Question: We pre-tested some polytomous items under one timing condition and have now collected a very small sample of that same data under another timing condition.  I would like to be able to use the item/step difficulties from the first condition as priors to estimate item/step difficulties for the second condition.

 

Answer: Winsteps does not accept prior distributions, so we need to simulate them as additional data.

 

For instance, we could use the pre-test dataset, down-weight it using PWEIGHT=, and analyze it again together with the new data.

 

When we have a suitable set of item/step difficulties, then export them with IFILE=if.txt, SFILE=sf.txt

 

Then analyze only the new data with anchored items and steps IAFILE=if.txt, SAFILE=sf.txt to get a report on only the new data.