estat gof not valid after -GSEM-

[Anat Tchetchik]

Dear all,
I have run a simple gsem model (3 measurement models and two structural)

gsem (  prventive -> q18_7 q18_8 q18_9..,)(protective -> q18_1 q18_2 q18_3 ..,)  (aoc -> q5_1 q5_2 q5_3, ) (aoc <-   finm18 q38_6 news_sev_day) (prventive <- aoc)(protective <- aoc), covstruct(_lexogenous, diagonal) latent(prventive aoc protective )

,. The model has converged ,all coefficients are significant
To obtain goodness of fit measures I ran :

estat gof, stats(chi2 rmsea ic all)

, and received

estat gof not valid

, any idea what might have been wrong?
Anat

[Richard Williams]

estat gof is not a supported option after gsem, presumably because many of the measures it computes would not be possible and/or appropriate. Instead, try

estat lcgof

Also see

help gsem postestimation

[Andrew Musau]

The following thread may be useful in explaining why you do not have goodness of fit statistics, in case you need to justify not reporting them.
https://www.statalist.org/forums/for...-modeling-gsem

[Anat Tchetchik]

Thank you very much Richard. I referred to gsem postestimation manual, the code

estat lcgof

returns the same error (not valid). I can only retrieve the information criteria AIC BIC using

 estat ic

Does it mean that I won't be able to compare GOF of different specification as in SEM (i.e. using CFI, TLI, RMSEA , SRMR etc.), but only based on the ll (model) and the information criteria?

[Richard Williams]

What version of Stata do you have? I’m not sure how long estat lcgof has been around.

[Anat Tchetchik]

I use Stata 15.1.. by the way, all the indicators I'm using in the measurement models are ordered categorical. I'm not sure if I have to use 'ologit' option in the measurement specifications (the latent variable itself is continuous).
In R, just assign them as "ordered", to let R "know" that they are ordered..

[Assane Diouf]

Hello, Iam working with stata 16 and I received the same msg: "estat gof not valid" after ologit. Thank you in advance for your assistance, best

[Andrew Musau]

Assane Diouf, have a look at the Stata Journal article https://journals.sagepub.com/doi/pdf...867X1701700308 and the accompanying command ologitgof.

search ologitgof

[Assane Diouf]

Thank you Andrew, well appreciated