Dear community,
I'm performing a one-way ANOVA analysis. As sample sizes and variances are unequal, I am calculating the Welch's F, and want to use Games-Howell to perform pairwise post-hoc tests. It is not clear to me whether the SPSS output displaying the Games-Howell post hoc-tests are corrected for multiple comparisons (like e.g. the Bonferroni or Tukey post-hoc tests do). Does anyone know if they are, or do I have to perform a correction for multiple comparisons manually? All input appreciated!
Best,
-Lasse
Hello, You might find the following link useful.
https://www.researchgate.net/post/How_can_I_calculate_Games-Howell_post-hoc_tests_for_factorial_ANOVA_using_SPSS
Dear Lasse,
The Games-Howell test is a test that does not assume equal variances. Your choice of using the test should be dependent on whether the variances across your groups are heterogeneous. You should first of all perform the Levene's test of homogeneity of variance. If the Levene's test gives a "Sig." value that is less than 0.05, it means that the assumption of the homogeneity of variance has been violated and considering the fact that your experimental groups have different sample sizes, you are on track to use the Games-Howell test as it gives you correct results which you can use to make comparisons. It is corrected in a way similar to the Tukey test.
Additionally, you can use a robust test such as the Welch's test.
- Badges
- Science topic
- Similar topics
- Mathematics
- Statistics
- Sampling Design
- Sample Size