Validation and comparison of homogeneous and occupational interest scales

Abstract

Occupational interest scales and homogeneous scales were developed and compared in a double cross-validation design. A sample of 3,072 Air Force personnel was randomly divided for study purposes. Homogeneous scales were developed in each half-sample using a backward item selection procedure. Occupational scales were developed by selecting items which discriminated satisfied personnel within a career field from a "men-in-general" group. A Bayesian classification procedure was then used to cross-validate the scales developed in each half-sample. The results indicated that the occupational scales were slightly superior in correctly classifying satisfied personnel versus men-in-general. However, correlations between all interest scales and job satisfaction indicated generally higher validity for relevant homogeneous scales than for occupational scales. It was concluded that where prediction of group membership or occupational choice is desired, occupational keying should probably be the preferred approach. However, when concern is with a broader range of criteria, homogeneous scales may offer more general utility.