NCJ Number
109346
Journal
Security Management Volume: 32 Issue: 2 Dated: (February 1988) Pages: 75-80
Date Published
1988
Length
6 pages
Annotation
This article offers guidelines to help security managers review psychological or honesty tests used to screen job applicants, with attention to evaluating research data supporting the tests.
Abstract
To make sure a test can withstand outside scrutiny by qualified experts, the security manager must first evaluate credentials of the vendor and test developer. Determining whether the test's author is a certified psychologist with a substantive publication record is especially important. The article provides 12 guidelines for evaluating a psychological test. It cautions against tests for which little or no validation research exists, ones that claim they can only replace the polygraph, and tests not based on the prediction model of validation. The guidelines also say that managers should be wary of studies that do not tell how many people were incorrectly predicted to have job problems and ones claiming to predict dangerous or violent behavior. Other signs of questionable validation research are studies that used small numbers of subjects, were not cross-validated, and did not use real job candidates as subjects. The article emphasizes that poor tests increase a company's liability.