Upon successful completion of the course, students will be able to apply a statistical technique called linear mixed-effects modeling (including multilevel analysis and hierarchical linear modeling) to address specific research questions. Students will be able to recognize, evaluate, and interpret various types of parameter estimates, and be able to judge the appropriateness of the use of mixed-effects models presented in the scientific literature. Students who successfully complete the course will be able to describe the theoretical underpinnings of mixed-effects models, utilize these methods in their subsequent research, and report such analyses and their results in scientific journal articles.
|
|
How does error-related slowing in a Stroop task change over time in participants having received brain stimulation versus controls? How does daily drinking on weekdays versus weekends differ between females and males? How do developmental trajectories in a longitudinal study differ between children with versus without ADHD? How are income level of school districts and teachers' teaching style related to students' math scores?
These are all examples of research questions that are analyzed with so-called linear mixed-effects models (LMMs; closely related terms are multilevel analysis and hierarchical linear modeling), as they involve one dependent variable and so-called "clustered errors" (due to repeated measures or hierarchical "nesting" of data such as students "nested" in schools). After having successfully finished this class, you will be able to use this type of statistical model in your own research and report such analyses and their results in scientific articles. This relatively novel analysis approach is quickly being adopted in virtually all fields of psychology and other behavioral sciences because they are flexible, powerful, can handle well unbalanced data and missing observations and, compared to more traditional analyses, avoid inflated Type I errors and allow to make inferences about processes that traditional analysis cannot.
You will also learn the theoretical basis of linear mixed-effects models, and their use not only with normally distributed data, but also other data types (e.g., binary). The main statistical packages in R that will be used are lme4 and brms.
Teaching methods: Lectures, computer lab sessions, and homework assignments.
Exam information
Course grades are based on the results of (a) the take-home exam (80%), which requires students to perform the appropriate analyses, and write up the results in APA style; and (b) weekly homework assignments (20%). The homework assignments will each be graded as pass/fail and the total points earned for this component will be determined as a function of the proportion of assignments that were successfully completed.
|
|
|
|
|
|
|