I ran an LGMM in which 3 covariates were used to predict class membership in a 4-class solution. Partial output is below.
At the risk of seeming ill-informed, I am confused by the output. The coefficients for the analysis suggest, for example, that AGE predicts C1 v C4 (p=.003). However, the OR below is only .924. These results seem inconsistent. What am I missing? Also, can the odds ratios, p value, and CIs be obtained directly in the output.
Categorical Latent Variables
C#1 ON AGE -0.079 0.026 -3.022 0.003 HLTH 3.169 0.518 6.115 0.000 INCHG 1.026 1.206 0.851 0.395
LOGISTIC REGRESSION ODDS RATIO RESULTS C#1 ON AGE 0.924 HLTH 23.792 INCHG 2.790
The size of the odds ratio is related to the size of the coefficient not the signficance. The odds ratio is the exponentiation of the coefficient. You can get confidence intervals for the odds ratios. Use the CINTERVAL option.