Racially Gerrymandered Police Tests

Linda S. Gottfredson

While President Clinton tries to take credit for putting 100,000 cops on the streets, his Justice Department is quietly pressing for policies that will drastically lower the quality of police throughout the nation. Under Assistant Attorney General Deval Patrick, Justice's Civil Rights Division is trying to force state and local governments to adopt a new police entrance examination that allows nearly equal percentages of black and white applicants to pass--and that achieves this result by stripping the test of crucially important reading, reasoning and judgment skills.

The new test arose out of a 1990 consent decree in which the Justice Department and the Nassau County, N.Y., Police Department agreed jointly to "design, develop and validate" a new test for would-be officers that "either does not have adverse impact upon blacks, Hispanics and females, or has been validated"--shown to be job-related. Four years and millions of dollars later, the county administered its new test. While the proportion of blacks who passed was nearly the same as for whites, the results make clear that the new examination had been created to avoid testing for job-related skills.

Many exceptionally well-qualified candidates received very low or failing scores. Some had years of experience as probation officers or as cops in other jurisdictions; others had or were pursuing graduate degrees in law or criminal justice. A look at those who passed the test is even more disturbing. According to close observers, a high proportion of top scorers not only have poor academic records, but also have outstanding arrest warrants, are unable to account for years of their work history or refuse to take a drug test.

To understand how this happened, you have to wade through a nearly impenetrable five-inch-thick July 1995 report in which the technical consultants hired by the Justice Department and the county describe how they developed and validated the test. The consultants were well aware that they faced a virtually impossible task: developing a test that measures important job skills but also produces proportional hiring despite big gaps in such skills between blacks and whites when they leave school. How these experts set about this task is a case study in racially gerrymandering the content of a test.

The report begins by explaining that experienced county cops rated "reasoning, judgment and inferential thinking" as crucial for good police work. It concludes by recommending a set of tests that measures none of these skills. Of the 25 tests the county administered in its day-long 1994 battery, the consultants decided that the scores from onlyeight of them- all personality tests--should be used in ranking candidates. These measured such qualities as "responsibility," "nondelinquency" and "openness to experience." The actual test questions are secret, but the report gives the following two examples:

1. When you were in high school, you were a member of a sports team.
A. Yes
B. No

2. Which of the two statements is most like you--A or B?
A. I'm always in a hurry at work to get things done.
B. At work, I think of myself as part of a smooth running machine.

The final test battery--the subset of the 1994 test that was actually used to determine who was eligible to be hired--replaced scores on a reading test with a pass/fail grade: Candidates merely had to read as well as the worst readers--the bottom 1%--among incumbent police officers.

The consultants' first step in denuding the test had been to exclude from consideration all traditional cognitive tests, despite their superior value in predicting job performance. Instead, they created their own "innovative" tests of judgment and reading comprehension. Mostly administered by video, these substitute tests seemed designed to reduce demand for skills. For example, the reading test used passages that applicants had available for study up to 30 days before the exam.

The consultants' second step was to test their exams for "disparate impact"--racial discrepancies in scores--before testing their ability to predict job performance. Normally, a test battery is first tried on a sample of experienced officers to determine which of its component tests best predict performance. Then, those that are most predictive are administered to applicants, at which point disparate impact can be assessed. In this case, the consultants threw out all the tests with substantial disparate impact--purportedly because they failed to predict job performance. Among those thrown out were all the "innovative" substitutes for traditional cognitive tests.

The consultants concealed the standard statistics that would have disclosed what they were doing. The most important data they kept secret were "zero-order validities," the raw correlations between test performance and job performance, which are required by federal guidelines and professional standards. Instead, they used improper correlations and gave them a patina of authority by making up a similar name--"simple validities." The consultants then committed a series of statistical errors that inflated the final battery's apparent value by over 100%, boosting it into the acceptable range for police exams.

Given the lack of appropriate data in the report, there is no way to know whether the Nassau County test works any better at selecting qualified police officers than simply flipping a coin. Yet John M. Gadzichowski, a Justice Department lawyer who oversaw the test's development, has testified, "I think it's beyond question that the exam . . . is valid. It predicts job success."

Meanwhile, the test provides the Justice Department with a new tool for coercing police departments into proportional hiring. The New Jersey State Police and the Suffolk County, N.Y., Police Department were among the first to feel the heat this year. The former has been sued by the NAACP and the latter subjected to a heavy-handed "compliance review" by the Justice Department. The alleged disparate impact of their exams is a focus of complaint in both cases, and both departments have been told that they might end their new legal troubles by adopting the Nassau County test.

The Nassau County test destroys what it purports to embody--merit hiring. In the name of fairness, it removes competence as an advantage, denying job opportunities to talented individuals of all races. Thus the quality of policing can be expected to fall. This new form of racial preferences poses a clear and present danger to public safety.