help in data analysis for physiotherapy bachelor thesis

Thank you very much. That is the output I receive...
but then, why so different from a one-way anova of difference scores (where it founds high significant interaction)??
and then, can I make a little trick and analyse only difference scores, declaring then in the section discussion the limits of this method?
Yes but I don't think it would be neither incorrect nor unfair.
Simply, "absolute" values I collected have too big standard deviation due to the high variation of ankle range of motion that exists amongst individuals. If I were to do the same experiment on, say, 100 or even 1000 subjects, I would still have an high standard deviation.
I mean high, when compared to the improvement subjects had, which is clinically relevant because obtained in just one treatment session (that means, the treatment could then be applied with the knowledge of being effective).
That is why I insist on asking if a one way anova can be applied on difference scores. Because it is the only way to focus on treatment effect, avoiding subject's range of motion natural variability.
Have I been a little bit clearer now? Do you still think it is not advisable the use of different score? And if you do, then I ask: if not advisable, is it at least allowed and not totally incorrect?

Thank you.
I had a talk today with a biostatician who is researcher in my university.
After having viewed all the data, he said that the way is to make a one-way ANOVA with difference scores (which gives F=24, p<0.001) and then pairwise comparison between groups, correcting the original p value (0.05) in 0.017 according to Bonferroni's rule.


Ninja say what!?!
I think what he has suggested if fine. However, I would still not recommend the Bonferroni rule. If you're looking for a way to adjust for comparing multiple groups, there are other methods you can use that are less conservative (do a google search). The Bonferroni method is VERY conservative and in my opinion should only be used if you're testing multiple different hypotheses.


Ambassador to the humans
What he means is that a Bonferroni correction is very strict in a sense. It basically says that you need a TINY p-value before you claim something is significant. There are other methods which make it so that you still get the protection you want when comparing multiple groups but you don't have to take quite as large of a hit in that the p-value doesn't need to be quite as small as it would with the Bonferroni correction.