Comparison between LD-Matrix in "mirt" and in IRT PRO

[yingyi....@gmail.com]

Hi Phil and everyone,

I have been trying to do the same analysis in IRT PRO and with "mirt" packages in R. 

In IRT PRO, it is suggested that LD > 10 would be a suspect of local dependence; while I am wondering what would be the cut-off value in R? 

I used the same mirt packege codes updated on March 2018, page 134. 

.residuals(data, suppress = .15)

I am wondering if suppress=.15 here is a recommended threshold? Is there any reference or rationale that I could track down?

Thanks in advance,

Yingyi

[Phil Chalmers]

The value is a correlation, which I think has a better overall interpretation than the cut-off of LD > 10 setup (where exactly did that come from anyway?). Standardized correlation residuals have a common lineage with traditional factor analysis, though the actual magnitude may have different meaning. 

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.
For more options, visit https://groups.google.com/d/optout.

[Phil Chalmers]

Who knows, I don't use IRTPRO so I'm not sure what exactly it reports/does. I just followed the Chen and Thissen article, and applied my own correlation statistic based on standardizing tables. 

Phil

On Wed, Mar 28, 2018 at 3:40 PM, <yingyi....@gmail.com> wrote:

The LS matrix tables are different (in terms of the numbers) than the mirt package in R, though I am using the same analysis. 

--

[yingyi....@gmail.com]

Hi Phil Thanks for the reply. Correlation makes sense here! the "LD > 10" comes from IRT PRO documentation, wherein I think it reported the standardized LD chi-square statistics. 

Thanks again! This is super helpful for me to compare the 2 softwares! :) 

Best,

Yingyi

Phil

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

[Fulya barış]

Hi Phil !

I'm trying to detect local dependency with LD matrix in R

I couldn't something

upper triangle of that matrix is correlation coefficient of residuals isn't it ?

lower triangle of that matrix LD values ?.. How can I interpret that values ?.. are they chi square values ?.. or do ı need to interpret them for local dependency ?

thanx

Phil

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