Racial profiling makes little sense, mathematically speaking. Using statistical analyses, William Press of the University of Texas at Austin has found that choosing people to screen based on ethnicity is no more effective than random checks, because nonterrorists vastly outnumber terrorists. The optimal way to screen would be to use “square-root-biased sampling,” so that someone nine times as likely to be a terrorist as the average traveler would be screened three times more frequently. This approach would turn up more terrorists in part by avoiding the repeated screening of the same innocent people who fit the profile. But because that strategy would be difficult to implement, Press says that mathematically, the more sensible method is not to profile at all. The study appears in the February 10 Proceedings of the National Academy of Sciences USA.
This article was originally published with the title "Math against Profiling" in Scientific American 300, 4, 31 (April 2009)