[Rasch] Differential Item Functioning: Help Needed

Rense Lange rense.lange at gmail.com
Wed Feb 6 17:53:27 EST 2013


About right - but I haven't seen enough dif > 1 logit to be sure about the larger differences. These are binary items, correct?

Rense

On Feb 6, 2013, at 12:05 AM, Mike Linacre <rmt at rasch.org> wrote:

> Folks:
> 
> Here is a DIF nomogram showing the Reference and Focal group sizes to 
> obtain p=.05 statistical significance with various logit sizes of DIF on 
> dichotomous items (MCQs, etc.). For example, the nomogram indicates we 
> need groups of around 100 to detect DIF of 1 logit, and groups of about 
> 1500 (log-scaled) to detect DIF of 0.25 logits.
> 
> It is based on statistical theory, but with approximations and assumptions.
> 
> Does it match your experience? Are the indicated group sizes too small, 
> too large or about right?
> 
> Mike L.
> 
> DIF Nomogram is at http://www.winsteps.com/a/difnomogram.png
> _______________________________________________
> Rasch mailing list
> Rasch at acer.edu.au
> Unsubscribe: https://mailinglist.acer.edu.au/mailman/options/rasch/rense.lange%40gmail.com




More information about the Rasch mailing list