Data from three separate experiments were employed. College undergraduates were taught laws of logarithms by means of the same lesson material, and received the same twenty-item posttest after they were instructed under one of three learner control conditions. In the complete learner control condition, students chose both the sequence and number of primary presentation forms displayed to them. In the partial learner control condition, students chose how many instances were displayed to them. In the no learner control condition, students received a fixed sequence and number of primary presentation forms. Stepwise and all possible regressions routines were run using six basic presentation characteristics and various combinations, powers, and interactions of these as predictors of student posttest performance. There were no significant posttest score differences between the three learner control conditions. The predictive equations produced for the data of each learner control condition were similar to one another, but, the models produced for the complete learner control and no learner control data accounted for more of the posttest score variance than the predictive models produced for the partial learner control and combined data. Minimal posttest score variance was accounted for by any of the predictive equations that were investigated. (Author/GK)