[R] No fit statistics for some models using sem

John Fox jfox at mcmaster.ca
Wed Mar 7 20:02:17 CET 2007


Dear David and Ista,

I haven't looked at this model carefully, but the fact that the df are
0 suggests that the model is just-identified and therefore necessarily
perfectly reproduces the covariances among the observed variables.
Removing a parameter would over-identify the model, making possible the
computation of the missing fit statistics.

Regards,
 John

On Wed, 7 Mar 2007 18:31:09 +0000
 "David Barron" <mothsailor at googlemail.com> wrote:
> It's not the correlation as such that is the problem; it's because
> you
> only have 10 degrees of freedom "available" with four observed
> variables, and you are estimating 10 parameters, which is why you get
> a chi square of zero.  When you remove any one free parameter (such
> as
> the correlation), the model becomes identified.
> 
> On 07/03/07, Ista Zahn <istazahn at gmail.com> wrote:
> > Hi,
> >
> > New to both R and SEM, so this may be a very simple question. I am
> > trying to run a very simple path analysis using the sem package.
> > There are 2 exogenous (FARSCH, LOCUS10) and 2 endogenous (T_ATTENT,
> > RMTEST) observed variables in the model.  The idea is that T_ATTENT
> > mediates the effect of FARSCH and LOCUS10 on RMTEST. The RAM
> > specification I used is
> >
> > FARSCH -> T_ATTENT, y1x1, NA
> > LOCUS10 -> T_ATTENT, y1x2, NA
> > FARSCH -> RMTEST10, y2x1, NA
> > LOCUS10 -> RMTEST10, y2x2, NA
> > T_ATTENT -> RMTEST10, y2y1, NA
> > FARSCH <-> FARSCH, x1x1, NA
> > LOCUS10 <-> LOCUS10, x2x2, NA
> > T_ATTENT <-> T_ATTENT, y1y1, NA
> > RMTEST10 <-> RMTEST10, y2y2, NA
> > LOCUS10 <-> FARSCH, x2x1, NA
> >
> > This model runs, but using the summary function does not return the
> > usual model fit statistics, only the following:
> >
> > Model Chisquare =  0   Df =  0 Pr(>Chisq) = NA
> >   Chisquare (null model) =  8526.8   Df =  6
> >   Goodness-of-fit index =  1
> >   BIC =  0
> >
> > If I omit the last line from the RAM specification(i.e., delete
> > "LOCUS10 <-> FARSCH, x2x1, NA"), I DO get all the usual statistics:
> >
> >   Model Chisquare =  1303.7   Df =  1 Pr(>Chisq) = 0
> >   Chisquare (null model) =  8526.8   Df =  6
> >   Goodness-of-fit index =  0.95864
> >   Adjusted goodness-of-fit index =  0.58639
> >   RMSEA index =  0.30029   90% CI: (NA, NA)
> >   Bentler-Bonnett NFI =  0.84711
> >   Tucker-Lewis NNFI =  0.082726
> >   Bentler CFI =  0.84712
> >   BIC =  1294.1
> >
> > My understanding is the you should always put in the correlation
> > between exogenous predictors, but when I do this I don't get fit
> > statistics. Can anyone help me understand what is happening here?
> >
> > Thank you,
> >
> > Ista
> >
> > ______________________________________________
> > R-help at stat.math.ethz.ch mailing list
> > https://stat.ethz.ch/mailman/listinfo/r-help
> > PLEASE do read the posting guide
> http://www.R-project.org/posting-guide.html
> > and provide commented, minimal, self-contained, reproducible code.
> >
> 
> 
> -- 
> =================================
> David Barron
> Said Business School
> University of Oxford
> Park End Street
> Oxford OX1 1HP
> 
> ______________________________________________
> R-help at stat.math.ethz.ch mailing list
> https://stat.ethz.ch/mailman/listinfo/r-help
> PLEASE do read the posting guide
> http://www.R-project.org/posting-guide.html
> and provide commented, minimal, self-contained, reproducible code.

--------------------------------
John Fox
Department of Sociology
McMaster University
Hamilton, Ontario, Canada
http://socserv.mcmaster.ca/jfox/



More information about the R-help mailing list