Error in D3.LRT(object, h1 = h1, useImps = useImps, asymptotic = asymptotic, : D3 test statistic could not be calculated. Try the D2 pooling method.

[Macsue Jacques]

I am using Multiple imputation data to fit parallel models. I have checked variable models individually to ensure best fit but when I combine some of the variables into a parallel model I get this error, which I am not sure what it means, or how to fix it.

Another few questions:

1- After plotting some of my variables I see that there is a positive trend and then a decline afterward, as well as extremely large variability in response within same participant. Would this lead to negative variances in time points and/or slope & intercept variance? Do I need to worry about that? Would this be considered a Heywood case as well?

2- This might be a silly question but I don't quite understand why to models that are perfectly fitted when analyzed individually lead to a bad fit when combined into a parallel model. What should I do in this case? Some of my data is extremely variable as a group as well as within the same participant, how do you deal with this issue? I've tried every different possibility  in order to get the best fit (i.e. quadratics, picewise, etc.)

3- Is also not uncommon  that when running parallel growth models with multiple imputed data I might get:

Negative test statistic set to zero 

Negative pooled test statistic was set to zero, so fit will appear to be arbitrarily perfect. 

Why is that? How do I fix this? Should I worry about overfitting if my RMSEA=0, CFI=1 etc?

Any help on this would be much appreciated!

[Macsue Jacques]

Em quarta-feira, 22 de julho de 2020 16:01:42 UTC+10, Macsue Jacques escreveu:

I am using Multiple imputation data to fit parallel models. I have checked variable models individually to ensure best fit but when I combine some of the variables into a parallel model I get this error (

Error in D3.LRT(object, h1 = h1, useImps = useImps, asymptotic = asymptotic, : D3 test statistic could not be calculated. Try the D2 pooling method.), which I am not sure what it means, or how to fix it.

[Yves Rosseel]

> combine some of the variables into a parallel model I get this error (
> Error in D3.LRT(object, h1 = h1, useImps = useImps, asymptotic =
> asymptotic, : D3 test statistic could not be calculated. Try the D2
> pooling method.), which I am not sure what it means, or how to fix it.

See ?runMI documentation in the semTools package. It would seem you need
to set test="D2".

Yves.

[macsuej...@gmail.com]

Great! Thank you!