Opinion
No. 03 Civ. 2833 (SAS).
November 3, 2004
Martha M. McBrayer, Esq., Benedict P. Morelli Associates, P.C., New York, NY, for Plaintiffs.
Theresa Crotty, Corporation Counsel of the City of New York, New York, NY, for Defendants.
OPINION AND ORDER
Craig and Georgia Bonton, are suing individual employees of the New York City Administration of Child Services ("ACS") as well as the City of New York under 42 U.S.C. §§ 1981, 1983. The Bontons have demanded a jury trial. The Bontons allege that in July, 2001, the individual defendants sought and obtained a court order from the Family Court placing the Bontons' infant twins, Rosella and Craig, Jr., in foster care for approximately a year and three months in violation of the Bontons' civil rights and further allege that this violation is a consequence of ACS's policy or custom of singling out African-American families for such treatment. Defendants have moved to preclude the testimony of the Bontons' statistical expert, Dr. Harriet Zellner, who has provided an opinion to support the contention that ACS discriminates against African-Americans. For the reasons stated below, this proposed expert testimony is inadmissible and, therefore, must be precluded.
The Bontons also bring claims under the New York State Constitution's Bill of Rights as well as state law claims of false arrest, false imprisonment and negligence. See Complaint ¶¶ 63-106.
See id. at 1.
I. BACKGROUND
A. Dr. Zellner's Report
The Bontons propose to use the testimony of Dr. Harriet Zellner ("Zellner"), who holds a Ph.D. in economics from Columbia University, to assist in proving its Monell claim against the City. The purpose of the report is "to determine whether there were statistically significant disparities in 2000 and 2001 in the rate at which children of African American and white families investigated by [the] City of New York Administration for Children's Services . . . were remanded to ACS custody." Zellner based her report on two sets of data, both of which were supplied by ACS: "(1) the racial identity of parents investigated by ACS in 2000 and 2001 and (2) the number of children — by race — remanded to the custody of the agency." The report displays a selection of this data in a table, according to which 90.4% of all children remanded to ACS custody in 2000 were African-American even though only 79.8% of all cases involved African-Americans. By contrast, in the same year, White children accounted for 9.6% of all children remanded, while 20.2% of all cases involved White parents. The report displays similar data for 2001. Zellner uses Fisher's exact test to calculate the probability that the disparity in remand rates is due purely to chance. She determines that the likelihood is less than one in 1,000 and, consequently, concludes that "the observed disparity is highly significant statistically."
See Monell v. Department of Soc. Servs. of the City of New York, 436 U.S. 658, 694 (1978) (holding that a municipality may be sued under section 1983 only when a municipal policy or custom caused the plaintiff's injury).
See 4/1/04 Report of Dr. Harriet Zellner ("Zellner Report"), Ex. S to 10/8/04 Declaration of Theresa Crotty in Support of Defendants' Motion to Preclude Plaintiffs' Expert and to Bifurcate the Trial ("Crotty Decl.").
Id. at 2.
Id.
See Table 1, Zellner Report. In fact, the table omits data for parents whose race was identified by ACS as Hispanic or "Other," thus skewing upwards the percentages for both African-Americans and Whites. See 4/13/04 Report of Dr. Philip Bobko ("Bobko Report"), Ex. T to Crotty Decl. at 5-6. This omission has no effect, however, on the disparity in remand rates between African-Americans and Whites, which is the focus of Zellner's report as well as the Bontons' claim against the City.
See Table 1, Zellner Report. It should be noted that Plaintiffs' Memorandum of Law egregiously misstates the data contained in Dr. Zellner's report. According to the Bontons' attorney, David S. Ratner, "in 2000, only 9.6% of the White families investigated by ACS had their children remanded to ACS custody whereas 90.4% of the African-American families investigated by ACS had their children placed in foster care." Affirmation in Opposition to Defendant's Motion to Preclude the Testimony of Plaintiffs' Statistical Expert and To Bifurcate the Trial ("Pl. Opp.") at 5-6. According to the data in Zellner's table, the correct percentages are 2.7% with respect to Whites and 6.5% with respect to African-Americans. The question arises whether Mr. Ratner is intentionally trying to deceive the Court or if he simply cannot grasp the findings of his own expert's report.
