FW: [Rasch] Common persons equating

Stephanou, Andrew Stephanou at acer.edu.au
Tue Jan 6 13:18:11 EST 2009


From: Stuart Luppescu [mailto:slu at ccsr.uchicago.edu]
Sent: Mon 1/5/2009 11:59 AM
To: rasch
Subject: Re: [Rasch] Common persons equating
The attached message has been automatically discarded.
 
On ?, 2009-01-05 at 16:06 -0500, Burga León Andrés Alberto wrote:
> Hello to everybody on the list.
> 
> I'm trying to equate two test using a common persons approach. I'm not
> sure how to handle extreme scores: should I elimanate persons with
> extreme scores on test A and B from the calibrations and only
> calculate the equating contants with non-extreme persons?

If you calculate the equating constant as the difference between the
means of the person measures, you can include all people if you
calculate weighted means using the precision (1/se) as the weights.
Since the extreme people will have large standard errors, they will be
down-weighted in the calculations.

--
Stuart Luppescu -=- s-luppescu .at. uchicago.edu       
University of Chicago (^_^)/ CCSR
???????? -=-=- Kernel 2.6.25-gentoo-r
How wonderful opera would be if there were no
 singers.





-------------- next part --------------
An HTML attachment was scrubbed...
URL: https://mailinglist.acer.edu.au/pipermail/rasch/attachments/20090106/19acd8d2/attachment.html 


More information about the Rasch mailing list