The Selection on Air Traffic Control Specialists: Two Studies Demonstrating Methods to Insure an Accurate Validity Coefficient for Selection Devices.

摘要

There are several conditions that can influence the calculation of the statistical validity of a test battery such as that used to selected Air Traffic Control Specialists. Two conditions of prime importance to statistical validity are recruitment procedures and the accuracy of the data base. The recent edition (1978) of the Federal Uniform Guidelines on Employee Selection Procedures places considerable emphasis on recruitment practices and their effect on validity. In the first of two studies, Monte Carlo techniques were employed to demonstrate the frequently overlooked effect that recruitment procedures can have on the validity coefficient. It was shown how highly specific recruitment results in a more homogenous group of applicants which produces a small applicant group variance on the selection test scores. It was further shown how a small applicant group variance considerably reduces the validity coefficient when the coefficient is corrected for selection effects; commonly termed restriction in range. This paper suggests a procedure that eliminates this recruitment problem and results in compliance with the Uniform Guidelines. The second study describes a statistical procedure to use when it is necessary to eliminate erroneous data. The procedure employs the notions of statistical distance and probability to identify data that has an extremely small likelihood of belonging to the population of the remaining data set. (Author)