FSS-DSOC Seminar: “A Road Map for the Regression Modeling of Ordinal Response Variables”, by Prof. Andrew S. Fullerton

Ordinal variables are measured characteristics with categories ranked from low to high on a single trait but unknown spacing between the categories. Social science datasets, such as the General Social Survey, are replete with ordinal variables.  As a result, quantitative social scientists frequently analyze response variables (i.e., outcomes) originally measured on an ordinal scale. Statisticians developed ordinal regression models in the 1970s. However, researchers continue to analyze ordinal outcomes using a number of alternative models, including binary logit, multinomial logit, and linear regression.  In light of these different approaches and the lack of justification for the decisions involved in this process, I propose a set of guidelines for researchers analyzing ordinal outcomes. This roadmap for ordered regression modeling emphasizes two key principles.  First, the theory of the data-generating process should be consistent with the regression model’s theoretical assumptions.  Second, decisions researchers make throughout the process of analyzing data should integrate empirical, practical, and theoretical considerations.  I also introduce a new theory of ordinality based on the idea of a latent, objective probability underlying many ordinal outcomes, including those based on a Likert scale.  I illustrate the benefits of this roadmap with an example from the General Social Survey (job insecurity and health).  I conclude the talk with a discussion of best practices for applied ordered regression modeling and a broader discussion of the process of publishing research in generalist sociology journals based on my experiences as the coeditor of The Sociological Quarterly (2020 to 2024) and Social Problems (2024 to present).

• How to publish in The Sociological Quarterly or Social Problems?
• What authors should do and avoid doing while submitting/writing their papers?
• Q&A

Andrew Fullerton is the Professor of Sociology at Oklahoma State University (OSU). He joined the OSU Sociology department in 2007 after completing his Ph.D. in Sociology at the University of Connecticut (2007). He also has a MA in Survey Research from the University of Connecticut (2003) and a MA in Sociology from Indiana University (2001). His research has been published in journals such as Social Forces, Social Problems, Sociological Methods and Research, and Social Science Research. He has also published a book on ordered regression models with Chapman and Hall/CRC and a book on neoliberalism and insecurity with Routledge. Prof. Fullerton teaches courses in Quantitative Methods, Social Stratification, and Social Problems. He currently serves as the co-editor of Social Problems.