Effects of search strategies on fault diagnosis performance

This paper investigates the effects of two diagnostic search strategies (decision table search and hypothesis-and-test search) on fault diagnosis performance, and whether cognitive styles could mediate these effects. A simulated nuclear power plant main control room prototype was employed as a test case for fault diagnosis experiments. Performance was measured by diagnosis accuracy, completion time, frequency of display switching occurrences, mental workload, and self-reported confidence level in the diagnosis task. Results revealed a significant effect of search strategy on the number of display switches, with significantly fewer occurrences for the group using the decision table search strategy as compared to that of the hypothesis-and-test search strategy group. No significant effect of search strategy was observed on the other diagnosis performance measures after controlling for possible effects of cognitive styles. The results suggest a potential advantage of using a decision table search strategy for fault diagnosis given its smaller associated magnitude of display switching.