Two-way repeated measures ANOVA in SPSS: one-within, one-between (March 2020)

This paragraph introduces the topic of the video, which is a tutorial on performing two-way repeated measures ANOVA using SPSS. The focus is on scenarios with one between-subjects factor and one within-subjects factor. The video also mentions the availability of a data file and a PowerPoint supplement for the audience's reference. The explanation begins with a review of repeated measures ANOVA and its application in comparing groups over time, leading to the hypothesis of interaction effects between the repeated factor and a grouping variable. The example used in the video involves testing anxiety scores across three time points for different treatment groups.

The second paragraph delves into the results of the multivariate and univariate tests from the ANOVA analysis. It discusses the implications of significant multivariate test results for the main effect of time and the interaction between time and treatment group. The paragraph also addresses the assumption of equality of variance-covariance matrices and the sensitivity of the Box's test to multivariate non-normality. The robustness of the multivariate test is considered in the context of equal or nearly equal group sizes, and the potential impact of non-normality and sample size on the validity of the test results is highlighted.

This section of the script examines the within-subjects effects, focusing on the main effect of time and the interaction between time and treatment group. It discusses the use of tests such as Mauchly's test for sphericity and the implications of its violation on the likelihood of type I errors. The script also explores the use of the Greenhouse-Geisser and Huynh-Feldt adjustments for tests of within-subjects effects, providing guidance on which tests to use based on the degree of departure from sphericity. The significance of linear and quadratic trends over time is also highlighted, with the suggestion that the data may follow a more complex, curvilinear pattern.

The fourth paragraph discusses the tests of between-subjects effects, which involve comparing the average anxiety scores across the different treatment groups. It notes the statistical significance of these tests, indicating differences in average anxiety levels among the groups. The paragraph also covers the estimated marginal means for anxiety at different time points, irrespective of group membership, and the pairwise comparisons that reveal significant differences between time points. The importance of considering the results of Levene's test for equality of error variances when interpreting these findings is emphasized.

The final paragraph of the script outlines the process of conducting simple effects tests to further explore and describe the nature of interactions found in the analysis. It provides a step-by-step guide on modifying the SPSS syntax to request these tests, which involve comparing mean differences in anxiety over time within each treatment group. The results of these tests, including both multivariate and pairwise comparisons, are presented, showing significant mean differences within each group over time. The paragraph concludes by discussing how to compare groups within each time point, offering a comprehensive understanding of the interaction effects.