The generalizability of earlier findings on training objectives for military specialties was studied. Thirteen tasks or jobs, involving mechanical, electrical, general, or administrative aptitude, were rated by senior noncommissioned officers on: recommended field training emphaisi; present school emphasis; consequences of inadequate job performance; delay tolerance; and task difficulty. Occupational survey data--concerning percent of members performing each task, percent of time s pent on task, learning difficulty, and average grade of task performers--were also made available. Policy equations were used as models to "capture" the rankings of the judges' recommended training emphasis, and to validate the training emphasis scale. Factor and job analysis indicated that the two policies used yielded reliabilities of .72 and .64. Interrater agreement was also found to be related to predictive efficiency of the equations. Only a moderate number of raters (about 40) were needed to provide stable data, and the performance criterion used in this study was found to be reliable and valid. Further research is recommended for those complex specialties which had low predictive efficiency and poor interrater agreement. (GDC)