I want to investigate differences in 6 tasks, in one sample composed by only one group.
I can't use RM ANOVA because data are not normally distributed and the sample is small.
Can I use non parametric ANOVA (unpaired)? Or i have to use a paired test?
I don't know if in a paired test is possible to compare the observations that are more than 2 (in this case are 6).
Martina Gnerre , you cannot use any type of t-test if your data is not normally distributed, see:
Article More about the basic assumptions of t-test: normality and sample size
If you have more than 2 groups to compare, and if your data is not normally distributed, you can use the Kruskal-Wallis test (the Kruskal-Wallis is a non-parametric version of ANOVA):
https://en.wikipedia.org/wiki/Kruskal%E2%80%93Wallis_one-way_analysis_of_variance
The repeated measure is the distinguishing feature of paired tests. As Rolando Gonzales Martinez said, normality is a presumption for paired t-test and RM ANOVA. Friedman's test is a good alternative.
If you are pairing, the normality that matters is the normality of the *differences* not of the data--and they are likely to be better. Also, ANOVAs are quite robust wrt the normality assumption; you'll lose some power, but probably not as much as switching to a nonparametric method.
- Badges
- Science method
- Similar topics
- Mathematics
- Statistics
- Statistical Modeling
- Analyses of Variance
- ANOVA