Theoretical considerations suggest that the differential effectiveness of teaching models and associated feedback procedures stems from their distinctive cueing properties. This led to the development of three treatment conditions which may be labeled "rating" (rehearsal of key discriminations), "observation" vicarious reinforcement), and "direct practice." Specific predictions were that increases in the response strength of probing (the dependent variable) would be greatest for the Direct Practice Group, next the Observer Group, and finally the Rater Group, and that within groups, Rater subjects would demonstrate the most significant within-repertoire shifts of sub-skills. Forty undergraduate teacher trainees were randomly assigned to three experimental and one control group. All were pre- and posttested in 5-minute video taped lessons which they taught for four fifth grade pupils. Each subject spent a total of 75 minutes in three treatment sessions over a 3-week period. Two trained raters coded relevant behaviors. Data treatment included analyses of variance and covariance and t tests. The first hypothesis was supported; the second was not although data was suggestive. Results provide further support for the theoretical assumptions. By increasing the distinctiveness of relevant model stimuli, observational learning is facilitated. (JS)