SEM Output Model Test User Model = 0 / Incorrect model specification?

[Ash Smith]

Hi! I'm new to R and SEM, my output doesn't seem correct. Thanks in advance for any guidance you can provide!

I've specified my model and tried to run with my full dataset, dataset with only variables of interest, and a covariance-variance matrix of my data, and I'm still getting the same output, where my Model Test User Model test statistic and df both = 0, and the statistics comparing the User Model and Baseline Model are either 0 or 1. 

So I assume that I haven't correctly specified the null model for lavaan to compare to my hypothesized model? 

mmodel <- '
   # regressions
     TOSCA_total_shame.norm ~ 1 + HHRDS_mean.norm + RecruitmentSite.logical
     DSISS.total.binary ~ 1 + TOSCA_total_shame.norm + HHRDS_mean.norm + RecruitmentSite.logical
   # covariance
     HHRDS_mean.norm ~~ RecruitmentSite.logical'
 fit.mmodel <- sem(mmodel, data = shame_SI_nmmmodeldata, model.type = "sem")
 summary(fit.mmodel, fit.measures=TRUE)

OUTPUT

lavaan 0.6-12 ended normally after 38 iterations

  Estimator                                         ML
  Optimization method                           NLMINB
  Number of model parameters                        14

  Number of observations                           133

Model Test User Model:
                                                     
  Test statistic                                 0.000
  Degrees of freedom                                 0

Model Test Baseline Model:

  Test statistic                                35.359
  Degrees of freedom                                 6
  P-value                                        0.000

User Model versus Baseline Model:

  Comparative Fit Index (CFI)                    1.000
  Tucker-Lewis Index (TLI)                       1.000

Loglikelihood and Information Criteria:

  Loglikelihood user model (H0)               -328.007
  Loglikelihood unrestricted model (H1)             NA
                                                     
  Akaike (AIC)                                 684.014
  Bayesian (BIC)                               724.479
  Sample-size adjusted Bayesian (BIC)          680.195

Root Mean Square Error of Approximation:

  RMSEA                                          0.000
  90 Percent confidence interval - lower         0.000
  90 Percent confidence interval - upper         0.000
  P-value RMSEA <= 0.05                             NA

Standardized Root Mean Square Residual:

  SRMR                                           0.000

Parameter Estimates:

  Standard errors                             Standard
  Information                                 Expected
  Information saturated (h1) model          Structured

Regressions:
                           Estimate  Std.Err  z-value  P(>|z|)
  TOSCA_total_shame.norm ~                                    
    HHRDS_mean.nrm            1.061    0.506    2.098    0.036
    RcrtmntSt.lgcl            0.096    0.183    0.525    0.599
  DSISS.total.binary ~                                        
    TOSCA_ttl_shm.           -0.156    0.035   -4.469    0.000
    HHRDS_mean.nrm           -0.275    0.207   -1.329    0.184
    RcrtmntSt.lgcl           -0.202    0.074   -2.734    0.006

Covariances:
                     Estimate  Std.Err  z-value  P(>|z|)
  HHRDS_mean.norm ~~                                    
    RcrtmntSt.lgcl      0.005    0.008    0.602    0.547

Intercepts:
                   Estimate  Std.Err  z-value  P(>|z|)
   .TOSCA_ttl_shm.    2.745    0.349    7.867    0.000
   .DSISS.ttl.bnry    1.168    0.170    6.859    0.000
    HHRDS_mean.nrm    0.652    0.016   41.539    0.000
    RcrtmntSt.lgcl    0.489    0.043   11.275    0.000

Variances:
                   Estimate  Std.Err  z-value  P(>|z|)
   .TOSCA_ttl_shm.    1.109    0.136    8.155    0.000
   .DSISS.ttl.bnry    0.180    0.022    8.155    0.000
    HHRDS_mean.nrm    0.033    0.004    8.155    0.000
    RcrtmntSt.lgcl    0.250    0.031    8.155    0.000

[Jasper Bogaert]

I have three comments/remarks:

1. Am I correct to assume that TOSCA_total_shame.norm  and DSISS.total.binary are observed variables? In that case you are performing a path analysis (with observed variables) and your model is just saturated. This implies that your fit is perfect, as can be seen from the fit measures.

2.  You use an endogenous binary variable, so I would suggest using the ordered argument (see e.g. https://lavaan.ugent.be/tutorial/cat.html). Not sure whether it is also necessary for path analysis with observed variables, but I just wanted to mention this,

3. I am not sure whether and how to use the model.type argument in the sem function (you might want to try to leave it out once).

If your goal is performing path analysis with observed variables, it seems to be ok to me. 

Hope this helps.

Best wishes,

Jasper

Op zaterdag 30 juli 2022 om 02:25:30 UTC+2 schreef ash.m.s...@gmail.com: