Problem using polymirt

[Jordache Ramjith]

Please see the output. I installed the package mirt and then tried. Why can't it find polymirt?

*** arch - i386
*** arch - x64
* DONE (mirt)
> data=read.table("paul.txt", header=T)
> library(mirt)
Loading required package: stats4
Loading required package: lattice
> polymirt(d)
Error: could not find function "polymirt"

[Matthew Sigal]

Hi Jordache,

From the mirt documentation, "The function (polymirt) is currently depreciated and instead should be run by using the confmirt function."

Best,

Matt

--
Matthew J. Sigal, BA (Hons), MA
Quantitative Methods, PhD Candidate
Department of Psychology
262 Behavioural Science Building
York University, 4700 Keele St.

Toronto, ON, Canada M3J 1P3

(416) 736-2100 x66163

msi...@yorku.ca

www.matthewsigal.com / www.dfconsulting.org

--
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.

[Phil Chalmers]

Sorry for the confusion, but confmirt() and polymirt() have been completely removed from the package. It's probably best to update to the latest CRAN version and pass everything through mirt() with a suitable method argument (method = 'MHRM' will create equivalent code to confmirt and polymirt). Cheers.

Phil

[Jordache Ramjith]

Thank you Phil. I was using the paper from the journal of statistical software as it's easier to follow. I only started using R recently and only started reading up on IRT last week. I'm pretty sure there's multidimensionality in my data. The data items are basically dichotomous test scores. Simply adding the total provides such little explanation in the variation of the exam mark of these students. Wanted to see if performing regression using latent variable estimates are better. Hopefully I figure this out, but the basic question is then (at the risk of sounding stupid): Can mirt help me estimate the latent abilities for the different factors that must exist in this test? Are there any applicational examples of anything similar?

[Phil Chalmers]

No problem, the package comes with a lot of examples which can be found by flipping through the index help files or called directly using the ? operator (e.g., ?mirt). In the lastest release I've automatically generated the output from the examples, and they can be navigated here: http://philchalmers.github.io/mirt/mirt-vignettes.html.

Yes, all latent ability scores can be generated from the fscores function; see the help files for examples and details. Cheers.

Phil 

[Jordache Ramjith]

Hi, sorry just a quick Q. Shouldn't the length of the dataset correspond to the number of different estimates for each latent
ability in a multidimensional model? I somehow get a 100 less estimates for each latent ability.

[Jordache Ramjith]

Could it be because from the 539 people there are only 460 distinct responses?

[Jordache Ramjith]

Okay yes I was being stupid. I just saw the full.scores option. Thank you :-)