Placement of prompts, length of task, and level of retardation in learning complex assembly tasks.

Vocational rehabilitation clients were divided into two retardation groups (mildly and moderately mentally retarded). Each subject was taught four different assembly tasks (lawn mower engine, bicycle brake, carburetor, and drill), each utilizing one of four different teaching methods (short task--preresponse prompting, long task--preresponse prompting, short task--error-correction prompting, and long task--error-correction prompting). Modeling of correct selection and placement of parts was used either before the subject's response or after an error had been made. Preresponse prompts yielded fewer errors than did error-correction prompts in training trials. Training by error-correction, however, was better than preresponse prompting for producing fewer errors on test (probe) trials for the moderately retarded subjects. Mildly retarded subjects performed equally well in both prompting conditions on test trials. Prompting conditions and level of retardation did not interact with length of the tasks. Results were discussed in terms of modification of the three-term contingency and covert-stimulus comparison.