Generalized Graded Unfold Model

[Seongho Bae]

Hello all,

I want to try to use the Generalized Graded Unfold Model (GGUM) in my data.

I already know software what supports the Generalized Graded Unfold Model when build 2004, but It doesn't run in 2014 now.

Anyone know what software can support the Generalized Graded Unfold Model even in R?

And, the Generalized Graded Unfold Model can be supported by multidimensional item response theory paradigm?

Cheers.

--

Seongho Bae

[Phil Chalmers]

Hi Seongho,

I've thought about adding the GGUM into mirt, but am not sure how to generalize it to multidimensional models. There is a multidimensional ideal point model for dichotomous data in mirt already, but not for polytomous. Part of the reason I haven't implemented it to date is that I couldn't figure out how the probability trace lines are supposed to look given the parameters (if you could supply a working one with comments, that would help save me a lot of time). 

And no, I don't think GGUM has been generalized to multidimensional IRT in the literature, or that other R packages have the ability to estimate the unidimensional GGUM. Cheers.

Phil

--
You received this message because you are subscribed to the Google Groups "mirt-package" group.
To unsubscribe from this group and stop receiving emails from it, send an email to mirt-package...@googlegroups.com.
For more options, visit https://groups.google.com/d/optout.

[Nathan Carter]

Hi Phil,
First, thank you for creating and maintaining the 'mirt' package, it is such a great and needed piece of software .

Second, I've been holding my tongue until this got accepted (I was a reviewer) but here is an article detailing the MGGUM for between-items multidimensionality .. The within-items case gets very hairy (I.e. What is a multidimensional ideal point anyway?)

I know for me it would be extremely exciting to have a MGGUM software ! Link below

http://m.apm.sagepub.com/content/early/2015/08/29/0146621615602855.abstract

[Phil Chalmers]

Hi Nathan,

Thank you for your kind words and for the reference. Some side work has been done on these types of models with mirt, though I'm not at liberty to discuss the details as I'm not the one spearheading the project. 

That being said, this parametrization looks like it will fit very well within the current code-base, so it's probably worth exploring. I'm going to be a little busy over the next few months (bogged down with my own obligations and work), so I probably won't have time to work on it too seriously. However, adding models to mirt with pure R code is now extremely straightforward (basically all done in the 02-item_methods.R file, where a working stub has been provided), so pull requests/collaboration for developing new IRT models are always welcome. Cheers.

Phil 

[Garett Howardson]

Not to be a curmudgeon, but this statement does not sound very open-source and/or 'R like." Is mirt planning to move toward private source code? I most certainty understand that people have to publish and in the applied measurement field your code often sets you apart from the competition, but this seems like a very fuzzy boundary to cross. It seems that if you have push toward full open source than an individuals using any pieces of mirt should also have to abide by the same open source guidelines. There are, of course, different around ways around the publication/competition paradigm but it doesn't seem like mirt has been in that business before. 

[Seongho Bae]

Hello, Garett.

Please calm down. The mirt is under GPL 3 License. That means this software is fully free software when born to forever.

Mx. Phil just requested code commit using 'GitHub' ecosystem for preventing crash among codes and more improvements with debates. Cheers.

--

Seongho Bae

2015년 9월 8일 화요일 오후 6시 21분 26초 UTC+9, Garett Howardson 님의 말:

[Phil Chalmers]

Apologies if my previous message alluded to a closed-source version of the package. No, I plan to keep future versions public as I find the distribution of quasi-open source packages for R really weird and off-putting.

As an aside about how others choose to develop software, I really can't stop people from modifying the package source code for their own purposes, and I think we all have to respect that some people just don't want their code to be open-sourced (especially at the outset and prior to publication of their ideas). All we can do is hope that after developers are finished writing and publishing they will want to contribute back to open-source projects like mirt. I certainly have used private versions of the code-base in the past for my own work and experiments (and still do) but always with the intention to merge things back if/when they work out. IMHO, not every stage has to publically transparent throughout the writing and process, but I would hope the end-goal is to make it that way after the details/publication specifics are all nailed down. Cheers.

Phil

[Garett Howardson]

Thank you, Phil, for your response. I had not intended that to be an inflammatory post but it seems like some may have interpreted it as such, so I apologize. I see now what you were saying in the earlier message and that was clearly a misinterpretation on my part, apologies. I'm still trying to understand some of the subtleties of the applied measurement literature (not my primary field) and you've definitely shed some light on that for me, thanks!

Best, 

Garett 

[Brenton Wiernik]

I was just wondering what the status of adding the MGGUM to mirt was? Has any progress been made since your 2014 post? There have now been several dissertations that have examined estimation procedures for the MGGUM, in addition to the APM paper linked to previously.

https://smartech.gatech.edu/handle/1853/53411
https://www.ideals.illinois.edu/handle/2142/46897

http://journals.sagepub.com/doi/abs/10.1177/0146621615602855

[Phil Chalmers]

We're looking to release a beta version of this model sometime during the summer. So just a little longer I suppose.....

Phil

--
You received this message because you are subscribed to the Google Groups "mirt-package" group.

To unsubscribe from this group and stop receiving emails from it, send an email to mirt-package+unsubscribe@googlegroups.com.

[Brenton Wiernik]

Great to here. Looking forward to it!