What is a hierarchical model?
Hierarchical and Mixed Effects Models in R
Richard Erickson
Data Scientist
Course overview
- Components for mixed-effect models
- Applying and interpreting linear mixed-effect models
- Generalized linear mixed-effect models
- Repeated measure models
Why do we use a hierarchical model?
- Data nested within itself
- Pool information across small sample sizes
- Repeated observations across groups or individuals
Other names for hierarchical models
- Hierarchical models: Nested models, Multi-level models
- Regression framework
- "Pool" information
- "Random-effect" versus a "fixed-effect"
- "Mixed-effect" (linear mixed-effect model; LMM)
- Linear mixed-effect regression (lmer)
- Repeated sampling: "Repeated-measures", "Paired-tests"
School test scores
Meta-data:
- Gain in math scores for individual students from kindergarten to 1st grade
- Part of a national-level assessment in US
- Subset of data from West, Welch, and Galecki
Student-level variables:
- Student ID:
childid - Math test-score gain:
mathgain - Math kindergarten score:
mathdind - Student's sex:
sex - Student's minority status:
minority
School test scores
Classroom-level variables:
- Classroom id:
classid - Teacher's math training:
mathprep - Teacher's math test knowledge test score:
mathknow - Teacher's years teaching:
yearstea
School-level variables:
- School ID:
schoolid - School's household poverty level:
housepov - School's socioeconomic status:
ses
Let's practice!
Hierarchical and Mixed Effects Models in R
