If you use these materials, please include the following citation.
MeredithRolfe, , 2013, "Multiple Regression and OLS Violations", http://opossem.org/content/multiple-regression-and-ols-violations
Type of Material:
Lecture Notes and Slides
Creative Commons License:
CC BY-NC-SA 3.0
This lecture is a continuation of the introductory regression lecture also available on OPOSSEM. It introduces multiple regression and briefly reviews the common OLS violations. As the lecture is intended for delivery to a large course, heavy use is made of IRT instant voting technology (including a competition amongst members of diferent sections.) Embedded in the lecture is an IRT based review of p values and significance tests based on published tables in articles, as students struggled to get p-values and coefficients interpreted consistently.
The purpose of this module is to introduce students to bivariate regression analysis. The module begins with an example of how the relationship between two continuous variables can be depicted using a scatterplot--in particular, how the variance in a continuous dependent variable can be decomposed into systematic and unsystematic components. Readers are introduced to the linear model, including a discussion of the Y-intercept and slope coefficients, predicted values and errors, the least-squares principle, the conditional mean, and goodness of fit. The module concludes with a discussion of hypothesis testing for bivariate regression, including significance tests for the R2 and for the unstandardized slope coefficient.