Hi there,
I have been given the advice that for large samples (>100) power issues are not a problem. (FYI there has been sufficient power for my two independent t-tests and for a 2x2 chi-square analysis). Therefore, with this rationale, I was advised we do not need to use Bonferroni corrections. Is this accurate?
I am asking this question as I am confused; I believed power was an issue of Type II error, and that Bonferroni protected against Type I (family-wise) error.
How would having sufficient power protect against Type 1 error?
Thanking you in advance.
A lot of issues in what you raise here:
1) power concerns arise from the types of analyses, the size of effect you are looking for and number of variables as much as the overall sample size. I think it is just good practice even if you probably won't have power issue to be able to statistically express that you have sufficient power
2) Bonferroni issues arise mainly from looking at the number of comparisons you make and the significance levels of significant findings. Again sample size is not the main issue. Small differences over multiple comparisons may just be due to change, the correction helps with assessing that concern
- Badges
- Science topic
- Similar topics
- Psychology