The primary purpose of this study was to determine if self-instructional data analysis sessions could be used effectively as traditional college laboratories to reinforce students' understanding of certain principles of mechanics. The secondary purpose was to determine whether any of the 140 college physics students involved had a dependence on the mode of instruction based on their literary and mechanical scores on the Kuder Preference Record. The self-instructional material was presented in two modes: audio-mechanical and literary. The self-instructional sessions used no physics apparatus. The control session was conducted with short oral presentation, laboratory session, and directed activities. The Latin Square and two-factor mixed design were the statistical analyses used. Results showed that the students used statistically less time in the audio-mechanical and literary modes than in the control mode for two of the three experiments. Statistically equal amounts of time in the three modes were used for the third experiment. No significant statistical interaction between the students' preference of mode and Kuder literary and mechanical difference scores was evidenced. It was concluded that self-instructional data analysis sessions were effective and could be beneficial. (Author/EB)