Sensitivity Analysis of Selected Variables in the CEM.
Abstract
The designed fractional factorial experiment is presented for a five-factor experiment with each factor having two levels. The advantages of factorial experimentation over the old classical one-factor-at-a-time experimentation are briefly discussed. A computer-simulated experiment was conducted according to the design using the CONAF (Conceptual Design for the Army in the Field) Evaluation Model (CEM) and the CONAF II data deck. Results of the analyses are presented on the sensitivity of four model output variables: (1) Forward Edge of Battle Area (FEBA), (2) Blue State, (3) Red State, and (4) Blue Casualties, to change in five input factors: (1) Boundary Shift Limit, (2) Prepared Defense Threshold, (3) Indices of Firepower Potential, (4) Red Replacement Policy, and (5) Estimated Outcome Thresholds. Analyses are performed on the output data at four periods of game play (12, 24, 36, and 48 days of battle). None of the four output variables were found to be sensitive to change in the input factor Boundary Shift Limit. All four output variables, however, are sensitive to change in the Red Replacement Policy.
Document Details
- Document Type
- Technical Report
- Publication Date
- Jan 01, 1974
- Accession Number
- AD0780807
Entities
People
- Carl B. Bates
- Jerry Thomas
Organizations
- Center for Army Analysis