[Rasch] evidence of validity?
puryabaghaei at gmail.com
Fri Jan 4 02:37:49 EST 2013
The Rasch model is a latent trait model. Latent trait models help
verify that all the items (observed variables) centre around one
reference point which is hopefully our construct of interest. That is
the reason why latent trait models are essential for validation and
validity theories in which latent trait models do not play a central
role are problematic. Fit to the Rasch model is evidence that
variability in the responses is caused by the location of the
individuals on the latent trait (see Denny Borsboom’s papers).
However, latent trait models won’t tell us what exactly the trait we
are measuring is. For this we need substantive construct theory.
On 1/3/13, Mike Linacre <rmt at rasch.org> wrote:
> Item maps are informative about construct and content validity. For
> instance, for the construct validity of an arithmetic test, we expect
> the "addition" items to be at the "easy" end of the map, and the
> "division" items to be at the "hard" end of the map. For its content
> validity, we don't expect to see "geography" items anywhere on the map.
> Mike L.
> Rasch mailing list
> Rasch at acer.edu.au
Purya Baghaei, Ph.D
Islamic Azad University,
Ostad Yusofi St.
Phone: +98 511 6635064-5
Fax: +98 511 6634763
More information about the Rasch