I have 5 samples cut from the same epoxy film. I perform FTIR-ATR on them 4 times per film before and after exposure to water. So in all, I have 20 FTIR spectra of one film in one condition. If I want to compare the water (OH) peak, should I average all the spectra first and then normalize the resultant spectrum, or should I normalize the OH peak in each spectra and then average it? If I do the latter, the standard deviation comes out to less than 0.1.