Using the Constrained Ordinal Models for Likert-Based Outcomes

• Activity Number: 499
• Type: Contributed
• Date/Time: Wednesday, August 7, 2013 : 8:30 AM to 10:20 AM
• Sponsor: Section on Statistics in Epidemiology
• Author(s): Ana W. Capuano*+ and R. William Field and Marizen R. Ramirez and Jeffrey D. Dawson
• Companies: Rush University Medical Center and The University of Iowa and The University of Iowa and The University of Iowa
• Keywords: ordinal ; proportional odds model ; logistic regression ; likert ; trend odds model ; violence

Abstract

Likert scales are commonly used in observational studies, such as surveys involving the frequency of school violence, an important public health issue associated with adverse health outcomes. We demonstrate how the proportional odds model and the trend odds model can be applied to data measured in Likert scales, allowing for random cluster effects. We use survey data from 28,882 students in 81 middle schools from a large urban school district to examine simple and adjusted effects of binary, continuous and multinomial predictor variables that have been found to be associated with violence exposure. We show, for example, that the trend odds model indicates that children with disability have a higher risk of severe violence, and that the odds is more pronounced (e.g., a cumulative odds ratio that increases by 66%) at the higher Likert levels. Different latent distributions of the underlying Likert scale distributions are also presented and discussed.