Hi,
I found a significant relationship in one group and non-significant relationship in the other group. But after running a chi-square difference in a multi-group analysis, the difference was not significant. In this case, is it because of a small size of sample? (N=100 for each) or a small effect size in one group that has a significant result? How can I interpret this?
Will be appreciated to anyone who can give me some comments.
First of all, if you're going to run a multi-group analysis, you should have run it first. If you don't find a significant p-value with it, you stop. If you find a significant p-value, then you would do pairwise testing. For this pairwise testing, you should adjust the alpha level for the multiple testing. The Bonferroni correction is probably the most commonly used adjustment but a little too conservative in my opinion. I prefer the false discovery rate (FDR) adjustment when it is appropriate.
- Badges
- Science topic
- Similar topics
- Biological Science
- Kinesiology
- Running