[Rasch] Validation & Rasch
Mike Linacre (RMT)
rmt at rasch.org
Fri Nov 23 06:46:33 EST 2007
Thank you for classifying our technical activities regarding test validity.
For the traditionalists, please add:
Please also add these conceptual (and more important) activities:
A. Construct validity: Investigating what the test is actually measuring,
usually through the item hierarchy from less to more.
B. Predictive validity: Investigating whether what the test is measuring
accords with the person hierarchy from less to more.
C. Misfit diagnosis: Investigating what treasure (to be mined) or trash (to
be thrown out) is also present in our test responses.
D. Communicating findings: no matter how valid a test is psychometrically,
if its consumers misunderstand or misuse its findings, its practical
validity is compromised.
At 11/22/2007, you wrote:
>Looking over papers which have used Rasch methodology to validate tests, I
>found out a number of procedures for validating
>1. Studying fit stats.
>2. Checking threshold ordering
>3. Assessing unidimensionality using PCA of residuals
>4. Investigating DIF
>5. Checking person measure invariance over subsets of the test
Editor, Rasch Measurement Transactions
rmt at rasch.org www.rasch.org/rmt/ Latest RMT: 21:2 Autumn 2007
-------------- next part --------------
An HTML attachment was scrubbed...
More information about the Rasch