MEASUREMENT AND PREDICTION OF COGNITIVE LOADINGS IN CORRECTIVE MAINTENANCE TASKS: II. BAYESIAN ANALYSIS OF A SAMPLE OF CLASS B ELECTRONICS TECHNICIANS' BEHAVIORS.
Abstract
Thirty-six advanced Navy electronics technicians (Class B Electronics School students) were given a symptom-malfunction (S-M) matrix completion test on a blocking oscillator circuit. Next, via a card-simulation format, each technician attempted to solve six troubleshooting problems in the same circuit; records were kept of each voltage and resistance reading made and of each component replacement choice. After the troubleshooting session, the subjects took a retest on the S-M matrix completion test. Using the technician's S-M matrix values, a Bayesian computation was performed for each performance step; this computation yielded Bayesian likelihoods for each replaceable component in the circuit. The Class B technicians showed notably superior performance to an earlier Class A technician sample in terms of troubleshooting time, steps to solution, and the number of correct solutions. The retest showed that as a technician works on search problems in this oscillator, he improves his original subjective S-M matrix on that circuit and appears to learn as a consequence of troubleshooting. The advanced technicians were not appreciably more Bayesian-like in component-replacement choices (about 55% resembled Bayesian performances compared to 52%) than the earlier Class A technician sample. The earlier finding, that S-M matrix quality is related to Bayesian-like component-replacement choices, and several competence indicators were confirmed in this study. (Author)
Document Details
- Document Type
- Technical Report
- Publication Date
- Apr 01, 1967
- Accession Number
- AD0652363
Entities
People
- Anthony K. Mason
- Douglas M. Towne
- Joseph W. Rigney
- Nicholas A. Bond Jr.
- Robert H. Cremer
Organizations
- University of Southern California