Table 1.
Solution | Log likelihood | Bayesian information criterion | Entropy | Lo–Mendell–Rubin Likelihood Ratio Testa
|
|||
---|---|---|---|---|---|---|---|
2 Times the log likelihood difference | p value | ||||||
3-class | −7915.5 | 16,701.0 | 0.917 | 561.9 | .01 | ||
4-class | −7,785.6 | 16,733.1 | 0.906 | 259.9 | .61 | ||
| |||||||
Results of 3-class solution | Class 1 | Class 2 | Class 3 | ||||
| |||||||
Based on estimated model | 67.2 (17.4%) | 149.8 (38.7%) | 170.0 (43.9%) | ||||
Based on most likely class membership | 66 (17.1%) | 150 (38.8%) | 171 (44.2%) |
Lo–Mendell–Rubin likelihood ratio test (LMR LRT) is a test of probability that the data were generated by a model with one fewer latent class.