400002 SE SE Methods for Doctoral Candidates (2014S)

This course provides a practical and applied introduction to ordinary least squares (OLS) regression models, one of the most widely-used statistical methods in the social sciences. By the end of this course, you will be able to construct and interpret OLS regression models. You will have a firm understanding of the assumptions of the model, the differences between various types of independent variables and how to identify and address possible dangers and problems. You will also be able to evaluate critically OLS models used in scholarly journals.

The course will cover all the basic aspects of this technique in five sessions. We will begin by reviewing basic statistical concepts, such as comparing means and testing hypotheses, before moving on to the analysis of the association of two continuous variables. The next session will present simple linear regression and the assumptions underlying OLS regression. The final three sessions cover the core of this method. First, we examine in detail multiple regression models, concentrating on the practical interpretation of results. Then, different types of explanatory variables are introduced, with a focus on binary/nominal variables and interaction effects. In the final session, an overview of possible problems and their remedies is provided, and we will consider how to approach model-building in OLS regression.