cancel
Showing results for 
Show  only  | Search instead for 
Did you mean: 
Choose Language Hide Translation Bar
gail_massari
Community Manager Community Manager
Get the skinny on mixed models from developer and author Elizabeth Claassen

Mixed effects models are useful for data with more than one source of random variability. For example, when we have an outcome measured more than once on the same person, we need to account for both within-person and across-person variability.

JMP Developer Elizabeth Claassen and co-author of JMP for Mixed Models takes a practical approach to making sense of data. When she supported students at the University of Nebraska-Lincoln Statistical Consulting Lab, she frustrated those who sought a mixed model solution to a designed experiment and who just wanted a quick answer. “To determine the correctness of a mixed model, you first have to explain the relationship between the factors and response in your experiment,” she would say. “Let's start by sketching that on paper.”

On Dec. 2 and Dec. 9, Elizabeth brings this practical approach to her Developer Tutorials, where she'll share her expertise about the JMP Pro mixed model capabilities she helped develop and provide examples of when, and why, the techniques are useful. She’ll debunk (or perhaps confirm) some mixed model rumors.

Rumor 1: How do you report the results of a linear mixed model? Don't get distracted by p-values. They guide, not dictate, decision making.

Rumor 2: How do you report how variable the effect is between individuals? Look at the random effect estimates.

Rumor 3: Negative variance components do not allow for mixed-model interpretations.

Rumor 4: Are the fixed effects estimates valuable? Yes, they represent the best-guess average effects in the population.

Rumor 5: Mixed models are no good when you have missing data or unequal group sizes.

Chris Gotwalt, JMP Chief Data Scientist, also credits Elizabeth for implementing the Satterthwaite confidence intervals in the JMP Pro 16 Fit Mixed platform, making interval estimation of linear combinations of variance components much easier for JMP users. Clearly a multi-talented person, Elizabeth revealed how her outside interests relate to her passion for programming. Come on, knitting is like coding? Really?

See the videos, including discussions, from the 2021 Dec. 2 and Dec. 9 live webinars..

Last Modified: Dec 19, 2023 5:06 PM