Social science and minority "set-asides."

Public Interest, Wntr, 1993 by George R. Lanque

Anecdotal data is typically used to make a case, rather than analyze a problem. Rarely is any attention paid to such social science fundamentals as sample construction, response rate, and the replication of results. In the $400,000 Dade County study, for example, both questionnaires and interviews were used to gather anecdotes to confirm discrimination in the local construction industry; but only 15 percent of the questionnaires were returned and only black businesses were interviewed. When a white-owned construction company sued the county over its MBE program (a case in which I served as an expert witness for the plaintiffs), it was disclosed at trial that the questionnaires had been destroyed and that assurances of confidentiality had been given those interviewed so that no third party could examine the unedited data. Andrew Brimmer, who conducted the Dade County study as well as the one in Atlanta, also had to concede that he did not have enough information about norms in the contracting industry to know whether anecdotes about slow payment, late notice of subcontracting opportunities, and other difficulties were evidence of racially disparate treatment or were common to all small businesses. Finally, he had to admit that he had no way of verifying whether any of the anecdotes were actually true.

Anecdotal data appear to be carefully selected to confirm the studies' arguments. In Milwaukee, for example, 68 percent of Hispanic contractors and 60 percent of women contractors surveyed did not claim that discrimination affected their businesses. But the study quoted only those owners who did allege discrimination. Sometimes the authors of studies are forced to add considerable interpretation to obtain the desired conclusion. The Seattle study confessed that M/WBEs were "reluctant to reach general conclusions absent direct evidence of bias, which is seldom available." Consequently the study's authors decided that their own interpretation of events would be authoritative in deciding if an incident was discriminatory.

Anecdotal approaches are popular because their outcomes can be easily controlled. But they are not always very persuasive in the courtroom. In 1991, the Ninth Circuit remanded an MBE case from King County, Washington, back to the district court, with instructions to focus on statistics rather than the boilerplate affidavits Seattle officials had gathered.

Statistical approaches

Statistical proof of discrimination is complex. Reasonable people can disagree about the proper data and tests. Justice O'Connor's opinion in Croson focuses on the importance of finding a statistical disparity between the availability and the utilization of qualified minority contractors willing and able to perform a service. How should this availability be measured? At a minimum, the availability data should take into account the actual geographical area from which the jurisdiction's public contractors are drawn. The data should also be broken down by type of market, and by the various groups the M/WBE program covers. Discrimination against Asians in heavy construction, for example, does not prove that other minorities suffer discrimination in heavy construction. Nor does it prove that Asians suffer discrimination in other markets. Finally, the analysis should take into account Croson's "willing and able" language by considering the capacity of firms to take on jobs of differing sizes. Large firms are better able to bid for large contracts. In the construction market, just 12 percent of the nation's firms have five or more employees, yet their work accounts for 78 percent of total revenues. Since M/WBEs are generally newer and smaller than non-M/WBEs, a failure to take the size of firms into account will exaggerate the availability of M/WBEs.