Dear community,
I want to ask you a methodological/statistical question.
We performed ten independent hierarchical linear regressions divided by gender for an article writing (N=1502 participants, 39.9% men and 60.1% women). Could Type I error significantly arise through ten independent hierarchical linear regressions? Or do we have enough statistical power for each subgroup to guarantee that we have enough statistical power? I would also appreciate references that can help us make clear this matter.
Thank you so much!
Hi,
Running ten independent hierarchical linear regressions raises the risk of Type I error. Correct for multiple comparisons using methods like Bonferroni or Holm. Your sample size is large, but statistical power also depends on effect size and model complexity. Always conduct power analyses for each subgroup.
Hope this helps.
Don't worry about whether there is Type 1 error, but simply make an adjustment to minimise it, such as doing what Hsin-Yuan Chen said :)
- Badges
- Science topic
- Similar topics
- Psychology
- Mental Processes
- Learning