See Table 1, Zellner Report.
Fisher's exact test is a calculation used by statisticians to determine the statistical significance of an observed value as compared to the prediction of a pre-established hypothesis. See David H. Kaye David A. Freedman, Reference Guide on Statistics, REFERENCE MANUAL ON SCIENTIFIC EVIDENCE 83, 164, 172-73 (Federal Judicial Center 2000).
See Zellner Report at 3.
Id. at 4.
Zellner acknowledges, however, a shortcoming in her analysis that bears on her conclusion:
The reader will note that — with only the data provided — we can not control statistically for any existing between-race differences in factors like family income or parents' employment status that might generally influence the ACS decision. In effect, our statistical analysis assumes that such differences were either non-existent or unimportant to ACS. The greater their actual importance to ACS decision-making . . . the less insight the simple analysis we have reported above can provide. . . . As regards the issues under analysis here, controls for variables such as family income, parents' education, parents' occupation, parents' employment status and the number of times ACS has investigated the family over the last several years would be very useful.
Id.
B. Dr. Philip Bobko's Response
The defendants have provided a response to Zellner's report by Dr. Philip Bobko ("Bobko"), who has a Ph.D. in economics and social statistics from Cornell University. While Bobko does not perform his own statistical analysis, he presents a number of criticisms of Zellner's report. First, he challenges the way Zellner presents the data in the table appended to her report. Second, Bobko asserts that Zellner states the result of her statistical significance test "in a potentially misleading and exaggerated manner." Third, Bobko contends that "counting problems" in regard to the number of cases referred to ACS during 2000 and 2001 affect the validity of Zellner's data. Fourth, and most significantly, Bobko argues that Zellner's analysis has little, if any, probative value because it fails to account for significant explanatory variables.
See Bobko Report at 2.
See id. at 5-6.
Id. at 6.
Id. at 9. According to Bobko, the numbers that Zellner considers as cases in fact refer to reports of child abuse or neglect. Bobko argues that because there may be several reports of child abuse involving the same family over the course of a given year, Zellner's data — as well as the resulting analysis — are flawed. See id. at 7-9.
See id. at 9-11. I do not discuss any of Bobko's first three criticisms as my analysis of the fourth criticism requires preclusion of the report.
II. LEGAL STANDARD
Rule 702 of the Federal Rule of Evidence states the following requirements for the admission of expert testimony into evidence:
If scientific, technical, or other specialized knowledge will assist the trier of fact to understand the evidence or to determine a fact in issue, a witness qualified as an expert by knowledge, skill, experience, training, or education, may testify thereto in the form of an opinion or otherwise, if (1) the testimony is based upon sufficient facts or data, (2) the testimony is the product of reliable principles and methods, and (3) the witness has applied the principles and methods reliably to the facts of the case.
A district court must act as "a gatekeeper to exclude invalid and unreliable expert testimony." Under Rule 702 and Daubert v. Merrell Dow Pharms., Inc., the trial judge must determine whether the proposed testimony "both rests on a reliable foundation and is relevant to the task at hand."
Bickerstaff v. Vassar College, 196 F.3d 435, 449 (2d Cir. 1999) (quotation omitted).
509 U.S. 579 (1993).
Id. at 597. See also Kumho Tire Co. v. Carmichael, 526 U.S. 137, 141 (1999) (extending Daubert's general holding setting forth the trial judge's gatekeeping obligation in regard to expert testimony based on scientific knowledge to testimony based on "technical" and "other specialized" knowledge).
In addition, Rule 403 states that relevant evidence "may be excluded if its probative value is substantially outweighed by the danger of unfair prejudice, confusion of the issues, or misleading the jury." "Expert evidence can be both powerful and quite misleading because of the difficulty in evaluating it. Because of this risk, the judge in weighing possible prejudice against probative force under Rule 403 . . . exercises more control over experts than over lay witnesses."
Fed.R.Evid. 403.
Daubert, 509 U.S. at 595 (quotation omitted).
The proponent of expert evidence must establish admissibility under Rule 104(a) of the Federal Rules of Evidence by a preponderance of the proof.
See Bourjaily v. United States, 483 U.S. 171, 175-76 (1987).
