[Rasch] Re: Rasch Digest, Vol 41, Issue 1
juho.looveer at gmail.com
Wed Dec 3 13:19:50 EST 2008
A few years ago I advised on a project where we assessed satisfaction in an
educational setting in one year, to establsih a scale of satisfaction. This
was based on a large set of pooled items (each campus could administer a
different set of items, but included a common core set of items). One of
the purposes of the scale was to identify areas where students were not
satisfied, and to implement changes to address these areas. We were to go
back a few years later, re-assess with a different cohort of students, and
look for change.
i.e. we were using a calibrated scale to look for changes - a bit different
to usual Rasch methodology.
The challenge was to first identify whetehr items on the scale had changed
location (i.e. whetehr the intervention strategies had worked), and then to
assess whether satisfaction poverall had changed.
First step - have the items changed?
Several ideas suggested. One was to include the year or assessment time as
a variable and investigate differential item functioning for this (and I
prefer using RUMM for dif, as you get a visual which allows you to consider
the whole range of "ability" rather than one overall statistic that can hide
non-uniform dif). Another idea was to anchor the items from the first
assessment, and then re-analyse and assess item fit for new application of
items to the original scale.
OIf you determine that the items have changed (location), then you treat
them as separate items when looking for overall change.
Happy to discuss further if you like..
Dr Juho Looveer
Juho.Looveer at gmail.com
On Wed, Dec 3, 2008 at 12:00 PM, <rasch-request at acer.edu.au> wrote:
> Send Rasch mailing list submissions to
> rasch at acer.edu.au
> To subscribe or unsubscribe via the World Wide Web, visit
> or, via email, send a message with subject or body 'help' to
> rasch-request at acer.edu.au
> You can reach the person managing the list at
> rasch-owner at acer.edu.au
> When replying, please edit your Subject line so it is more specific
> than "Re: Contents of Rasch digest..."
> Today's Topics:
> 1. Longitudinal Rasch (Petroski, Greg)
> Message: 1
> Date: Tue, 2 Dec 2008 12:27:07 -0600
> From: "Petroski, Greg" <PetroskiG at health.missouri.edu>
> Subject: [Rasch] Longitudinal Rasch
> To: <rasch at acer.edu.au>
> 0510DC719E56F64BB2AD84EE64CE6BAD02BB5931 at UM-XMAIL06.um.umsystem.edu>
> Content-Type: text/plain; charset="us-ascii"
> I know I asked a related question long ago but after after changing
> computers I can not fine the responses.
> I want to score an existing scale using the partial credit model and use
> these scores to examine change over time. This is a medical-type
> evaluation, is NOT a self-report, and the items are always the same.
> The final analysis will be complex with lots of covariates and a nested
> data structure, so I would like to score the tool with WinSteps or
> ConQuest, and use the save the trait estimates as the outcome in the
> subsequent analysis. What I am not sure of is how to handle the
> repeated measurements on the same people. Each participant will be
> evaluated at least four times. Here is my thinking,
> 1) Ignore the time period and estimate model parameters and "abilities"
> just once as though the data was cross-sectional.
> 2) N is large, so estimate the item parameters using the baseline
> assessment and treat them as fixed and estimate only abilities for each
> subsequent time point.
> Two issues - Will the software I have do this? Does one need an
> equating step?
> 3) Separate calibrations and use an equating procedure to put abilities
> on the same metric.
> 3) Forget Rasch/IRT and use the summated score!
> Any and all thoughts are welcome.
> Gregory F. Petroski, PhD
> Office of Medical Research
> School of Medicine
> University of Missouri -Columbia
> Columbia, Mo. 65212
> -------------- next part --------------
> An HTML attachment was scrubbed...
> Rasch mailing list
> Rasch at acer.edu.au
> End of Rasch Digest, Vol 41, Issue 1
-------------- next part --------------
An HTML attachment was scrubbed...
More information about the Rasch