Overview and Fundamental Hurdles
What are some common pitfalls commonly encountered when doing regression analysis? How can we detect them?
This session aims to discuss the some fundamental hurdles we face in applied economic studies.
Summary
In this session, we first categorize steps of applied econometric analyses: research question, design, modeling, analysis. We also quickly discuss what constitutes a good research question to allow us to focus on the other sus-mentioned steps in the reminder of the course. We then review logistical aspects for this class. Next, we present some of the fundamental hurdles of applied economic research: spurious correlation, reverse causality, confounders, model misspecification, external validity, lack of statistical power. We also examine how to avoid them: by learning our econometrics, deriving the maths or, what might be more fun, implementing some data visualization and running simulations. Finally we build an example of simulation for regression analysis by studying how omitting a variable may affect the conclusion of an analysis.
Session Outline
- Introduction and steps of applied econometric analyses
- Good research questions
- Logistics
- Fundamental hurdles
- Avoiding hurdles
- Simulations: usefulness through the example of omitted variable bias
- Summary
Materials
Exercise and simulation
During the lecture, we did a simple simulation exercise, exploring the drivers of omitted variable bias in a simple regression. You can find the document for this simulation here.
In addition, you have an assignment for next class.
Specific resources for this lecture
- More on good research questions:
- A fun website with examples of spurious correlations: https://tylervigen.com/