The degree of transfer of technical training to workplace behavior was evaluated using a Solomon Four-Group experimental design. Additionally, all groups received retrospective pretests. Subjects were 103 technicians in an electronics company. Supervisors rated technicians on behaviorally anchored rating scales which were developed and labeled as behavior description scales for simplicity. Analysis of variance revealed no effect for training nor pretest. A training-pretest interaction effect was revealed for one dimension (Communication with Support Groups). Analysis of covariance revealed main effects for pretesting for two dimensions (Problem Solving and Communication with Supervisor) and a pretest-training interaction-for one dimension (Problem Solving). Except for one dimension, t tests revealed no significant differences between traditional pretests and retrospective pretests, thus negating a hypothesized response shift bias.