Tables for sem in Stata

1 Introduction

A quick demo of making tables with sem in Stata.

2 Get Data

Data from Multilevel Thinking

use "https://github.com/agrogan1/multilevel-thinking/raw/main/simulate-and-analyze-multilevel-data/simulated_multilevel_data.dta", clear

3 Run sem Model

Outcome is a function of physical punishment, parental warmth and the intervention. Physical punishment is potentially an outcome of the intervention.

sem (outcome <- physical_punishment warmth intervention) ///
(physical_punishment <- intervention) // sem model

est store M1 // store estimates from this model

Endogenous variables
  Observed: outcome physical_punishment

Exogenous variables
  Observed: warmth intervention

Fitting target model:
Iteration 0:  Log likelihood = -23247.767  
Iteration 1:  Log likelihood = -23247.767  

Structural equation model                                Number of obs = 3,000
Estimation method: ml

Log likelihood = -23247.767

------------------------------------------------------------------------------------------
                         |                 OIM
                         | Coefficient  std. err.      z    P>|z|     [95% conf. interval]
-------------------------+----------------------------------------------------------------
Structural               |
  outcome                |
     physical_punishment |  -.9422686   .0831977   -11.33   0.000    -1.105333   -.7792041
                  warmth |   .8360494   .0599923    13.94   0.000     .7184666    .9536322
            intervention |   .6706488   .2266906     2.96   0.003     .2263434    1.114954
                   _cons |   51.49974   .3380981   152.32   0.000     50.83708     52.1624
  -----------------------+----------------------------------------------------------------
  physical_punishment    |
            intervention |  -.0513873   .0497018    -1.03   0.301    -.1488011    .0460265
                   _cons |   2.503555   .0345895    72.38   0.000     2.435761    2.571349
-------------------------+----------------------------------------------------------------
           var(e.outcome)|   38.43413    .992365                      36.53753    40.42919
var(e.physical_punishm~t)|   1.850885   .0477897                       1.75955    1.946962
------------------------------------------------------------------------------------------
LR test of model vs. saturated: chi2(1) = 0.04                        Prob > chi2 = 0.8347

4 Make Table With etable

As long as the variables have variable labels, etable (with a few options) will automatically make a nice regression table.

etable, ///
estimates(M1) /// using these estimates 
showstars showstarsnote /// show significance stars & note
showeq // show equations

                                        1    
---------------------------------------------
beneficial outcome                           
  physical punishment in past week  -0.942 **
                                   (0.083)   
  parental warmth in past week       0.836 **
                                   (0.060)   
  recieved intervention              0.671 **
                                   (0.227)   
  Intercept                         51.500 **
                                   (0.338)   
physical punishment in past week             
  recieved intervention             -0.051   
                                   (0.050)   
  Intercept                          2.504 **
                                   (0.035)   
/                                            
  var(e.outcome)                    38.434   
                                   (0.992)   
  var(e.physical_punishment)         1.851   
                                   (0.048)   
Number of observations                3000   
---------------------------------------------
** p<.01, * p<.05