Rate only poor tests in rule-based aiding: full quality ratings were counterproductive¶

Strategy · All strategies
Evidence · no claims cited

Description¶

The rule-based model enabled a form of computer-aided training in which subjects received immediate feedback after each test about the quality of the rules the model inferred they were using. Evaluation within Experiment Six concluded that "rulebased aidingwascounterproductivebecausesubjectstendedto misinterpret the quality ratings theirtests received." The article therefore suggests that ratings indicating unnecessary or otherwise poor tests might be helpful, while full quality ratings should not be given.

Design Implications¶

Context¶

Requirements¶

• A rule-based model capable of inferring which rules a subject is using after each test.

Constraints¶

• Evaluated only within Experiment Six with FAA trainees transferring to the FAULT task; full quality ratings were found counterproductive in that setting.

Target Learners¶

• maintenance trainees learning troubleshooting

Target Learning Goals¶

• efficient, low-cost fault diagnosis test selection

Affordances¶

• Rule Based Heuristic Model Fault Diagnosis

Related Strategies¶

Examples¶

-

Key Sources¶

• Rouse, W. B., Rouse, S. H., Hunt, R. M., Johnson, W. B., and Pelligrino, S. J. (1980). Human Decision-Making in Computer-Aided Fault Diagnosis. Technical Report 434, Coordinated Science Laboratory, University of Illinois. https://eric.ed.gov/?id=ED192743