doc. No. 34 at 9.) Kennedy responds, arguing that Dr. Kidwell provided a statistical analysis “showing the City’s hair testing policy caused a higher percentage of false positives in African American officers.” (Doc. No. 35 at 8.) These arguments require the Court to provide an in-depth explanation of Dr. Kidwell’s analyses. Dr. Kidwell analyzed a data set from the City that provides detail on the combination hair and urine drug tests the Police Department conducted from January 1, 2014 through December 31, 2019. (Doc. No. 35-3 at 1; Doc. No. 36-13.) The data set includes the following information on each of the 2,395 combination hair and urine drug tests conducted during the relevant period: • Biographical information about the officer being tested;
• The date the tests were conducted; • The reason the tests were conducted (random, reasonable suspicion, reinstatement, return from medical leave); • The result of the hair test; • Drugs found in the hair test, if any; • The result of the urine test; • Drugs found in the urine test, if any; and • The disposition of positive tests (officer dismissed, officer resigned, positive result cleared by medical review). (See generally Doc. No. 36-13.) A separate data set lists the race of every officer tested during this period.7 (See Doc. No. 34-7.) Dr. Kidwell proffers several cuts of data in an attempt to ascertain some statistically significant connection between the challenged practice and the purported disparate impact.8
7 The City’s data indicates that each of the officers tested self identifies as one of the following races: Caucasian, African American, Indian, Latino, Asian, or Other, but Dr. Kidwell states that this data set is problematic because “the Federal Government considers Latino as a nationality rather than a race.” (Doc. No. 35-5 at 2.) To remedy this deficiency, Dr. Kidwell divided the data on those officers who identified as “Latino” equally into the African American and Caucasian groups. (Id. at 5.) 8 Dr. Kidwell’s report also details purported issues with the way the Police Department’s contracted laboratory conducted the March 2019 drug test of Kennedy’s hair, but the fact that Kennedy alone may have suffered an adverse impact as a result of the City’s practice of conducting hair tests is not relevant to or sufficient to establish her disparate impact claim. See Gyda v. F.B.I. Crime Lab., Civil Action No. 13–7591, 2015 WL 4077009, at *3–4 (E.D. Pa. July 6, 2015) (holding that the plaintiff’s disparate impact claim failed where he alleged only that the defendant would not hire him but did not allege that similarly situated applicants were unable to gain employment with the defendant). (Doc. No. 35-5 at 3–6.) First, Dr. Kidwell considers every positive result from hair tests conducted during the relevant period and finds that African American officers were 1.65 times more likely to test positive than Caucasian officers. (Id. at 3.) Given the small sample size, however, an odds ratio of 1.65 is not statistically significant. (See id. at 3 (not arguing that 1.65 odds ratio in this data
set is statistically significant); id. at 5 (conceding that an odds ratio of 1.75 is not statistically significant).) Indeed, Dr. Kidwell concedes that an odds ratio of 2.14 would be statistically significant but an odds ratio of 1.82 or lower would not be given the relatively small sample size in the case at hand. (Doc. No. 35-3 at 5.) The insignificance of this finding is further bolstered by the fact that this analysis considers all positive results, not just false positives, and many of the positive results factored into this analysis were likely correct (especially those initiated on reasonable suspicion of drug use or those confirmed by the urine test). Second, Dr. Kidwell calculates the odds ratio for every hair test conducted during the relevant period that yielded a positive result and was not cleared by medical review.9 (Id. at 5.)
