Section 4 Review of OLS

This section will provide a review of OLS.

OLS is the workhorse of empirical political science. We will learn a lot beyond OLS, but OLS is often “good enough” and sometimes preferable for explaining the relationship between variables. That is to say, MLE will expand your toolkit, but OLS should remain a big part of your toolkit.

I recommend that you review the following readings to familiarize yourself with regression. I will make note within this section where particular readings are most relevant. These readings are available on Canvas in the modules- Week 1 section.

  • Wheelan, Charles. 2012. Naked Statistics. W.W. Norton. Chapter 11. This provides an accessible overview of regression and the interpretation of regression results.

  • Gelman, Andrew, and Jennifer Hill. 2006. Data analysis using regression and multilevel/hierarchical models. Cambridge University Press. Chapter 3. This is a slightly more technical overview and includes some R code for running regressions.

  • Building models and breaking models.

    • (Optional) Fox, John. 2015. Applied Regression Analysis and Generalized Linear Models, 2nd Edition. Sage. Chapter 11. This reading describes diagnostic tests to probe whether the model is a good fit of the data. We won’t go into detail about this in this class, but is material classes focused on linear regression will generally cover.
    • Messing, Solomon. “How to break regression.”
    • Lenz, G., & Sahn, A. (2020). “Achieving Statistical Significance with Control Variables and Without Transparency.” Political Analysis, 1-14. doi:10.1017/pan.2020.31. This paper talks about how to build a regression model, and in particular, why adding more and more controls isn’t always a good thing.