[Rasch] Collapsing categories

Stuart Luppescu slu at ccsr.uchicago.edu
Fri Nov 19 11:13:09 EST 2010

Hello fellow Raschies, We had some questionnaire items with an underused
category, something like this: 1) Never to 10% of the time, 2) 10% of
the time to 15% of the time, 3) 15% of the time to 30% of the time, etc.
The second category was underused so we collapsed 1 and 2. Then in a
subsequent survey we kept the same item but made the response categories
1) Never to 15% of the time, 2) 15% of the time to 30% of the time, etc.

I seem to remember learning in Ben Wright's class that the original item
with the collapsed categories, and the later item with the reworked
response categories could not be assumed to be equivalent. So, my
questions are:
1) Is anyone aware of research on whether items with collapsed
categories and reduced response categories (corresponding to the
collapsed categories) could be considered equivalent?
2) How would I test whether response patterns are significantly
different? I could do a chi-square test on the numbers in categories,
but since these data come from different people at different points in
time, how would I control for this non-equivalence of the samples?

Stuart Luppescu <slu at ccsr.uchicago.edu>
University of Chicago

-------------- next part --------------
A non-text attachment was scrubbed...
Name: smime.p7s
Type: application/x-pkcs7-signature
Size: 3573 bytes
Desc: not available
Url : https://mailinglist.acer.edu.au/pipermail/rasch/attachments/20101118/3b88320f/attachment.bin 

More information about the Rasch mailing list