Selecting Optimal Training Programs in a Hospital for the Mentally Retarded

This paper describes the development and application of a quantitative model to aid in selecting optimal training programs for mentally retarded children in a California state hospital. The model is a constrained version of R.A. Howard's Markov decision model. In this application five training programs are compared on a benefit-less-cost measure of performance. Tradeoffs between programs are found in terms of net benefit and annual cost to the hospital. The model is found to be additionally useful in evaluating contemplated new programs or modifications to existing programs where only their expected effects may be known.