[Rasch] Modeling content delivered

Mike Linacre rmt at rasch.org
Sat Sep 25 23:06:20 EST 2010


You describe an exciting situation. A couple of questions:

1. Could the math items be spiralled so that different students respond to 
different items on different days?

2. Is there an intervention during the administration (such as math teaching)?

Let's assume the items are spiralled in some way, and there is an 
intervention which operates uniformly across the students, then we could 
estimate this Rasch model:

(baseline ability of person n) + (learning effect up to day d) - 
(difficulty of item i) -> observation

This parallels studies of changes in rater leniency across rating sessions.

Mike Linacre
mike at winsteps.com

At 9/25/2010, you wrote:
>Can anyone point me to an example where a Rasch model was used to evaluate 
>unique test items administered over a series of days? For example, could a 
>Rasch model be used to calibrate a math test delivered two items per day 
>over the course of a couple of weeks? If this is a reasonable goal, what 
>are the pitfalls?
>Thank you for your input.
>Jacob Kean

More information about the Rasch mailing list