Online Program Home
My Program

Abstract Details

Activity Number: 531 - SPEED: Statistical Computing: Methods, Implementation, and Application, Part 2
Type: Contributed
Date/Time: Wednesday, July 31, 2019 : 11:35 AM to 12:20 PM
Sponsor: Section on Statistical Computing
Abstract #307956
Title: Using Information Criteria to Select Among Polynomial and “truly” Nonlinear Multilevel Models
Author(s): Wendy Christensen* and Jennifer Krull
Companies: University of California, Los Angeles and University of California, Los Angeles
Keywords: Nonlinear models; Information criteria; Multilevel modeling

Multilevel modeling is a popular analytic choice for behavioral researchers seeking to model longitudinal change. Such researchers often choose polynomial models, such as quadratic or cubic models, when the expected trend is nonlinear. While polynomial models are suitable in many situations, it is possible to fit a variety of nonlinear models. “Truly” nonlinear models, such as logistic and exponential models, have been used to model nonlinear change over time in many natural phenomena and are potentially applicable to theories about behavior. Unlike polynomial models, however, truly nonlinear models are non-nested, meaning that likelihood ratio tests cannot be used for model selection. Information criteria are a flexible and accessible method for non-nested model selection, but there are currently no guidelines for their use when truly nonlinear models are included in the set of candidate models. In this simulation study, efficient (AIC/AICC) and consistent (BIC/CAIC/HQIC) information criteria were evaluated on their ability to select among polynomial and truly nonlinear models. Specific recommendations for sample sizes (measures and individuals) for different models are provided.

Authors who are presenting talks have a * after their name.

Back to the full JSM 2019 program