# Which test should I choose ?

#### SimonP

##### New Member
Hello everyone. Quick question, if I may.
I am trying to calculate the statistics of the following situation. I have two groups of people :
1) group A (10 people) following one specific therapy. I will measure their pain level of a specific muscle at three different times : t1, t2 and t3
2) group B (10 different people) follows another therapy. I will also measure their pain level at three different times : t1,t2 and t3
My questiosn :
- Which test should I use to compare the evolution within each group ? I would like to perform a Friedman test, because the resultats are not normal. Is this correct ?
- Which test should I use to compare t1 of group A VS group B, t2 of group A VS group B and t3 of group A VS group B ? Can I make a "Mann-Whitney" test ?
Thank you in advance

#### Karabiner

##### TS Contributor
The Friedman test is not useful here if you want to compare changes over time between
groups. Friedman is for repeated measures of 1 group only.

You could maybe rely on the robustness of the analysis of variance with regard to
normality/non-normality, and perform a"mixed" analysis of variance with a within-subject
factor and a group factor.

Or you could specify which changes you are interested in, e.g. only t1 to t2, and t1 to t3.
In that case, you could calculate individual differences for the 20 participants, and compare
them between groups.

Why do you want to make the three cross-sectional comparisons?

With kind regards

Karabiner

#### SimonP

##### New Member
The Friedman test is not useful here if you want to compare changes over time between
groups. Friedman is for repeated measures of 1 group only.

You could maybe rely on the robustness of the analysis of variance with regard to
normality/non-normality, and perform a"mixed" analysis of variance with a within-subject
factor and a group factor.

Or you could specify which changes you are interested in, e.g. only t1 to t2, and t1 to t3.
In that case, you could calculate individual differences for the 20 participants, and compare
them between groups.

Why do you want to make the three cross-sectional comparisons?

With kind regards

Karabiner
Hello and thank you for the answer....nonetheless :
- a Mixed ANOVA would not work, I think, if my data do not follow an abnormal distribution. Is that correct ?
- the changes I want to investigate are :
1) is there a difference over time within each group ?
2) is there a difference at each time between each group ?

What do you think ?

#### Karabiner

##### TS Contributor
1) is there a difference over time within each group ?
2) is there a difference at each time between each group ?
I don't how much sense these research questions make in the context of your study,
but for the first you could use 2 Friedman tests, for the second 3 U-tests.

With kind regards

Karabiner