This third of three volumes reports on analytic procedures conducted to address various aspects of the scalar properties of the Device Effectiveness Forecasting Technique (DEFT). DEFT, a series of microcomputer programs applied to data gathered from rating scales, is used to evaluate simulator devices used in U.S. Army weapons training. The analytic procedures included Monte Carlo simulations to assess the interpretation of DEFT data output; sensitivity of DEFT parameters; comparison of outputs; stability; and interrater agreement. Results indicated that it would be necessary to include assumptions regarding expected distributions of input variables in order to meaningfully interpret DEFT output. Also, the Monte Carlo analyses demonstrated the sensitivity of output scores to variations in inputs, and assessed the effects of various assumptions regarding measurement error on output scores. Interrater reliability was high for all three DEFT levels on ratings of three actual training devices, using several different raters. (GDC)