Relative Merits of Several Missing Data Estimators in Personnel Selection Procedures

摘要

The primary criterion for selection of recipients of Army ROTC scholarships in an applicant's rank on an order-of-merit list based on his Whole Man Score (WMS). The WMS is a linear combination of variables believed useful as predictors of criteria such as academic success, aptitude for military science, leadership potential, and predisposition toward a military career. Four predictors are used: (1) high school class standing score; (2) Scholastic Aptitude Test (SAT) or American College Test (ACT) score; (3) extracurricular, athletic, and leadership activity score; and (4) personal interview score. The predictors are assigned arbitrary weights of 3:3:3:1, respectively (TRADOC Circular 145-5, 1975). This paper documents the psychometric rationale underlying the method of computing the high school class standing portion of the WMS. The method involves adjusting for class size by converting class rank to percentile rank. Percentile rank is then transformed to a normalized standard score by introducing an a priori distributional assumption. Several alternative methods for estimating missing class standing scores are compared.