ANOVA question

[Giovanni]

Hi,
I'm analysing organism health using different treatments. I'd like to analyse my data using percentage treatment of my control but query whether this violates assumptions in repeated measures anova? ie each control rep would vary slightly from the 100% but the mean would be 100% at each time measurement. Each treatment would be a percent of the control at each time.

I've found one paper which states that this is not appropriate BUT have found many other papers that actually go ahead and do it this way. I've looked through stats books but cannot seem to find an answer to this. Geting confused about compound sphericity etc.

Was hoping some Biology/ Ecology people out there may have faced this problem before. Is it okay to go ahead with the analysis this way? Or should I revert to the raw (non-percentage) data?

Thanks