PREDICTING THE POTENTIAL FOR ACTIVE DUTY SUCCESS OF REHABILITATED AIR FORCE PRISONERS
PERSONNEL RESEARCH LAB LACKLAND AFB TX LACKLAND AFB
Pagination or Media Count:
The report documents the progress in developing and validating a prediction device for use in aiding decisions to return to active duty or discharge Air Force prisoners sent to the Retraining Group. First, there is an extensive review of the methodology and results of efforts to predict delinquency, recidivism, and military unsuitability. Then, two multiple regression analyses made on a sample of 1,303 former retrainees are reported. Each of the analyses yielded encouraging results in an initial cross-validation on 138 more recent retrainee cases for which actual criterion data were available. The cross-validation procedure was limited to making predictions on only 71 cases where the value of the multiple regression predicted score was sufficiently high or low to assure satisfactory accuracy. The best of the two regression equations, a 13-predictor system, was 77.4 per cent accurate in predicting successful return to duty and 72.5 per cent accurate in predicting unsuccessful return to duty. Details for applications of such a system, once adequately validated, to the operational decision-making process of the Retraining Group are given.