Friedman's interpretation

Hi all,

I'm having some real difficulty interpreting the output of a Friedman's ANOVA. I wonder if anybody can help?

Essentially I have a group of 21 people who received an intervention for tremor. A measure of their tremor severity was performed at baseline (t0), during the intervention (t1) and at the end of the intervention (t2). Higher values = worse tremor. It is a ratio measure but distribution is non-normal (very positively skewed), hence choosing Friedman's.

Here is the SPSS output:


This suggests a significant change, but in which direction? Looking at the raw data, only 5/21 participants had an increase in their STT score, the rest decreased. But the median value increased from 4.3 to 4.6. The mean ranks decreased between baseline and time 1. I'm confused.

I know I need to then examine baseline --> time 1 in more detail so I performed a Wilcoxon Signed Rank Test but I'm still confused as to whether this shows a decrease over time or an increase? Which statistic should I be looking at?


Many thanks in advance!