III. DISCUSSION
Zellner's report is inadmissible because it will not assist the trier of fact. The Court does not doubt, nor do defendants challenge, Zellner's qualifications as an expert in statistics. The Court also commends Zellner for acknowledging the limitations of her methodology and confining her conclusion accordingly. Nonetheless, I find that Zellner's conclusion — that the disparity in remand rates for African-Americans and Whites cannot be attributed to chance — will be of no assistance to a jury charged with determining whether it is ACS's policy or custom to place African-American children in foster care on account of their race. In the absence of any other evidence in the record suggesting that ACS followed a discriminatory policy, Zellner's testimony would invite the jury to engage in impermissible speculation.
See Bickerstaff, 196 F.3d at 448 (noting that in determining whether evidence is admissible for purpose of summary judgment motion, courts must "carefully distinguish between evidence that allows for a reasonable inference of discrimination and evidence that gives rise to mere speculation and conjecture").
There is no question that the data in Zellner's report are troubling. As noted earlier, in 2000, 6.5% of investigations of African-American families resulted in the children being placed in foster care, whereas the figure for White families is only 2.7%. That means African-American children are over twice as likely to be removed from their parents as the result of an investigation by ACS as are White children. This disparity points to a sad state of affairs in New York City. In and of itself, however, it does not say anything about whether ACS discriminates against African-Americans.
The plaintiffs argue that Zellner's testimony is relevant to their allegation that their civil rights were violated "as a consequence of ACS' routine practice of singling out African-American families for . . . discriminatory treatment." Plaintiffs contend that "Dr. Zellner's statistical analysis of ACS' discriminatory actions does have a tendency to make the fact that the Bontons were discriminated against [on account of race] more probable." This contention is incorrect.
Pl. Opp. at 7.
Id. at 8.
Zellner's own conclusion is that "the data strongly support the charge that African-American children were more likely than White children to be placed by ACS in foster care over the 2000-2001 period." Zellner expresses no opinion as to whether African-American children were more likely to be placed in foster care on account of their race. The distinction is critical. In order to succeed in their Monell claim against the City, the Bontons must show a causal link between the violation of their constitutional rights and an impermissibly race-based official municipal policy or custom. However, Zellner's statistical evidence does not make it any more or less likely that it is ACS's policy or custom to discriminate against African-Americans.
Zellner Report at 4-5.
Zellner's only opinion as to causation is that the disparity in remand rates cannot be attributed to chance. See id. at 4.
See Monell, 436 U.S. at 694.
The reason for this is that other factors may account for the disparity. Zellner readily acknowledges as much. She identifies a number of factors generally revolving around socio-economic status that might influence the outcome of an ACS investigation. Although she does not speculate on whether these factors are related to race, she does admit that controlling for such variables would be "very useful." Bobko goes farther, insisting that a correlation between race and potential socio-economic control variables is clear. He argues that such variables may explain the disparity in remand rates.
See Kaye Freedman, supra note 10, at 138 ("[T]he association between two variables may be driven largely by a `third variable' that has been omitted from the analysis. For an easy example, among school children, there is an association between shoe size and vocabulary. However, learning more words does not cause feet to get bigger, and swollen feet do not make children more articulate. In this case, the third variable is easy to spot — age.").
See Zellner Report at 4.
Id.
See Bobko Report at 11 n. 2 ("Even a cursory perusal of data readily available on the Internet indicates that potential control variables such as income or living below the poverty level are related to race in New York City.").
See id. at 10-11.
To determine whether there is causal link between race and the observed disparity in remand rates, it is necessary to conduct a multiple regression analysis to control for explanatory variables such as parents' income level or employment status. While Bobko does not demonstrate the effect of controlling for such variables by means of his own regression analysis, it is impossible for the Court to agree with Zellner's assumption that these factors are "unimportant" to an analysis of the outcomes of ACS investigations. As a result, for a jury to determine solely on the basis of Zellner's report that a causal link exists between the observed disparity and an alleged policy of racial discrimination would require a logical leap that amounts to mere speculation.
A multiple regression analysis is a statistical tool for determining the effect of two or more explanatory variables on a variable to be explained, called the dependent variable. This tool allows the expert to determine the causal relationship, if any, between the explanatory variables and the dependent variable. See Daniel L. Rubinfeld, Reference Guide on Multiple Regression, REFERENCE MANUAL ON SCIENTIFIC EVIDENCE 179, 181 (Federal Judicial Center 2000).
