USER’S MANUAL
2. How to interpret the results
Table 1. Model fit of the result
Model Fit 

FIT 
.535 
AFIT 
.532 
GFI 
.993 
SRMR 
.078 
FIT_M 
.606 
FIT_S 
.168 
Table 1. provides six measures of model fit.
 FIT indicates the total variance of all variables explained by a model specification. The values of FIT range from 0 to 1. The larger this value, the more variance in the variables is accounted for by the specified model.
 AFIT(Adjusted FIT) is similar to FIT, but takes model complexity into account. The AFIT may be used for model comparison. The model with the largest AFIT value may be chosen among competing models.
 (Unweighted leastsquares) GFI and SRMR (standardized root mean square residual). Both are proportional to the difference between the sample covariances and the covariances reproduced by the parameter estimates of generalized structured component analysis. The GFI values close to 1 and the SRMR values close to 0 may be taken as indicative of good fit.
 FIT_M signifies how much the variance of indicators is explained by a measurement model.
 FIT_S indicates how much the variance of latent variable is accounted for by a structural model. Both FIT_M and FIS_S can be interpreted in a manner similar to the FIT.
Table 2. Parameter Estimates of Loadings
Loadings 


Estimate 
SE 
95% 
95% 
cei1 
.781 
.025 
.716 
.831 
cei2 
.825 
.021 
.786 
.854 
cei3 
.770 
.029 
.703 
.820 
cei4 
.804 
.030 
.727 
.868 
cei5 
.801 
.028 
.741 
.846 
cei6 
.843 
.028 
.796 
.883 
cei7 
.776 
.024 
.728 
.824 
cei8 
.801 
.033 
.705 
.856 
ma1 
.787 
.026 
.731 
.825 
ma2 
.758 
.025 
.700 
.805 
ma3 
.637 
.039 
.547 
.704 
ma4 
.823 
.028 
.761 
.803 
ma5 
.811 
.024 
.757 
.849 
ma6 
.743 
.040 
.675 
.793 
orgcmt1 
.748 
.036 
.675 
.795 
orgcmt2 
.790 
.024 
.747 
.834 
orgcmt3 
.820 
.020 
.773 
.857 
orgcmt7 
.707 
.033 
.620 
.778 
orgcmt5 
.796 
.029 
.721 
.842 
orgcmt6 
.709 
.050 
.597 
.796 
orgcmt8 
.782 
.030 
.726 
.843 
Table 2 displays the estimates of loadings and their bootstrap standard errors (SEs), and 95% confidence intervals. (Note: when indicators are formative, their loading estimates will not be reported)
Table 3. Parameter Estimates of weights
Weights 


Estimate 
SE 
95% 
95% 

cei1 
.150 
.010 
.132 
.170 

cei2 
.160 
.010 
.141 
.179 

cei3 
.157 
.010 
.140 
.177 

cei4 
.147 
.009 
.126 
.163 

cei5 
.162 
.011 
.143 
.184 

cei6 
.168 
.009 
.153 
.191 

cei7 
.150 
.009 
.130 
.167 

cei8 
.154 
.009 
.140 
.168 

ma1 
.219 
.021 
.182 
.263 

ma2 
.211 
.020 
.175 
.252 

ma3 
.194 
.017 
.163 
.233 

ma4 
.261 
.019 
.228 
.293 

ma5 
.237 
.019 
.199 
.267 

ma6 
.184 
.021 
.150 
.238 

orgcmt1 
.302 
.018 
.263 
.348 

orgcmt2 
.330 
.016 
.296 
.371 

orgcmt3 
.364 
.019 
.337 
.400 

orgcmt7 
.303 
.018 
.257 
.344 

orgcmt5 
.453 
.025 
.403 
.500 

orgcmt6 
.387 
.026 
.328 
.441 

orgcmt8 
.467 
.031 
.420 
.529 

Table 3 displays the estimates of weights and their bootstrap standard errors (SEs), and 95% confidence intervals
Table 4. Parameter estimates of path coefficients
Path 


Estimate 
SE 
95% 
95% 
LV_1 
.362 
.067 
.260 
.465 
LV_2 
.614 
.042 
.549 
.678 
LV_2 
.404 
.049 
.515 
.288 
Table 4. shows the estimates of path coefficients and their bootstrap standard errors (SE) and 95% confidence intervals.
Table 5. R squares of latent variables
R squares of Latent Variables 

LV_1 
0 
LV_2 
.131 
LV_3 
.377 
LV_4 
.163 
Table 5 provides the R square values of each “endogenous” latent variable, indicating how much variance of an endogenous latent variable is explained by its exogenous latent variables. In the present example, the first latent variable (LV_1) is exogenous, so that its R square is zero.
Table 6. Reliability and validity of measures

Cronbach’s alpha 
DillonGoldstein’s rho 
AVE 
Number of eigenvalues greater than one 
LV_1 
.8242 
.8956 
.7414 
1 
LV_2 
.7173 
.8419 
.6404 
1 
LV_3 
.7492 
.8567 
.6659 
1 
LV_4 
.6418 
.8071 
.5828 
1 
In Table 6 Cronbach’s alpha and DillonGoldstein’s rho (or the composite reliability) can be used for checking internal consistency of indicators for each latent variable. The average variance extracted (AVE) can be used to examine the convergent validity of a latent variable. The number of eigenvalues greater than one per block of indicators can be used to check unidimensionality of the indicators.
Table 7. Correlations of Latent Variables (SE)
Correlations 


LV_1 
LV_2 
LV_3 
LV_4 
LV_1 
1 



LV_2 
.362 
1 


LV_3 
.388 
.614 
1 

LV_4 
.209 
.404 
.461 
1 
Table 7 shows the correlations among latent variables.