COMPARISON OF CLASSIFICATION STRATEGIES BY COMPUTER SIMULATION METHODS.
Abstract
The effectiveness of several methods of classifying enlisted men for A-school training were compared using computer simulation methods. Using actual school quotas, nine different assignment methods were compared on seven assignment criteria using computer simulation methods with a sample of 587 enlisted men from NTC, San Diego. Of these men, 251 were assigned to A-school, and the remainder were assigned to the fleet. The seven criteria were: (1) aptitude test selector sums, (2) predicted final school grade, (3) probability of graduation, (4) training cost (without pay and allowances), (5) training cost with pay and allowances, (6) manning level ratio, and (7) criticality. Seven assignment strategies were used to maximize each of the above criteria. Two additional strategies were: (1) actual NTC assignment, and (2) random assignment. Random assignment was included as a base against which to evaluate improvement over chance. The average payoff on any classification criterion is best for the strategy directed toward that criterion. The present operational classification strategy is far superior to random assignment. The best overall strategy, under the conditions of the present study, is assignment to maximize the probability of graduation from A-school. This strategy comes close to maximizing the other strategies investigated, and also results in relatively homogeneous talent groupings in the various schools. Classification to maximize probability of graduation is a strategy worth further investigation. (Author)
Document Details
- Document Type
- Technical Report
- Publication Date
- Jun 01, 1968
- Accession Number
- AD0674370
Entities
People
- Edward F. Alf
- John H. Wolfe
Organizations
- Naval Health Research Center