[Rasch] Unidimensionality and Rasch measurement

Looveer, Juho Juho.Looveer at det.nsw.edu.au
Mon Jul 24 11:05:17 EST 2006


Unidimensionality applies to a scale, or a set of items  e.g. a test on

The first issue often is face validity, which should be checked during
test construction - do the items come from the same domain?  Is there a
theoretical basis for the items to be from the same "dimension" -
whether a specific topic like algebra, or as a general indicator of some

When you analyse results from a different population, the statistics
from common Rasch software (e.g. RUMM, Winsteps) will routinely provide
some indicators of fit to the model. 

If the results verify that the items are unidimensional for one
population, that seems fine.
If the same items do not appear to be unidimensional for a different
population (i.e. there are problems with the fit of some items), this
would suggest some differential item performance to me.  Some items have
elicited different responses from the two populations; there is some
underlying reason for this which needs to be considered. Perhaps a
language issues, or the second group had not been taught a certain topic
that was included in the test, etc.

Give me a call sometime if you wish to discuss further.  It may be a
specific or unique problem or sets of data to which you are referring.



Dr Juho Looveer 
Manager, Data Collection
Planning and Innovation 
NSW Department of Education and Training 
Level 5, 35 Bridge St Sydney 
GPO Box 33 
Sydney NSW 2001 
work phone: 956 18192 
fax: 956 18055 
Juho.Looveer at det.nsw.edu.au 

-----Original Message-----
From: rasch-bounces at acer.edu.au [mailto:rasch-bounces at acer.edu.au] On
Behalf Of Jim Athanasou
Sent: Monday, 24 July 2006 10:46 AM
To: rasch at acer.edu.au
Subject: [Rasch] Unidimensionality and Rasch measurement


I have been asked some questions following a seminar and would
appreciate some advice before responding.

The main question is about unidemensionality and that unidimensionality
found in one sample may not hold elsewhere. 

Do we really have to investigate unidimensionality and then re-calibrate
the items (and people) always starting from zero whenever we want to
apply Rasch measurement, or, do we calibrate our scale with confidence
on one group and then apply those findings elsewhere even in (quite)
different populations?

Any advice would be appreciated as well as any comments on novice or
mistaken assumptions that I have made in phrasing the questions.

Jim Athanasou
Rasch mailing list
Rasch at acer.edu.au

This message is intended for the addressee named and may contain
privileged information or confidential information or both. If you
are not the intended recipient please delete it and notify the sender.
-------------- next part --------------
An HTML attachment was scrubbed...
URL: https://mailinglist.acer.edu.au/pipermail/rasch/attachments/20060724/0b84fe6e/attachment.html 

More information about the Rasch mailing list