29 Alternatives to the Repeated Measures ANOVA

If we have a repeated measures with a single independent variable, but a small sample size and/or have violated the assumption of normality, it is advisable to use the non-parametric test, Friedman’s test.

We can select this in jamovi by selecting ANOVA, then Repeated Measures ANOVA – Friedman, under Non-Parametric. Move the levels of the independent variable to the Measures box, select Pairwise comparisons (Durbin-Conover) if you would like post hoc tests, and get descriptive statistics as well.

With the Broca’s aphasia example, the output will look like this:

Note that this is an interesting situation where our non-parametric pairwise comparisons turned out to be more powerful than the parametric version of the test. Under many circumstances, the parametric test will be more powerful, but sometimes the non-parametric equivalent ends up being more powerful.

When we go to write up the results, we should report the median as well as the mean, because this is the measure of central tendency used in the Friedman test. The medians are obtained by selecting Descriptives when we run the test.

Friedman’s test was performed examining how three tasks affected word recognition in patients suffering from Broca’s Aphasia. Task type significantly affected word recall, Χ²(2) = 6.64, p = .036. Pairwise comparisons using Durbin-Conover indicated that participants recognized significantly more words in the speech task (M = 7.17, Mdn = 7.50) than participants in the syntax task (M = 4.33, Mdn = 6.50; p = .006). There were no differences between the conceptual task (M = 6.17, Mdn = 6.50) and both the speech and syntax tasks.

Now that we have run both these tests, which should we write up? In this situation, given the small sample size, it would be wise to report the non-parametric test.