He finds that African American officers are 1.75 times more likely to test positive and not be cleared by medical review than are their Caucasian counterparts but again notes that an odds ratio of 1.75 is not statistically significant given the small sample size. (Id.) Third, Dr. Kidwell considers the number of officers who tested positive for marijuana in hair or urine tests. (Id.) However, in the relevant period, only five officers tested positive for
9 A positive test is excused, or “cleared by medical review,” where the officer has legitimately been prescribed and is taking the drug for which he tested positive. (See Doc. No. 34-9 at 13.) For instance, if an officer had been prescribed OxyContin and tested positive for oxycodone, that positive result would likely be cleared by medical review and the officer would not face any ramifications for that positive test. marijuana: “one was identified by urine alone, one by hair and urine, and three by hair alone.” 10 (Id.) Of the three officers who tested positive by hair alone, two were African American and one was Caucasian. (Id. at 5–6.) Although this yields an odds ratio of 2:1,11 Dr. Kidwell concedes that this “is insufficient with only three people for statistical analysis.” (Id. at 6.) Finally, Dr. Kidwell considers the number of positive results in hair tests that were
conducted at random—that is, this set does not include data on officers who were tested for reinstatement, return from medical leave, or reasonable suspicion. (Id. at 6.) This data set includes all positive results, not just false positives. (Id.) It also includes positive results that were later cleared by medical review, with no explanation by Dr. Kidwell about how these results are relevant to the analysis. (Id.) In fact, 10 of the 14 positive results in this sample were later cleared by medical review.12 (See Doc. No. 36-13 at 39–40.) The odds ratio is 14.8, which indicates that when the data is framed in this very specific way, African American officers are 14.8 times more likely than their non-African American counterparts to have a positive hair test. (Doc. No. 35-5 at 6.)
Despite these efforts to slice the data, Dr. Kidwell has not shown any relevant statistical significance. In conducting a statistical significance analysis, the Court must consider data that is relevant to the alleged disparate impact; “we cannot compare apples to oranges.” Meditz v.
10 This runs counter to the statement both parties have agreed to that, during the relevant period, five officers tested positive for marijuana by urinalysis (see Doc. No. 34-1 ¶ 25; Doc. No. 35-2 ¶ 25); however, because the discrepancy is with the number of positive results from urine tests, it is not relevant to the Court’s analysis. 11 It is not clear to the Court why this odds ratio is presented as a true ratio (2:1) while the other odds ratios are presented as figures (e.g., 1.65). 12 Of the “Positive HAIR List” data set, the following rows provide data on tests that were initiated at random and yielded a positive result: 2, 18, 19, 28, 30, 33, 38, 39, 41, 47, 62, 67, 75, and 79. (See Doc. No. 36-13 at 39–40.) But, of those random tests yielding a positive result, the tests detailed in the following rows were later cleared by medical review: 2, 18, 19, 30, 33, 38, 39, 41, 47, and 79. (Id.) City of Newark, 658 F.3d 364, 370 (3d Cir. 2011). Here, Kennedy is challenging the use of hair tests because, she claims, it disproportionately yields false positive results for African American officers. (Doc. No. 1 ¶ 12; Doc. No. 35 at 8.) Given this, the only data the Court should be considering is data on the number of false positive tests by race. The court in Green v. City of Philadelphia, Case No. 2:19-cv-03156-JDW, 2020 WL
7695956 (E.D. Pa. Dec. 28, 2020), took this approach in a nearly identical case. The plaintiff, a former Philadelphia police officer, challenged the City’s policy of conducting hair testing, arguing that it had a disparate impact on African American officers because they are more likely to test positive for cocaine given the makeup of their hair. Id. at *1. The plaintiff provided a raw data set indicating that .038% of African American officers and .008% of Caucasian officers had a positive hair test. Id. at *3. The Court concluded that the plaintiff failed to establish a prima facia case because “th[e] raw statistical data does not tell the Court . . . whether any of those tests were false positives.” Id. (emphasis added). Here, Kennedy has not presented any relevant statistical evidence showing how much
