|
|
Person parameters for mixture IRT models |
|
Message/Author |
|
|
Hi, I just want to verify something about the printed person parameter estimates in Mplus when a mixture IRT model is fitted. To my knowledge, when a single-class IRT model is fitted and the FSCORES are requested in the SAVE command, MPLUS is printing EAP estimates for person parameters, right? When a mixture IRT model is fitted and the FSCORES are requested in the SAVE command, is Mplus using class specific item parameters to estimate EAP for each individual based on the assigned class membership? If a two-class model is fitted and Person X is assigned to Class 2 based on the predicted class probabilities, then EAP is estimated based on the response pattern of Person X and estimated Class 2 item parameters. Is this correct? Thanks for your time and clarification. |
|
|
Q1: Yes, if ML is used (as opposed to WLSMV) Q2: Both an overall factor score and the factor score for the most likely class are presented. |
|
Back to top |
|
|