[IPAC-List] Ricci v DeStefano - Adverse Impact?

[Dettling, Aaron]

Lance, I'm afraid this is an over-reading of the Ricci decision.  The scotus in no way endorsed the 4/5 rule alone as the measure of adverse impact or rejected statistical significance as part of a plaintiff's burden of proof.  Rather, the Ricci plaintiffs admitted that the city was faced with a prima facie case of disparate impact (maybe wrongly); therefore, the Court's statements about a prima facie case of disparate impact are mere dicta.  Let's remember at the end of the day, the Court's ultimate ruling was that there was NOT a strong basis in evidence that a disparate impact claim against the City would have been successful, i.e., there was not a strong basis in evidence to conclude that african-american plaintiffs would have ultimately been able to sustain their burden of proof on AI AND that the city could not have carried its burden of establishing that the test was job related and consistent with business necessity.

Aaron L. Dettling
Balch & Bingham LLP
1901 Sixth Avenue North, Suite 1500
Birmingham, Alabama 35203-4642
(205) 226-8723 - Phone
(205) 488-5699 - Fax
Download vCard<blocked::http://www.balch.com/people/Vcard.aspx?Attorney=318>
www.balch.com<blocked::http://www.balch.com/>

IRS CIRCULAR 230 DISCLOSURE: Unless explicitly stated to the contrary, this communication (including any attachments) is not intended or written to be used, and cannot be used, for the purpose of (i) avoiding penalties under the Internal Revenue Code or (ii) promoting, marketing, or recommending to another party any transaction or matter addressed herein. Click here for more information.<blocked::http://www.balch.com/resources/publications/Circular230ClientLetter.pdf>

CONFIDENTIALITY NOTE: This email and any attachments may be confidential and protected by legal privilege. If you are not the intended recipient, be aware that any disclosure, copying, distribution or use of the e-mail or any attachment is prohibited. If you have received this email in error, please notify us immediately by replying to the sender and deleting this copy and the reply from your system. Thank you for your cooperation.



________________________________
From: ipac-list-bounces at ipacweb.org [mailto:ipac-list-bounces at ipacweb.org] On Behalf Of Lance Seberhagen
Sent: Thursday, July 09, 2009 6:59 PM
To: Pluta, Paul
Cc: IPAC-List at ipacweb.org
Subject: Re: [IPAC-List] Ricci v DeStefano - Adverse Impact?

What is the standard for adverse impact?  Is it (1) the 80% Rule, (2) statistical significance, (3) either one, or (4) both together?  The Supreme Court's majority decision in "Ricci" (page 27) was notable for its strong endorsement of EEOC's "80% Rule" from the Uniform Guidelines on Employee Selection Procedures as the standard (or perhaps merely "a standard") for determining adverse impact, regardless of statistical significance.  My analysis below focuses on blacks vs. whites, but the results would be the same for Hispanics vs. whites.

Lieutenant Exam

The New Haven Lieutenant exam had two major stages: (1) test and (2) certification.  On the test, the passing rate of blacks was 31.58% (6/19), compared to 58.14% (25/43) for whites.  Thus, the adverse impact was 54.32%, which is significant under the 80% Rule.  But if one applies a statistical significance test (Fisher's Exact Test) to the passing rates of the test, the probability is 9.71% (or 1.66 SD units), which is not statistically significant.  The next stage of the exam was certification.  The certification rate for blacks would have been 0.00% (0/6), compared to 40.00% (10/25) for whites, resulting in an adverse impact of 0.00%, which is significant under the 80% Rule.  But using Fisher's Exact Test, the probability of the certification stage would have been 14.11% (or 1.47 SD units), which is not statistically significant.  Only if one combines the test and certification stages ("overall selection") is there a statistically significant difference between blacks and whites.  The overall selection rate of blacks was 0.00% (0/19), compared to 23.26% (10/43) for whites, resulting in an adverse impact of 0.00%, which is significant under the 80% Rule.  Using Fisher's Exact Test, the probability of the overall selection rates would have been 2.43% (or 2.25 SD units), which is statistically significant.

