can ata be used with grm/pcm?

[peterhalpin]

Hi! I think your package is really great! It is the only one I have found for ATA, without having to write the linear program manually.

I am not sure if ata with grm / pcm is currently supported? Can you please advise? As a suggestion, if ata is not available for non-binary responses, it would be helpful to indicate this in the documentation.

A follow-up question (because I am pretty sure its not supported)-- what do you think about the following strategy for the grm. Convert an item with c = 1, ..., C response categories into C-1 binary items with response functions P(Y > c | theta) and then writing a support function to require that C-1 categories must be selected together (like the opposite of item_enemy). Do you think that would work? or is there an easier way I am missing?

Thanks again for the great package. I also enjoyed your JEM 2020 paper :)

[xluo11]

Hello. Sorry for the late reply. It's been extremely busy at work recently. I was in the process of splitting the xxIRT package into smaller, standalone modules, making it easier to maintain, but haven't got a chance to publish the updates. So, this might be helpful: - https://github.com/xluo11/Rata. - https://github.com/xluo11/Rirt Let me know if you have any questions. Xiao

…

On Mon, Jul 19, 2021 at 4:26 PM Peter Halpin ***@***.***> wrote: Hi! I think your package is really great! It is the only one I have found for ATA, without having to write the linear program manually. I am not sure if ata with grm / pcm is currently supported? Can you please advise? As a suggestion, if ata is not available for non-binary responses, it would be helpful to indicate this in the documentation. A follow-up question (because I am pretty sure its not supported)-- what do you think about the following strategy for the grm. Convert an item with c = 1, ..., C response categories into C-1 binary items with response functions P(Y > c | theta) and then writing a support function to require that C-1 categories must be selected together (like the opposite of item_enemy). Do you think that would work? or is there an easier way I am missing? Thanks again for the great package. I also enjoyed your JEM 2020 paper :) — You are receiving this because you are subscribed to this thread. Reply to this email directly, view it on GitHub <#11>, or unsubscribe <https://github.com/notifications/unsubscribe-auth/ABZ3BU6XGTUZI2R4RTFQFELTYSDADANCNFSM5AUNE7TQ> .