more likely African American officers are to have false positive results compared to other officers. The Court appreciates that it may have been difficult to identify which positive test results were false positives since they were administered some years before Dr. Kidwell drafted his report, but Dr. Kidwell made no effort to exclude data from tests that were highly unlikely to be false positives. For instance, the sole analysis that yields a statistically significant result calculates the odds ratio for all positive hair results, even those that were later cleared by medical review. A positive result that is later cleared by medical review is almost certainly a true positive and has no relevance to whether African American officers are disproportionately likely to show false positive results in hair tests. Even assuming arguendo that the relevant point of comparison is all positive hair tests (and not just false positive hair tests), African American officers are only 1.65 times more likely to test positive than are Caucasian officers, which Dr. Kidwell himself admits is not statistically significant. (Doc. No. 35-5 at 3.) Unable to identify any statistical significance in the relevant data set, Dr. Kidwell
contends that “maybe . . . the best comparison of all” is positive hair tests that were initiated at random rather than for some reason. (Id. at 6.) This data set shows that African American officers are 14.8 times more likely to have a positive hair test than their non-African American peers. (Id.) But this data set is irrelevant to the Court’s analysis for several reasons. First, the data set is very small—it only includes 14 tests. See Green, 2020 WL 7695956, at *3 (explaining that a sample set of five positive tests was “very small,” so much so that “the Court [could] not infer a disparity). Second, the data set includes all positive results, not just false positive results. Finally, and perhaps most critically, 10 of the 14 positive tests in the data sample were later cleared by medical review. The 10 officers who tested positive but were
cleared did not suffer an adverse impact from the hair tests—their positives were, for all intents and purposes, negatives. Taken together, these considerations render the 14.8 odds ratio upon which Kennedy bases her disparate impact claim meaningless. Removing the positive tests that were cleared by medical review leaves only four positive tests in the sample. Kennedy has not proffered any statistical analysis of those four positive, random tests, nor could she, given the small sample size. See Massarsky, 706 F.2d at 121 (“An adverse effect on a single employee, or even a few employees, is not sufficient to establish disparate impact.”). Kennedy has failed to present any relevant statistical analysis showing a statistically significant connection between the Police Department’s use of hair tests and adverse impacts, such as terminations or forced resignations, on African American officers due to false positive results, so she has failed to establish a prima facie case. See Hanrahan v. Blank Rome LLP, 142 F. Supp. 3d 349, 355 (E.D. Pa. 2015) (holding that the plaintiff, a former Drexel law student denied employment at three major Philadelphia firms, failed to establish a prima facie case for disparate impact through statistical evidence where he considered only one years’ worth
of data and “arbitrarily limit[ed]” the statistical analysis to law students from Temple, Drexel, and Penn); McCutchen v. Sunoco, Inc., No. CIV.A. 01–2788, 2002 WL 1896586, at *6 (E.D. Pa. Aug. 16, 2002) (holding that the plaintiff failed to provide relevant statistical analysis where he provided data on the pay history of management-level employees but “fail[ed] to provide information regarding the pay history of employees at his level); see also Stagi, 391 F. App’x at 145 (explaining that statistical analyses must be “relevant and otherwise compelling” in addition to significant to support a prima facie case for disparate impact). Kennedy contends that Dr. Kidwell’s statistical analysis is sufficient to show a significant disparate impact because it is similar to the statistical analysis in Jones v. City of Boston, 752
F.3d 38 (1st Cir. 2014). In Jones, the plaintiffs, individuals who had been terminated from or denied employment by the Boston Police Department because they tested positive for cocaine, alleged the Department’s use of hair tests had a disparate impact on African American officers. Id. at 41. The plaintiffs presented evidence that more African American officers tested positive for cocaine than did Caucasian officers, even though there were more Caucasian officers. Id. at 42–43. The plaintiffs’ expert analyzed a data set detailing every positive cocaine hair test and found that there was a statistically significant chance that African American officers were more likely to test positive for cocaine in hair tests than were their Caucasian counterparts. Id. at 42– 44. Given this data, the First Circuit held that the plaintiffs had established a prima facia case. Id. at 46–47. Jones is readily distinguishable from the case at hand. The plaintiffs’ expert in Jones found a statistically significant correlation in the relevant data set—every hair test that was positive for cocaine. Id. at 44–45. Dr. Kidwell analyzed essentially the same data set for Philadelphia—every hair test that was positive for any drug. (Doc. No. 35-5 at 3.) But he found
