looks like you did it wrong; as for a matched t test you MUST compute the pre-post differences and do the T^2 test on them If some of the items are measuring the same thing, you would do better to average them. You might then follow up with individual t tests.
hope you've got lot's of obs with 17 variables. If this was a class, you probably WON't get significance. I think you were simply testing if the means of your variables were non-zero . . ================================================================= Instructions for joining and leaving this list, remarks about the problem of INAPPROPRIATE MESSAGES, and archives are available at: . http://jse.stat.ncsu.edu/ . =================================================================
