[Rasch] Average inmnsq?

rsmith rsmith at jampress.org
Thu Dec 16 01:24:59 EST 2010


No doubt that you can calculate a mean, but then the interpretation of the mean is a bit treacherous.  With a number that is bounded by zero at the lower end and infinity at the upper end a change of .05 does not mean the same thing above and below 1.0 (the expected value).

Richard M. Smith, Editor
Journal of Applied Measurement
P.O. Box 1283
Maple Grove, MN  55311, USA
website:  www.jampress.org
phone: 763-268-2282
fax: 763-268-2782 

-------- Original Message --------
> From: Mike Linacre <rmt at rasch.org>
> Sent: Tuesday, December 14, 2010 8:31 PM
> To: rasch at acer.edu.au
> Subject: Re: [Rasch] Average inmnsq?
> 
> Stuart:
> 
> Each mean-square is derived from an empirical variance divided by a model 
> variance.
> 
> So, if the mean-squares are based on roughly the same number of 
> observations, and we compute their arithmetic mean, then we obtain a 
> usefully approximate average of them.
> 
> OK?
> 
> Mike L.
> 
> At 12/14/2010, you wrote:
> >A quick question I have: Is it appropriate to calculate average infit
> >mean-squares, or do I need to square them first, combine then and then
> >take the square root?
> 
> Mike Linacre
> Editor, Rasch Measurement Transactions
> rmt at rasch.org www.rasch.org/rmt/ Latest RMT:  24:3 Winter 2010
> _______________________________________________
> Rasch mailing list
> Rasch at acer.edu.au
> Unsubscribe: https://mailinglist.acer.edu.au/mailman/options/rasch/rsmith%40jampress.org 






More information about the Rasch mailing list