08-19-2016 02:25 PM
I will be doing a validation analysis of a newly developed questionnaire against an old/gold standard one. Currently, we have data collected from both quesionnaires on the same patients (some item are binary and some are in likert format). And we have developed some hypotheses that which items on the new questionnaire would be correlated to which items in the old one.
To perform validity analysis, my initial data analysis would be performing some simple correlation analsyis between related items. However, there are many items to compare individually and family-wise error rate would be pretty high. Plus, this is a little to simple and can't validate the new quesionnaire as a whole.
So, my question is after performing some correlation analysis, what method/technique I should try to get an overal number/index or whatever statistics which will tell us clearly and easily how much convergent validity the new questionnaire has against the old one?
I am sure there are some satistical approaches out there (factor analysis?) for my purpose but I am not sure what it would be. Any suggestions would be appreciated.