no statistically significant connection between hair testing and a disparate impact on African Americans. (See id. (explaining that his analysis of the “hair testing positives” showed that African American officers were 1.65 times more likely to be “identified as positive” through a hair test than Caucasian officers but declining to describe the 1.65 odds ratio as “statistically significant”).) In Jones, the plaintiffs’ claims survived summary judgment because they presented evidence of statistical significance in analyzing the relevant data set, but Kennedy has failed to produce any such evidence, so her claim fails. Other Evidence Alternatively, Kennedy argues her claim does not fail because she can establish a prima
facie case of disparate impact without statistical evidence. (Doc. No. 35 at 9.) Kennedy is correct that she is not strictly required to put forth statistical evidence, but courts will excuse a plaintiff’s failure to proffer statistical evidence in a disparate impact lawsuit only in the most extreme circumstances, and such cases are “few, far between, and only where the raw numbers evince the most egregious disparities.” Luceus v. Rhode Island, C.A. No. 15–489 WES, 2018 WL 1626263, at *6 (D.R.I. Mar. 30, 2018), aff’d, 923 F.3d 255 (1st Cir. 2019). For instance, in Chin v. Port Authority of New York & New Jersey, 685 F.3d 135, 150–53 (2d Cir. 2012), the Second Circuit explained that the plaintiffs need not show statistical significance where no Asian candidates were promoted during the relevant period. Similarly, in Equal Employment Opportunity Commission v. Steamship Clerks Union, Local 1066, 48 F.3d 594, 605–06 (1st Cir. 1995), the First Circuit held that the plaintiff had established a prima facie case of disparate impact even though it had not presented any statistical analysis because, in a six-year period, 30 new white members were welcomed into a sponsorship-based union, but zero new minorities were. The First Circuit explained that it would have been more helpful had the EEOC presented
“processed” data but sustained its claim based on the total exclusion of minorities. Id. at 606–07. The case at hand does not present similarly extreme circumstances, and Dr. Kidwell’s conclusions that “hair drug tests have different outcomes for different races because of the composition of African American hair as well as the use of ethnic hair care products” does not show a disparate impact. “[T]he most that it establishes is that there is a risk of a disparate impact.” Green, 2020 WL 7695956, at *3; see also id. (explaining that the expert’s conclusions that African Americans may be more likely to have false positive results in hair testing merely created a risk of disparate impact but did not establish a prima facie case because there was no evidence that the risk was ever realized in the Police Department’s hair drug testing policy).
* * * It is possible that Kennedy never used marijuana. But “the City does not have to have a perfect testing methodology to comply with Title VII. It just has to have a methodology that does not discriminate.” Id. at *4. Because Kennedy has not shown that the Police Department’s policy of conducting hair tests has a statistically significant disparate impact on African American officers or that this is the rare, extreme case where statistical evidence is unnecessary, she has failed to establish a prima facie case, and her disparate impact claim fails. Accordingly, the Court grants the City’s motion for summary judgment on the disparate impact claim. B. Equal Protection Under 42 U.S.C. § 1983 Next, Kennedy alleges that the Police Department’s use of hair tests violates 42 U.S.C. § 1983. (Doc. No. 35 at 21.) The Constitution protects only against intentional discrimination, and disparate impact claims are rooted in allegations of unintentional discrimination, so they are not actionable under Section 1983. See Dunleavy v. New Jersey, No. Civ. No. 05–3865(DRD),
2008 WL 199467, at *5 n.2 (D.N.J. Jan. 18, 2008) (“[D]isparate impact claims are actionable only under Title VII (and not § 1983, since constitutional substantive due process protects only against intentional discrimination) . . . .”). The City argues that Kennedy’s Section 1983 claim fails because it is based on her disparate impact claim and “disparate impact claims are not actionable under Section 1983.” (Doc. No. 34 at 17.) Kennedy, however, argues that her Section 1983 claim is based on the City’s intentional discrimination; specifically, she contends that the challenged conduct is intentionally discriminatory because “the City actively uses a test where African Americans are 14.8 times more positive than Caucasians.” (Doc. No. 35 at 21.)
