552 - Teaching Workshop. Teaching Regression to Undergraduates: Why It's Necessary and How It's Possible

Session Submission Type: Workshop

Abstract

Why is social statistics a required course for sociology undergraduates? There are at least three answers to this question: (1) we want our students to be able to read at least some of the quantitative professional literature in sociology, (2) we want our students to be able to engage in elementary quantitative sociological research, (3) we want to give our students skills that will improve their employability. On all three counts the most important technique for students to learn is multiple linear regression, yet our undergraduate social statistics courses seldom cover it, or at best offer only a cursory treatment of it. Instead, scarce contact time is wasted on a series of techniques and skills (non-parametric methods, contingency table analysis, tails of normal distributions, reading t tables) that are rarely used in practice.

We advocate a modernized curriculum that gets students involved in regression analysis as early as possible, with the goals of having students interpreting regression models from sociological journals and running their own regression models before the end of the semester. We illustrate why such a curriculum is necessary and show how it is possible. The bulk of our workshop focuses on techniques for introducing students to regression that can either be worked into the existing curriculum or be used as the basis for a new curriculum. Even teachers who are tied to the existing 20th century curriculum can use the teaching tools we present to increase their students' comfort with regression.

Contrary to widespread belief, regression is a highly intuitive technique, much more intuitive than methods that are currently taught to undergraduates. The existing curriculum centers on techniques that are conceptually advanced but computationally simple, since the curriculum was designed with slide rules and hand calculations in mind. Regression is computationally advanced but conceptually simple. With computers doing the calculations, students need no advanced mathematical knowledge to undertake simple regression or even multiple regression analysis. We show how a regression-based undergraduate social statistics course can focus students' minds on serious sociological research questions and help support the rest of the sociology curriculum.

Sub Unit

• Teaching Workshop

Session Organizer

• Salvatore J. Babones, University of Sydney

Leader

• Salvatore J. Babones, University of Sydney

Co-Leader

• Thomas J. Linneman, College of William & Mary