8 Random Coefficients Model 2: Summary

8.1 Model 2: Summary

We note SPSS gives Warning while fitting the Model 2.

As we check the estimates of variance components, we can see the issues of estimating variance components.

8.2 Interpretation

We have explored LMM incorporating random intercept and slope as well as with and without considering the correlation between random intercept and slope. Model 1 shows that the model is over-parameterised with the correlation between intercept and slope equals to 1.00. When we excluded the correlation term, the model did not converge as well as the estimate of random intercept equals to zero. We conclude that we do not need random slope term for the model. Hence, a model with random intercept of patients, as we fitted earlier, is adequate for this data.

If the model converges, then we should decide if we need more complex model based on model selection criteria for the random effects as we discussed earlier.