Why did you compare group A pre with group C post? I cannot imagine any meaningful research
question which can be answered by such a comparison.
The most efficient way to analyse such data would include the development between baseline
and follow-up. Either by using mixed analysis of variance (with one repeated-measures factor and
one grouping factor). Or simply by calculating the pre-post difference for each subject and then
comparing them by oneway analysis of variance, with post-hoc tests for pairwise comparison.
Or is the measurement of "arm strength" strictly ordinal scaled?