Testing Aptitude for Computer Programming

Abstract

This study explored the usefulness of a simulated work sample of programming (the Robot Test developed by the Bureau of the Census) as a criterion for validating programmer selection tests. In two samples of Civil Service employees in the United States Public Health Service, this type of criterion appeared quite predictable, with multiple correlations of .60 ( N = 135) and .71 ( N = 118). The best predictors of those studied were Parts 2 and 3 of the Programmer Aptitude Test, a frequently used selection test for programming. In one sample, the addition of the Numerical part of the Federal Service Entrance Examination significantly improved prediction of the criterion. Further research on the Robot Test is recommended.