ABSTRACT
The overall design of the Project A/Career Force program was intended to be fundamentally different from the conventional paradigm that dominated personnel research from 1906 to 1982. In 1906, Hugo Münsterberg was credited with conducting the first personnel selection research study when he attempted to evaluate the validity of a new procedure for selecting streetcar operators for the Boston transit authority. For a sample of streetcar operators, a new test of psychomotor ability was correlated with criterion measures of performance and yielded a significant relationship (Münsterberg, 1913). With this study, the classic prediction model was born and it has dominated personnel research ever since (Campbell, 1990). The modus operandi became the estimation of the Pearson correlation co-efficient when a single predictor score, or a single predictor composite score, and a single criterion measure of performance were obtained for a sample of job incumbents and the bivariate normal model was imposed on the distribution. Literally thousands of such estimates have been generated during this century (e.g., Ghiselli, 1973; Hunter & Hunter, 1984; Nathan & Alexander, 1988; Schmidt, 1988; Schmidt, Ones, & Hunter, 1992; Schmitt, Gooding, Noe, & Kirsch, 1984).
