Calculating the effectiveness of an educational intervention using a pretest-posttest design. I plan to run an ANCOVA on the scores with the pretest as the covariate. The six subcatagories on the test give the raw score of correct answers out of the total possible on each subcatagories. The overall score is given as a scaled score to equalize any differences on the pretest and posttest versions. The number of questions in each subcatagory and the overall test vary from the pretest and the posttest. Do I need to convert the scores to z scores to run the ANCOVA or can I use a percent score for each subcatagory and the scaled score for the overall score?