Captain Exam

Nothing was statistically significant in the Captain exam.  The New Haven Captain exam had two major stages: (1) test and (2) certification.  On the test, the passing rate of blacks was 37.50% (3/8), compared to 64.00% (16/25) for whites.  Thus, the adverse impact was 58.59%, which is significant under the 80% Rule.  But if one applies a statistical significance test (Fisher's Exact Test) to the passing rates of the test, the probability is 23.79% (or 1.18 SD units), which is not statistically significant.  The next stage of the exam was certification.  The certification rate for blacks would have been 0.00% (0/3), compared to 43.75% (7/16) for whites, resulting in an adverse impact of 0.00%, which is significant under the 80% Rule.  But using Fisher's Exact Test, the probability of the certification stage would have been 26.32% (or 1.12 SD units), which is not statistically significant.  Looking at the test and certification combined, the overall selection rate of blacks was 0.00% (0/8), compared to 28.00% (7/25) for whites, resulting in an adverse impact of 0.00%, which is significant under the 80% Rule.  Using Fisher's Exact Test, the probability of the overall selection rates would have been 15.43% (or 1.42 SD units), which is not statistically significant.

Discussion

The Court ruled that both exams had adverse impact, even though there were no statistically significant differences in either the test or the certification components of either exam.  The overall selection (test + certification combined) rates had adverse impact on the Lieutenant exam but not on the Captain exam.  Thus, it looks as if adverse impact does not require both the 80% rule and statistical significance, and we're not too sure about statistical significance.

On the Lieutenant exam, there were no statistically significant differences on the test alone or certification alone, but the overall selection rates (test + certification combined) were statistically significant.   However, under Section 703k(1)(B)(ii) of Title VII, "If the respondent demonstrates that a specific employment practice does not cause the disparate impact, the respondent shall not be required to demonstrate that such practice is required by business necessity."  In "Ricci," if the City did not have to show the business necessity of either the test or certification, would the Title VII have required the City to show the business necessity of the overall selection process (i.e., test + certification combined)?  That issue would have been worthy of Supreme Court clarification, but I guess we'll have to wait for another case to get the answer to that one.

Another issue is whether the Supreme Court has now established the "80% Rule" as a firm 80% standard.  If the selection rate of one group is 85% of the rate of another group, but the difference is statistically significant or very large numbers of people are selected, is there adverse impact?  We'll have to wait for for another case to get the answer to that one, too.

Lance Seberhagen, Ph.D.
Seberhagen & Associates
9021 Trailridge Ct
Vienna, VA 22182
Tel 703-790-0796
www.seberhagen.com<http://www.seberhagen.com>



Pluta, Paul wrote:

The courts have already weighed-in on voluntary affirmative action programs.  However, there is no indication that the City of New Haven had a voluntary affirmative action program in place.  If it did, perhaps it would have developed a different selection plan to better achieve its goals.  Affirmative action plans are usually strategic and address how a manifest racial imbalance is to be corrected over a specified period of time.  A post hoc race-based decision is in no way an application of affirmative action: it is a knee-jerk reaction that was taken to achieve some non-specific goal.  Had the City and its employees acted more deliberately and consistently, they might not have looked like such buffoons and we wouldn't have all been put through this nonsense.

Don't even get me started with the strict application of the 4/5 rule with small sample sizes.  You would think that nobody ever heard of sampling error.  It is interesting that we also run into that problem in LA where whites are usually not the majority group; at least not a very large one.  With small sample sizes (i.e., small examinations) impact calculated using the 4/5 rule-of-thumb shifts across different groups from administration to administration.  However, if one looks at the results across time, the impact tends to disappear.  Fancy that!

Paul E. Pluta, MA, SPHR
Human Resources Analyst III
Los Angeles County Department of Human Resources
Workforce Planning, Test Research, & Appeals Division