Fixed effects, random coefficients, and centering

Steve Raudenbush, Lewis-Sebring Professor of Sociology, Chairman of the Committee on Education at the University of Chicago, and main author of the well-known HLM program, discusses the fixed effect random coefficient (FIRC) model and its implications for causal inference for multilevel data. A new feature in HLM, this feature is computationally convenient, with no dummy variables and no centering. The Tennessee STAR data is used as illustration in this 45 minute webinar.

This is the first in a series of webinars by Professor Raudenbush. A second webinar, discussing the use of the newly implemented profile likelihood plots feature in HLM in inference for variance components in hierarchical linear modeling, will be available soon.