As a threshold matter and as discussed above in great detail, see supra Section III.A.1, the “14.8 times more likely to test positive” statistic is misleading and based on irrelevant data. But even if African American officers were so much more likely to have positive hair tests than their Caucasian peers, such disparity does not evidence intentional discrimination. The Police Department’s policy of using hair tests is facially neutral: African American officers are not tested more frequently than their colleagues, nor are they subject to different cut-off levels for what constitutes a positive test. In fact, Kennedy concedes that the Police Department’s practice of “us[ing] hair drug testing for its officers” is “a facially neutral standard.”13 (Doc. No. 35 at 6.)
13 At oral argument, Kennedy’s counsel confirmed that the policy is facially neutral. “To prove intentional discrimination by a facially neutral policy, a plaintiff must show that the relevant decisionmaker (e.g., a state legislature) adopted the policy at issue because of, not merely in spite of, its adverse effects upon an identifiable group.” Pryor v. Nat’l Collegiate Athletic Ass’n, 288 F.3d 548, 562 (3d Cir. 2002) (cleaned up). But Kennedy has not offered any evidence to show (and does not even contend) that the Police Department conducts hair tests
because of their potentially discriminatory impact on African American officers. See El v. Se. Pa. Transp. Auth., 418 F. Supp. 2d 659, 673 (E.D. Pa. 2005) (granting summary judgment on an intentional discrimination claim based on a facially neutral policy because “there [was] no evidence whatsoever on this record that SEPTA enacted or imposed the criminal record policy at issue . . . for the purpose of discriminating against African-Americans”). Contra Jackson v. Se. Pa. Transp. Auth., Civil Action No. 08–4572, 2009 WL 637460, at *6 (E.D. Pa. Mar. 10, 2009) (holding that the plaintiff stated a prima facie case of intentional discrimination where he alleged that “Defendant adopted and continued to use its facially neutral Criminal Background policy in an effort to intentionally discriminate against African American job applicants”).
Because Kennedy challenges a facially neutral policy and has not demonstrated that the policy was implemented because of its potentially discriminatory impact, she has failed to show intentional discrimination. Accordingly, her Section 1983 claim is not actionable, and the Court grants the City’s motion for summary judgment. C. Municipal Liability Under 42 U.S.C. § 1983 Finally, Kennedy brings a Section 1983 municipal liability claim against the City, alleging that the Police Department’s policy of conducting hair tests is unconstitutional and “led to the loss of her employment.” (Doc. No. 35 at 20.) To establish municipal liability, a plaintiff must show “(1) a constitutional violation by a municipal actor that (2) was caused by a municipal policy or custom.” Boyle v. City of Philadelphia, 476 F. Supp. 3d 101, 114 (E.D. Pa. 2020) (citing Monell v. Dep’t of Soc. Servs. of N.Y., 436 U.S. 658, 694 (1978)). “[A] municipality cannot be found liable on a Monell claim where there has been no underlying violation of rights . . . .” Johnson v. City of Philadelphia, 105 F. Supp. 3d 474, 482 (E.D. Pa. 2015). Kennedy has failed to establish that the Police Department’s use of hair tests disparately
impacts African American officers or that the Police Department implemented the policy with the specific intent to discriminate against African American officers, so her municipal liability claim fails. See Green, 2020 WL 7695956, at *3 (entering summary judgment on municipal liability claim because the plaintiff “proffered no evidence that he was treated differently than other police officers based on his race” or that the defendant intentionally discriminated against him through its use of hair tests); see also Boyle, 476 F. Supp. 3d at 115 (granting the defendant’s motion for summary judgment on a municipal liability claim where the plaintiff failed to present statistically significant evidence that the challenged policy disparately impacted minorities).
IV. CONCLUSION For the reasons above, the Court grants the City’s motion for summary judgment. An appropriate Order follows.