Modeling Mixed Effects for Binary and Count Response Data using Generalized Linear Mixed Models (GLMM)
jiancao
| Updated on
Generalized Linear Mixed Models were introduced in JMP Pro 17, where you specify two distributions - Binomial and Poisson.
GLMM combines two approaches: the linear mixed model and generalized linear model frameworks . GLMM is useful for three types of model structures:
- Randomized complete and incomplete block designs
- Split-plot experiments
- Random coefficient models
See how to:
- Model mixed effects for count data using Poisson Regression and interpret results
- Model mixed effects for grouped data using Binary Logistic Regression and interpret results
- Model mixed effects for individual data using Binary Logistic Regression and interpret results
Questions answered by Jian Cao @jiancao and Byron Wingerd @Byron_JMP during the live webinar demo:
Q: How do you assess performance of GLMM?
A: Here are the full details.
Q: Would failure to include the random effects necessarily always lead to an increase in possibliity of overdispersion?
A: The random effects, when included, make it less likely that you will over- or under-estimate.
Q: So since we have random effects why would we not always use the conditional prediction?
A: We usually want the marginal model profiler. The Conditional Model Profiler goes into some more detail on factors without doing a separate analysis.
Q: Is GLMM supported within MSA platform?
A: The data generated from the MSA builder could be analyzed with GLMM, but it is not automatically integrated with the variability platform or MSA.
Q: Does the order of the Y’s matter? I’m guessing the numerator is the top number.
A: Yes.
Q: Are repeated measures designs specified the same way as in the mixed model platform?
A: If you mean mixed model platform not GLMM, see https://www.jmp.com/support/help/en/17.1/index.shtml#page/jmp/statistical-details-for-repeated-measu....
Resources
- Documentation and examples of JMP Pro Generalized Linear Mixed Models (GLMM)
- Developer Tutorial video on GLMM