Cf. Sobel v. Yeshiva Univ., 839 F.2d 18, 34 (2d Cir. 1988) (holding that an attack on the omission of variables from a regression analysis need not consist of a competing regression, as long as the attack shows the relevance of the particular variables that ought to have been included).
Zellner Report at 4.
Courts have repeatedly held that statistical analyses that fail, as Zellner's does, to control for any nondiscriminatory explanations are inadmissible. Although multiple regression analysis is not a prerequisite for the admission of statistical reports in discrimination cases, the complexity of the social and economic forces at play in the etiology of child abuse and neglect necessitates the use of regression analysis in this instance. I conclude, therefore, that Zellner's proposed expert testimony would only confuse, rather than assist, the trier of fact.
See, e.g., Bickerstaff, 196 F.3d at 450 (holding that assumption that race bias tainted professor's course evaluation scores is untenable without attempting to control for other causes for low score); Smith v. Xerox Corp., 196 F.3d 358, 370-71 (2d Cir. 1999) (holding that plaintiff's statistical analysis failed on its own to support an inference of discriminatory treatment sufficient to withstand a summary judgment motion because the analysis did not account for any other causes for the fact that older workers were more likely to be terminated); Hollander v. American Cyanamid Co., 172 F.3d 192, 203 (2d Cir. 1999) (holding that expert report is inadmissible because its "inference of [age] discrimination solely on the basis of the raw numbers is impermissible in the absence of any attempt to account for other causes of the . . . anomaly"); Raskin v. Wyatt Co., 125 F.3d 55, 67-68 (2d Cir. 1997) (holding that expert report was inadmissible in part because it "assume[d] any anomalies in the . . . data must be caused by age discrimination, and [made] no attempt to account for other possible causes"). See also Bazemore v. Friday, 478 U.S. 385, 400 n. 10 (1986) (stating, in dicta, that "[t]here may, of course, be some regressions so incomplete as to be inadmissible as irrelevant").
See EEOC v. Venator Group, No. 99 Civ. 4758, 2002 WL 181771, at *2 (S.D.N.Y. Feb. 5, 2002) (reminding that "statistics come in infinite varieties and their usefulness depends on the surrounding facts and circumstances").
Finally, I reject the plaintiffs' contention that Zellner's failure to control for any other factors should be laid at defendants' feet. According to the plaintiffs' affirmation opposing the motion to preclude, "when the plaintiffs requested information beyond the basic racial breakdown, per the plaintiff's statistical expert, the defendants claimed that producing such documents `would be significantly more complicated.'" This statement is misleading. Plaintiffs applied to the court to compel production of data needed to produce a statistical analysis only days before the close of discovery. That letter to the Court did not specify, however, what kind of data was required. At a follow-up conference before Magistrate Judge Dolinger, plaintiffs revealed that they had not yet consulted a statistical expert to determine what data would be needed aside from a racial breakdown of ACS investigations and of cases where children were placed in foster care. Plaintiffs subsequently produced an affidavit from Zellner, who indicated that additional demographic information would be "very useful" to her analysis. Judge Dolinger declined to compel defendants to produce this additional information "since plaintiffs had never requested such production from defendants and the deadline for discovery [had] passed." Plaintiffs have only themselves to blame for the unavailability of the data required to conduct a regression analysis.
Pl. Opp. at 13.
See Bonton v. City of New York, No. 03 Civ. 2833 (S.D.N.Y. March 11, 2004) (Dolinger, J.), Ex. A to Pl. Opp. at 1 ("3/11/04 Order").
See 2/27/04 Letter from Ratner to the Court, Ex. C to Pl. Opp., at 1.
See 3/11/04 Order at 2.
3/10/04 Affidavit of Dr. Harriet Zellner, Ex. B to Pl. Opp., at 4.
3/11/04 Order at 3.
IV. CONCLUSION
For the foregoing reasons, the defendants' motion to preclude Dr. Zellner's expert testimony is granted. This opinion does not address defendants' motion to bifurcate the claims against the individual defendants from the Monell claim asserted against the City.
SO ORDERED.