The problem of preselection in weighted application blank studies

Abstract

The problem of preselection in predictions based on application blanks was studied on the basis of data pertaining to 291 applicants for clerical jobs in Prudential's Western Home office in Los Angeles. "Among persons actually hired, only five of the 19 items were found to discriminate between terminated and nonterminated employees at or beyond the .05 level of confidence. At least one of the 10 preselection items was found to be predictive of turnover at the .001 level… . Three possibilities suggest themselves: 1… . allow preselection to operate prior to utilizing the weights. 2. Apply 'restriction in range' corrections to individual item validity co-efficients, where assumptions can be met… . 3. Develop 'preselection weights,' based upon differences between those hired and those rejected." (PsycInfo Database Record (c) 2025 APA, all rights reserved)