Protocol Online logo
Top : New Forum Archives (2009-): : Bioinformatics and Biostatistics

Anova on data expressed as a ratio/% of control - (Jul/16/2012 )


I have several sets of data, each from a different individual. I was investigating the cell response to various treatments. Each set contained a control, and several treatment groups. Since the control response was variable, I normalised individual responses by expressing effects as a percentage change from control - set to 100%.

I would like to compare these groups but I'm not sure how. Firstly, how do I calculate the error of group means? and since the control means have been normalised, how do I express the error of control? Can I use anova or is this not right?

Your help is much appreciated


Yes you can use ANOVA. For each data set, you get control-normalized % change, and use it from all dataset to derive SD and do ANOVA.


Thank you for your reply. There is one part of the question I still don't know however. In previous papers that have used % from control, the vertical bar for control has an error attached to it. How can this be, since the control has been set to 100%. so for example...

Animal 1 - assayed in triplicate
mean control - 50 = 100%
mean treatment1 - 40 = 80%
mean treatment2 - 45 = 90%

Animal 2 - assayed in triplicate
mean control - 40 = 100%
mean treatment1 - 35 = 88%
mean treatment2 - 30 = 75%

Mean of means
Control - mean = 100%, SEM =?
Treatment 1 - mean = 84%, SEM = 4
Treatment 2 - mean = 83%, SEM = 7.5

Could somebody please help me,