Explores ways in which data are hierarchically organized, such as voters nested within electoral districts that are in turn nested within states. Provides a basic theoretical understanding and practical knowledge of models for clustered data and a set of tools to help make accurate inferences. Prerequisite: either SOC 504;SOC 505; and SOC 506, or equivalent; recommended: either CS&SS 505 and CS&SS 506, or equivalent. Offered: jointly with CS&SS 560/SOC 560.
This is an applied data analysis course that focuses on hierarchical linear models. It explores basic statistical models such as linear and logistic regression, generalized linear models before moving on to introducing their multilevel extensions. Key issues related to simulation methods, causal inference, analysis of variance, sample size calculations, model selection and missing-data imputation will also be covered in depth. Although the underlying theory behind hierarchical models is certainly important, the numerous examples we will discuss during the lectures will be crucial. At the end of this course the students should be proficient at analyzing hierarchical data and fully understand all the statistical issues involved.
Student learning goals
General method of instruction
Class assignments and grading
The homework will count as 70% of your grade. There will be a final class project that counts for the rest of your grade. There will be no midterm or final exams.