Analysis of Data from Pre- and Posttest Designs: A Comment