Old fixed bug: Two-way repeated measures ANOVA with matching by columns

We recently discovered a bug in Prism 3.00-3.02 (Windows) and 3.0a (Mac).

When you do RM two way ANOVA with matching by columns, Prism counts up the DF wrong for subjects, leading to wrong MS values, wrong F ratios and wrong P values. This problem only occurs if you match by columns and the number of columns does not match the number of rows.

You can work around this problem easily. From your data, click Analyze, then choose Transpose from the list of data manipulations. Then analyze the resulting table with two way ANOVA, but choose repeated measures by row. Now you get the correct results.

This bug is fixed in version 3.03 (Windows) and 3.0c and 3.0cx (Mac).