HUMAN FACTORS PREDICTION VALIDATION.
Abstract
This study concerned the predictive accuracy of personnel and training projections research conducted during new development cycles, and the development of an administrative tool for decision-making processes associated with allocating resources to future human factors research. Two measures of projection accuracy, termed Regret and Discrepancy were formulated. Data on a sample of projections made during the 5-year period 1959 to 1964 were collected. These data were analyzed to provide measures of accuracy (Discrepancy and Regret) and enumerate various possible predictor variables. Limitations within the data prevented collection of Regret scores. A validity index (to predict Discrepancy) was derived using curve fitting and multiple linear regression. Four important factors - types of projections, point-in-time (within the RDT and E cycle), system size, and number of replications of the research - were discovered. A final multiple regression formula (validity index) based on these four factors was produced and found to have a correlation of .66 with measured Discrepancy. Suggestions for refinement and utilization of the validity index, for further work with the Regret model of the index, and for improvement of reports of future human factors projections are included. (Author)
Document Details
- Document Type
- Technical Report
- Publication Date
- Aug 01, 1967
- Accession Number
- AD0819210
Entities
People
- Lawrence E. Langdon
- Richard S. Gebhard
- Robert A. Wood
- Victor J. Hernandes