I have performed a 2 x 2 x (2) split-plot analysis on raw scores of a questionnaire. The questionnaire was taken twice by participants, once before and once after the intervention in question. A reviewer has suggested:
(1) that the split-plot analysis does not sufficiently take account of 'spurious differences in baseline' scores
(2) that instead of the split plot analysis on raw scores, I should calculate a single variable which they have called 'individual proportion of pre-intervention score' (I can't tell if they mean
Firstly is the reviewer correct about point 1? I always thought that a split-plot analysis was explicitly an analysis of change, specifically difference in amount of change across individuals.
Secondly what advantage, if any, is there to transforming the variable to a proportion instead of using raw scores (given that the raw scores fulfill the usual assumptions of ANOVA)?
in amount of change.