[Rasch] Validity in Rasch

Stephen Humphry stephen.humphry at uwa.edu.au
Thu Mar 24 00:08:38 EST 2011


Hi Svend, putting aside more basic issues:

"2) Modelling the departure from the model. This is what they do in RUMM when they split items"

I understand what you mean, but wouldn't say the point of splitting items is intended to be modelling departure; the point is supposed to be to analyse data in a different way (with an item split for different person groups) to see if there is data-model fit even though there isn't when the data are analysed the original way (the item is not split, treated as 'functioning' the same for different groups of people). I'll leave it for you to decide whether you think this is splitting hairs ;)

Best, Steve

________________________________
From: rasch-bounces at acer.edu.au [rasch-bounces at acer.edu.au] On Behalf Of Svend Kreiner [S.Kreiner at biostat.ku.dk]
Sent: Wednesday, 23 March 2011 7:07 PM
To: rasch at acer.edu.au
Subject: Re: [Rasch] Validity in Rasch

Why can't we have it both ways?

The assumptions of the Rasch model defines sound (and therefore also valid) measurement
and
the assumptions have to be tested because we do not have sound measurement if they are violated.

There really is no conflict.

What we may disagree about is what to do when items do not fit the model.

In such cases we have three options:

1) Purification by elimination of items that do not fit the model,
2) Modelling the departure from the model. This is what they do in RUMM when they split items
3) Use the Rasch model anyway, hoping that measurement is robust to the item-model misfit.

Best

Svend

Den 23-03-2011 02:00, Jim Sick skrev:

Juanito,

I've written a short online article dealing with the "assumptions" of
the Rasch model, in which I recommend that we avoid thinking of them as
"assumptions which need to be tested a priori," as is the case for
statistical tests like ANOVA. It is often better to think of Rasch
assumptions as "requirements" of sound measurement. A primary goal of a
Rasch analysis is to determine to what degree the requirements (or
assumptions) have been approximated. Read my article here, if you like.
It may not answer your questions directly, but will give you a better
overview.

http://jalt.org/test/sic_5.htm

Best
Jim Sick
Tokyo
_______________________________________________
Rasch mailing list
Rasch at acer.edu.au<mailto:Rasch at acer.edu.au>
Unsubscribe: https://mailinglist.acer.edu.au/mailman/options/rasch/s.kreiner%40biostat.ku.dk



--
Svend Kreiner
Professor
Department of Biostatistics
University of Copenhagen

Øster farimagsgade 5, entr. B
P.O. Box 2099
DK-1014 Copenhagen K, Denmark

Email: S.Kreiner at biostat.ku.dk<mailto:S.Kreiner at biostat.ku.dk>
Phone: (+45) 35 32 75 97

Fax: (+45) 35 32 79 07




More information about the Rasch mailing list