Suppose a number of persons give Likert responses to [;p;] items on a questionnaire, repeating on a number of time points. I want to show what item-response [;j;] at time [;t;] best explains a change in item-response [;k;] at time [;t+1;].
One solution I've come across involves fitting a large number of vector-auto-regressive models to the data, pick the model that fits the best/does not violate model assumptions and then calculate which item-response Granger-causes another item-response.
I suspect this is not the best way to do it, one reason being the use of [;p;]-values (even though Bonferroni corrections were applied). Does anyone have an idea on how to tackle this problme?
[link][2 comments]