Which post hoc to apply?

#1
I am analyzing my data using graph pad prism. I have groups with unequal sample size (n=12, n=7,n=13,n=6). Independent between group one way Anova yielded a statistically significant results. Please suggest me which post hoc test would be more appropriate to apply. I applied tukey, but read somewhere that it's not appropriate for unequal sample size. Please note: Bartlett's test and Brown-Forsythe test are not significant. Also, in same situation with 3 groups can Fischer LSD be applied
 

hlsmith

Not a robit
#2
Yeah i am gonna imagine that some corrections may not be significant given sample sizes. That doesnt mean you cant use them. This is why you specify what correction you are going to use prior to conducting analyses, researcher degrees of freedom may bias you now.

PS, I typically use Bonferroni since it is easy to understand and over cautious.
 
Last edited: