[Rasch] Test Translation & DIF

Matt Schulz mschulz at pacificmetrics.com
Tue Jan 20 04:14:26 EST 2009



Under the assumption of measure invariance, you can (and should) analyze the
data separately for each group, and compare the various indices (esp item
and step difficulties).   This is a common item design under the assumption
of measure invariance. 







From: rasch-bounces at acer.edu.au [mailto:rasch-bounces at acer.edu.au] On Behalf
Of Anthony James
Sent: Monday, January 19, 2009 11:06 AM
To: rasch at acer.edu.au
Subject: Re: [Rasch] Test Translation & DIF


Thanks for your comments,

I?m aware of the problems that may occur in test translation. I wanted to
know what analytical and psychometric activities can be done to make sure of
translation equivalence after translating the instrument and carrying out
the qualitative works.

Matt?s paper is very helpful. He states that item fit, person fit, item
difficulty estimates and step calibrations should be comparable across
groups. Now a clerical question:

In order to compare item estimates one cannot run separate analyses, I
think. The two versions must be linked. I?m using Winsteps. Can I just enter
the items once and then enter the responses of the two language groups under
one set of item labels as if all have taken the same test and then run DIF
analysis? Or should I have a bilingual group who takes half of the items in
one of the languages and the second half in the other language and use this
group to link the items?




-------------- next part --------------
An HTML attachment was scrubbed...
URL: https://mailinglist.acer.edu.au/pipermail/rasch/attachments/20090119/9ee9509f/attachment.html 

More information about the Rasch mailing list