Re: st: a model that runs in LISREL, but not in Stata 12. Any suggestions?

[Stas Kolenikov <skolenik@gmail.com>]

Have you tried fitting it with my -confa-? I doubt it is any better
than -sem- in terms of numeric stability issues... but at least it has
a more straightforward syntax :).

On Mon, Nov 14, 2011 at 12:45 PM, Jonathan Richards
<richards.868@buckeyemail.osu.edu> wrote:
> Hey everybody-
>
> This is a follow-up to some messages last week. I ran a model in LISREL, and am trying to run it in Stata 12, but I can't get the model to converge in Stata 12. The LISREL and Stata 12 code are posted below.
>
> (Cam and Richard, if you see this, thanks for your feedback on this subject last week. To answer some of the questions that you two raised:
> -The normalizations of the latent factors are the same in the LISREL and the Stata 12.
> -I'm not specifying starting values in LISREL, and it converges. In Stata 12, I've run the model without specifying initial values, and specifying many different initial values - still no luck.
> -I am giving a stratifying variable and clustering variable in both.)
>
> I'm stumped. Any suggestions would be appreciated.
>
> Thanks-
>
> Brent
>
> PS
>
> LISREL code:
>
> DA NI=60 NO=2077 MA=CM CL=12 ST=22
> RA FI='C:\Documents and Settings\Brent Richards\My Documents\dissertation\noncogs\Young Lives\UKDA-5307-stata8\stata8\forLISREL1.psf'
> SE
> 16 17 21 19 20 /
> MO NX=5 NK=2 TD=DI
> LK
> lnc5 lc5
> FR LX(2,1) LX(3,1) LX(5,2)
> VA 1 LX(1,1)
> VA 1 LX(4,2)
> PD
> OU
>
> Stata 12 code:
>
> svy: sem (LNC5@1 -> perprescmc) (LNC5 -> travalonmc) (LNC5 -> bratmc) (LCCG5@1 -> ppvtmc)
> (LCCG5 ->mathmc) ///
> (perprescmc <- _cons@0) (travalonmc <- _cons@0) (bratmc <- _cons@0) (ppvtmc <- _cons@0)
> (mathmc <- _cons@0), ///
> covstructure(e.perprescmc e.travalonmc e.bratmc e.ppvtmc e.mathmc, diag) nocons ///
> iterate(1000) latent(LNC5 LCCG5) var(e.ppvtmc, init(441)) var(e.mathmc, init(6.656)) ///
> var(e.perprescmc, init(3.37)) var(e.bratmc, init(0.18)) var(e.travalonmc, init(0.49)) ///
> method(ml)
>
>
>
>
> *
> *   For searches and help try:
> *   http://www.stata.com/help.cgi?search
> *   http://www.stata.com/support/statalist/faq
> *   http://www.ats.ucla.edu/stat/stata/
>



-- 
Stas Kolenikov, also found at http://stas.kolenikov.name
Small print: I use this email account for mailing lists only.

*
*   For searches and help try:
*   http://www.stata.com/help.cgi?search
*   http://www.stata.com/support/statalist/faq
*   http://www.ats.ucla.edu/stat/stata/