Thank you for your help; this is a survey questionnaire with a Likert scale. I wanted to run the Wilcoxson sign rank test using proc univariate to do a few comparisons, such as comparing pre_arm with post_arm.
In the first design, if I create a column named total, which is the Sum(of Q1-Q24) for each patient_id, then how do I calculate (total score of post_arm – total score of pre_arm) for each id to place that in proc univariate?
but with the second design, I can create two columns named a total of pre_arm, which is Sum(of preQ1-preQ24), and a total of post_arm, which is Sum(of postQ1-postQ24), then calculate the difference b/n these measures to place that in proc univariate.
It seems that there would be easier way with long format, is n’t it?
... View more
