Assume that I conducted an experiment with 20 questions. The questions 1-10 are Type A questions whereas questions 11-20 are Type B questions.
I am wondering whether two types of questions differ from each other.
A natural approach is to calculate the average response for Type A and Type B questions for each participant. Then with a repeated measures t-test, I can test the hypothesis that whether two types of questions are similar or not.
Another approach is to transpose the data (switching rows to columns) and calculating the average response for each question from 1 to 20. Then an independent samples t-test can test whether the rows 1-10 are different from 11-20.
I know that the first approach is commonly used, but I don’t know why the second approach is not popular (or is ever used).