ConQuest Community

Help desk => Questions and Answers => Topic started by: snailwihs on November 14, 2014, 04:45:34 AM

Title: Grouping Variable
Post by: snailwihs on November 14, 2014, 04:45:34 AM
I am conducting MIRT, and want to have correlations between different dimensions differed by different groups (e.g. gender).
I fond the command 'group'. However, I cannot find more information about this command from the users' guide, such as whether the distributions for different groups were set the same and the example of using "group" in MITR.
Thank you so much and look forward to your reply.
Title: Re: Grouping Variable
Post by: Eveline Gebhardt on November 18, 2014, 03:26:29 AM
Hello

Thank you for your post. My colleague Greg Macaskill (macaskill@acer.edu.au) will explain the new group command to you.

Kind regards
Eveline
Title: Re: Grouping Variable
Post by: GREGTMAC on November 18, 2014, 05:31:44 AM
Hi,
     The group command does not affect the estimate command, but allows some outputs to be displayed by groups,such as descriptives and itanal, but these are based on results from the whole dataset. The dofor command in Conquest3 or later, will fit the model for each group. I enclose a small example of its use.

Greg Macaskill