A Computer Program for Trend Analysis in a Two- Or Three-Factor Experiment With Repeated Measures On One of the Factors1

THE use of experimental designs with repeated measures on one of the factors is extensive in the behavioral sciences. Moreover, the repeated measures factor in these designs, e.g., delay of reinforcement, length of isolation period, or dosage of a drug, often has levels which represent equally spaced intervals along an underlying continuum. Hence, additional information about the nature of the relationship between the repeated measures factor (treatment levels) and the dependent variable can be obtained by partitioning the treatment variation into nonoverlapping trend components, i.e., linear, quadratic, cubic, or quartic, through the use of orthogonal polynomials (Winer, 1962, pp. 353-369; Kirk, 1968, pp. 270-275). The program discussed in this paper performs an analysis of variance for a twoor three-factor experiment with repeated measures on one of the factors. More importantly, it provides the researcher with the option of having a trend analysis performed on repeated measures factors of the type described above.