Re: M2 statistics

[Yuen Wan Ho]

In your paper (mirt: A Multidimensional Item Response Theory Package for the R Environment), you used mirt( ) function to conduct the full-information factor analysis. In that example, results of G2 statistics and fit indices (e.g., RMSEA) were provided. I performed mirt ( ) function and used 2PL model to examine the goodness of model fit a 10 item scale (in dichotomous response format). I could not see G2 statistics and the value of RMSEA but values of Log-liklihood, AIC and BIC. I want to know if I need to type in other commands that can result chi-square statistics and /or G2 statistics to evaluate the model fit of the scale? I also saw the command for M2 statistics, I did type in and it gave the M2 statistics and fit indices, is it a big problem when I found M2 statistics was statistically significant but fit indices showed the good model fit (RMSEA = 0.05, CFI = .94, TLI = .03)?

Thank you very much!

[Phil Chalmers]

The G2 statistic provided in the paper is no longer supported because the data tables were too sparse (the paper is also quite dated as much has been changed since then). M2 is really the only suitable statistic to use.  

Phil

On Mon, Aug 1, 2016 at 10:10 AM, Yuen Wan Ho <spoo...@gmail.com> wrote:

In your paper (mirt: A Multidimensional Item Response Theory Package for the R Environment), you used mirt( ) function to conduct the full-information factor analysis. In that example, results of G2 statistics and fit indices (e.g., RMSEA) were provided. I performed mirt ( ) function and used 2PL model to examine the goodness of model fit a 10 item scale (in dichotomous response format). I could not see G2 statistics and the value of RMSEA but values of Log-liklihood, AIC and BIC. I want to know if I need to type in other commands that can result chi-square statistics and /or G2 statistics to evaluate the model fit of the scale? I also saw the command for M2 statistics, I did type in and it gave the M2 statistics and fit indices, is it a big problem when I found M2 statistics was statistically significant but fit indices showed the good model fit (RMSEA = 0.05, CFI = .94, TLI = .03)?

Thank you very much!

--
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.