[Rasch] Why isn't the person separation better?

Mike Linacre rmt at rasch.org
Thu Sep 9 20:08:59 EST 2010


Trudy and Imogene:

There is an ambiguous use of the word "Separation" in the Rasch community.

Ben Wright's "Person Separation Index" =  True S.D. / RMSE - This has a 
range 0 to infinity.

RUMM's "Person Separation Index" = reliability = True variance / Observed 
variance  - This has a range 0 to 1.

So, RUMM's PSI of 0.82 = Wright's PSI of 2.13
So, Wright's PSI of 0.82 = RUMM's PSI of 0.40

Trudy and Imogene, which numbers are you talking about?

Mike L

At 9/9/2010, you wrote:
>Dear Trudy,
>What person separation would you regard as good in your context? In the
>context of a student examination, for example, I would consider 0.82
>very good separation, and would be surprised to find significant
>improvement even with more items.
>Imogene

Mike Linacre
Editor, Rasch Measurement Transactions
rmt at rasch.org www.rasch.org/rmt/ Latest RMT:  24:1 Summer 2010
-------------- next part --------------
An HTML attachment was scrubbed...
URL: https://mailinglist.acer.edu.au/pipermail/rasch/attachments/20100909/7938d076/attachment.html 


More information about the Rasch mailing list