[Rasch] RE: using ability estimates instead of sum scores

Stephen Humphry stephen.humphry at uwa.edu.au
Thu Feb 11 22:20:10 EST 2010

Hi Felix,
You might need to clarify. It's standard according to theory to replace sum scores by estimates: the estimates are obtained from sum scores in the Rasch model (from weighted scores in the 2PL). This is provided there are no missing responses. If there are missing responses, you use ability estimates based on the responses there are and it's a bit more involved than replacing using one correspondence table of raw scores to ability estimates. The simplest kind of ability estimates have some bias in typical situations, particularly for people with more extreme (high or low) scores. For most 'practical' purposes it doesn't matter much at all. If it does matter, you can use unbiased estimates like Weighted Likelihood Estimates (WLEs).
Hope that helps.

From: rasch-bounces at acer.edu.au [rasch-bounces at acer.edu.au] On Behalf Of Fischer, Felix [Felix.Fischer at charite.de]
Sent: Thursday, 11 February 2010 4:04 PM
To: 'rasch at acer.edu.au'
Subject: [Rasch] using ability estimates instead of sum scores

Hello everybody,

can someone give me a hint about the effects of using ability estimates from an IRT model instead of classical derived sum scores (e.g. when using a quality of life measure)? I red that there might be a "bias", but i can't follow those advanced statistical papers in PSYCHOMETRIKA. Is it practical to just replace sum scores by ability estimates? Any literature regarding this topic would be a help, too.



Dipl. Psych. Felix Fischer

Medizinische Klinik mit Schwerpunkt Psychosomatik

Charité -- Universitätsmedizin Berlin

Luisenstr. 13a

10117 Berlin

Tel.: 030 - 450 553575

Email: felix.fischer at charite.de



Please consider the environment before you print
Rasch mailing list
Rasch at acer.edu.au

Please consider the environment before you print

More information about the Rasch mailing list