Model Summary (generalized linear mixed models)

Figure 1. Model Summary
Model Summary

The Model Summary views provide some statistical evidence that the linear mixed model is better than the linear regression.

  • Based on the information criteria, the linear mixed model with two random intercepts is preferred over the linear regression model because it has smaller AICC (10793.793 < 11053.347) and BIC (10811.765 < 11059.006) values.
  • Based on a likelihood ratio test, the linear mixed model with two random intercepts is still preferred. Under the null hypothesis that the variances for the two random intercepts is zero, the difference in the −2 restricted log likelihoods (−2LL) for the two models has a chi-square distribution with degrees of freedom equal to the difference in the number of model parameters. The linear mixed model has a −2LL of 10787.782 and 12 parameters; the linear regression has a −2LL of 11051.344 and 10 parameters. The p value for the likelihood ratio test is the probability that a chi-square random variate with 12−10 = 2 degrees of freedom is larger than (11051.345 − 10787.782 = 263.563), which is near 0.

Next