Hi all-

This may be a very simple question, but I’m not sure if a dependent t test is OK here.

I have a survey that was given to around 100 people before and after a training. On the survey, there are many items relating to 9 major areas. Most of the areas are measured by a number of items on a 5- point Likert scale. To test the differences on each of these areas before and after the training, I’m thinking to add up the item scores in an area for each participant, so I’ll have a before- and after- training total scores for that area for all the participants and run a dependent t test.

But for one particular area, there are only 3 questions/items, and these items have Yes/No choices, so I have only binary data (before and after training) for these 3 items. I wonder if I can give 1 to Yes and 0 to NO, add up the responses within the area for each participant and run a dependent t test like I plan to do with the other items. My concern is that if I add up the responses, each participant's area score can only go from 0 to 3, will the range be too small to meet the normality assumption?

Thanks in advance. Sandra