A missing value technique for the analysis of data with repeated measurements on each experimental unit