Case 3:20-cv-07499-VC Document 70 Filed 09/28/22 Page 1 of 4
`
`
`
`
`
`
`
`UNITED STATES DISTRICT COURT
`
`NORTHERN DISTRICT OF CALIFORNIA
`
`
`
`THOMAS LIU,
`
`Plaintiff,
`
`v.
`
`UBER TECHNOLOGIES INC.,
`
`Defendant.
`
`
`
`
`
`
`
`Case No. 20-cv-07499-VC
`
`
`ORDER GRANTING MOTION TO
`DISMISS
`
`Re: Dkt. No. 59
`
`The motion to dismiss the third amended complaint is granted.
`
`A plaintiff in a disparate impact case must allege three elements: “(1) a significant
`
`disparity with respect to employment for the protected group, (2) the existence of a specific
`
`employment practice or set of practices, and (3) a causal relationship between the identified
`
`practice and the disparity.” Liu v. Uber Technologies, Inc., 551 F. Supp. 3d 988, 990 (N.D. Cal.
`
`2021) (citing Freyd v. University of Oregon, 990 F.3d 1211, 1224 (9th Cir. 2021)).
`
`Liu alleges that terminations at Uber occur on a racially disparate basis because of the
`
`company’s reliance on customer ratings. In his first amended complaint, Liu plausibly alleged
`
`that racial discrimination could affect customer ratings, including in the rideshare industry. These
`
`allegations go primarily to the second and third elements. But Liu did not adequately allege that
`
`this legitimate concern about racial discrimination actually manifested itself in driver
`
`terminations at Uber, and so he did not satisfy the first element. Liu, 551 F. Supp. 3d at 991.
`
`In his second amended complaint, Liu attempted to address this problem by conducting a
`
`survey of Uber drivers. The survey asked drivers to check a box indicating their race, and it
`
`asked, “If you have been deactivated by Uber, was it because your ratings were too low?”
`
`

`

`Case 3:20-cv-07499-VC Document 70 Filed 09/28/22 Page 2 of 4
`
`
`
`According to Liu, the results of the survey demonstrated that minority drivers were terminated
`
`for their ratings at a higher rate than white drivers. But, as explained in a prior ruling, this survey
`
`was “essentially meaningless.” Liu v. Uber Technologies, Inc., No. 20-CV-07499-VC, 2022 WL
`
`1613285, at *2 (N.D. Cal. May 23, 2022). Because the survey focused only on drivers who had
`
`been deactivated, it could show a disparity between white and minority drivers, even if there was
`
`no actual disparity given the total driver population. See id. at *1. It was additionally flawed
`
`because it asked whether drivers identified as “Latinx,” a term that, as Liu himself concedes,
`
`likely led many confused Latino respondents to identify as “other.” Id.
`
`
`
`In response to the dismissal, Liu made one change to his complaint. In a footnote to his
`
`third amended complaint, Liu explains that his counsel sent a follow-up email to survey
`
`respondents who answered “no” to the initial survey. According to Liu, of those drivers who
`
`responded to that follow-up, 51.7% stated that they had not been deactivated at all.
`
`
`
`This new information makes Liu’s complaint worse, not better. As previously explained,
`
`“Liu is correct to emphasize that a survey described in a complaint alleging racial disparity need
`
`not be put to the same rigorous test that might apply at the summary judgment stage. But a
`
`survey must provide at least some information from which racial disparity can be plausibly
`
`inferred.” Id. at *2. The original survey was phrased in a way that seemed to elicit responses only
`
`from drivers who had been deactivated—not Uber’s total driver population. The fact that a
`
`number of Uber drivers misunderstood this confusingly worded survey makes it even less useful.
`
`It certainly does not help Liu plead the first element of a disparate impact claim—namely, “a
`
`significant disparity with respect to employment for the protected group.”
`
`Lest there be any misunderstanding, this is not to suggest that a plaintiff must always
`
`include information about the effect of an employment practice on the defendant’s workplace to
`
`plead the first element of disparate impact claim. Sometimes the disparate impact created by an
`
`employment practice will be self-evident, potentially obviating the need for further allegations
`
`about the resulting disparity. Take, for example, Jenkins v. New York Transit Authority. 646 F.
`
`Supp. 2d 464 (S.D.N.Y. 2009). There, the plaintiff’s religion required her to only wear skirts,
`
`2
`
`

`

`Case 3:20-cv-07499-VC Document 70 Filed 09/28/22 Page 3 of 4
`
`
`
`and she brought a disparate impact claim after she was fired from her job as a bus operator for
`
`refusing to wear pants instead. Id. at 467–68. The defendants argued that the plaintiff,
`
`notwithstanding her personal experience, had not stated a disparate impact claim because such a
`
`claim “requires an allegation that an employer’s policy has had a statistically significant,
`
`disproportionate, and negative effect on one protected group as compared to others.” Id. at 469.
`
`The court rejected this argument. Because it was “plain” that the policy would disparately affect
`
`women of the plaintiff’s religion, there was no need for statistical or other evidence about the
`
`results of that practice in the workplace. Id. at 470. One can imagine other policies that would,
`
`on their face, disparately affect a protected class, and a plaintiff would not need to point to any
`
`facts showing the resulting disparate impact to challenge those policies.
`
`Here, in contrast, Liu has plausibly alleged that racial bias could affect customer ratings
`
`in the rideshare industry, and so use of those ratings in Uber’s employment decisions could cause
`
`a disparate impact. But his allegations are not so strong as to create an inference that use of the
`
`ratings will always cause a disparate impact, in the same way that a ban on skirts will always
`
`cause a disparate impact for members of a religion that requires skirts. Therefore, Liu must
`
`include non-conclusory allegations about the impact of the challenged practice at the actual
`
`company he is suing. But his allegations about the effects of the practice at Uber are
`
`incoherent—even more so now than before.
`
`As stated in prior rulings, it is hardly fanciful to suspect that Uber’s practice of
`
`terminating drivers based on customer ratings negatively affects minority drivers. Nor, as
`
`previously noted, is a survey necessarily the only way a plaintiff in a case like this might satisfy
`
`their pleading burden with respect to the first element of a disparate impact claim. The problem
`
`is not with Liu’s theory, but with his failure to develop meaningful factual allegations in support
`
`of it (and with his counsel’s apparent failure to put thought or effort into the endeavor). Because
`
`the Court is not convinced that it would be impossible for Liu to state a disparate impact claim,
`
`he will be given one final chance to file an amended complaint. But if he fails to state a claim
`
`next time, dismissal will be with prejudice. Liu and his counsel have been given plenty of
`
`3
`
`

`

`Case 3:20-cv-07499-VC Document 70 Filed 09/28/22 Page 4 of 4
`
`
`
`chances—arguably more than enough—to state a claim. Leadsinger, Inc. v. BMG Music
`
`Publishing, 512 F.3d 522, 532 (9th Cir. 2008) (explaining that a district court may deny leave to
`
`amend after “repeated failure to cure deficiencies by amendments previously allowed”). Any
`
`amended complaint is due within 28 days of this order. If no amended complaint is filed by that
`
`date, dismissal of the third amended complaint will be with prejudice.
`
`IT IS SO ORDERED.
`
`Dated: September 28, 2022
`
`______________________________________
`VINCE CHHABRIA
`United States District Judge
`
`4
`
`