Computerized support of the pretrial confinement decision-making process in the Marine Corps

Abstract

This thesis explored the feasibility of developing an expert system to assist Marine Commanding Officers in making accurate pretrail confinement decisions. The predictive value of a number of factors was investigated, based on interviews with those who make or review pretrail confinement decisions regularly and on a preliminary study of eight battalions at Camp Pendelton, California. The study revealed eight factors with predictive value. Rank, prior unauthorized absence, year of service, age, receipt of unit awards and position page 11 entries manifested the types of association expected, based on the interviews. Seriousness of the charges showed a surprising inverse association with unauthorized absence, while GCT score suggested that those of average intelligence were less inclined to flee than those of either above average or below average intelligence. Findings were incorporated into a first-version prototype of the Pretrail Confinement Advisor (PCA) expert system using the M.1 Knowledge System Software Tool. Procedures for refinement and field testing of the prototype were recommended.