Application of Lightfoot's Cluster Evaluation System to Current Problems in Army Occupational Analysis
Final rept. Sep 95-Feb 97
HUMAN RESOURCES RESEARCH ORGANIZATION ALEXANDRIA VA
Pagination or Media Count:
The study objective was to build a prototype cluster structure validation methodology and to test it in a population data base of Army military occupational specialties. We developed a cross validation and internal validity CVIV procedure for estimating the cluster structures of empirical data bases. The major contributions of the CVIV procedure are that it can be used with many different types of empirical data and includes a statistical approach for identifying optimal cluster structure. We validated the CVIV procedure through an experimental design that allowed us to analyze the properties of the statistical test in terms of Type I error rate, power, and precision. The results provide strong support for the validity and utility of the CVIV procedure for estimating population cluster structure from sample data. First, the statistical test preserved the Type I error rate of .05. Second, the power of the test ranged between 86 percent and 100 percent across sample sizes. Third, 63 percent of the sample results matched the cluster structure of the Army job population data base. The CVIV procedure has wide application for the analysis of cluster structures in a range of data bases in both research and applied settings across the social and physical sciences.
- Military Forces and Organizations