Mixed effect models

Mixed effect models have both main effects and random effects; the random effect is what distinguishes this type of model. If your data are hierarchical, involve repeated measurements, or are otherwise nonindependent, then a mixed effect model might be appropriate.

 
 
This case study by Dr David Delaney explains how he used general and generalised mixed effect models to test whether painted turtles adjust their nesting beh...
 

Case Study: Mixed models used to assess nesting behavior of painted turtles.

In this guest post, Dr. David Delaney explains how he used mixed effect models to experimentally test whether painted turtles select nest site locations in a risk-sensitive way. Dr Delaney measured the distance away from the river (continuous measure) and whether or not the nest was successful (binary measure) before and after a simulated predation attempt. A random effect of turtle identity was fitted as a random effect to account for the repeated measures for each female (i.e. before and after).

Additional Resources

StatsTree - Mixed Models

Whitlock & Schluter - The Analysis of Biological Data

Chapter 15: pages 481-486, Chapter 16: page 623 [Sapling Ch 15, Sapling Ch 16]

 
 

Mixed Effects Modeling in R

Intermediate: Code example on mixed effect modeling in R.

 

Mixed effects models in R

Intermediate: Code tutorial video on mixed effects models in R.

 
 

Linear mixed effects models

Intro: When to choose mixed-effects models, how to determine fixed effects vs. random effects, and nested vs. crossed sampling designs. Requirements… [Read More]

 

 Linear Mixed Models

Intermediate: Documentation on mixed effects models in R and S-PLUS.

 
 

Review Questions

 

  1. A geneticist is interested in whether different fruit fly species- Drosophila melanogaster, D. suzukii, and D. simulans- have different hind-leg lengths. In this example, is the fruit fly species a fixed effect, or a random effect?

  2. The same geneticist has 20 different inbred lines from each of the above species (named Dmel1, Dmel2, Dmel3…, etc) that they have maintained in lab. If the geneticist is interested in the same question as above- do the species have different leg lengths?- is the inbred line # a fixed effect, or a random effect, in this case?

  3. The geneticist measured legs of 10 females from each inbred line (described above). Write an “R formula” for a linear model designed to analyze these data.

The Next Steps

Confused?

Let’s move down the tree and review these concepts.