2
$\begingroup$

I am doing an EEG study with a small sample size (N=13) and I have many variables that I need to compare (for example, onset of readiness potential, maximal point of RP, time or awareness, etc), so i need to perform many paired-sample comparisons. I am not sure though, whether I should be using a dependent t-test, or a non- parametric test such as Wilcoxon's signed rank? I was advised that my sample size is too small to reliably test for the normality of the distribution of the differences between my variables, but then also read elsewhere that normality tests should theoretically work on extremely small sample sizes, so I'm not sure what to believe! Can I test for normality on small sample sizes? or should I just use Wilcoxons for all the comparisons? thanks very much, Emily

Improve this question
$\endgroup$

1 Answer 1

Reset to default
0
$\begingroup$

That's right, t-tests are pretty robust to violations of normality, but so for violations of homogeneity of variance. I'd go for a dependent t-test and then access normality by inspecting a qq-plot and looking at boxplots of the data, instead of conducting any of those commonly used normality tests, like Levene's

Improve this answer
$\endgroup$

Not the answer you're looking for? Browse other questions tagged or ask your own question.