[Rasch] Intepreting Rasch-Andrich (or other ) thresholds in checklist scores (NOT likert scales)

Mike Linacre mike at winsteps.com
Mon Mar 25 18:50:53 EST 2013


Thank you for your question, Imogene.

Your checklist is acting as a testlet of 12 locally-dependent items. The 
testlet becomes a super-item with a "rating" equal to the score on the 
sub-test.

David Andrich comments on this situation in 
http://www.rasch.org/rmt/rmt251c.htm

As always, the Rasch measures are interval measures, but the meaning of 
the Rasch-Andrich thresholds depends on the degree of local dependency 
among the criteria. If you want to divide up the latent variable 
according to the count of correct criteria, then it would be easier for 
your audience if you use the Rasch-Thurstone thresholds instead of the 
Rasch-Andrich thresholds. See, for instance, 
http://www.rasch.org/rmt/rmt233e.htm .

Mike L.

On 3/24/2013 20:04 PM, Imogene Rothnie wrote:
>
> I have a checklist type assessment where say ANY 9 correct criteria 
> out of 12 would produce the same Rasch measure.  This is opposed to 
> the usual likert scale ratings of performance.
>
> I want to look at the Rasch-Andrich thresholds between checklist 
> scores achieved on each station. Is it valid to view increasing N 
> checklist items correct as interval measurement, or even in the way we 
> would interpret Rasch-Andrich thresholds for a likert scale type item?
>
> Many thanks
>
> Imogene Rothnie
>

-------------- next part --------------
An HTML attachment was scrubbed...
URL: https://mailinglist.acer.edu.au/pipermail/rasch/attachments/20130325/98078389/attachment.html 


More information about the Rasch mailing list