Session Ready
Exercise

Practice Computing Regression Discontinuity Effects

Let's continue with the NBA research example. On your workspace is a simulated data set, called NBA, that closely resembles the data analyzed by Berger and Pope. The data frame NBA contains game characteristics for over 18,000 NBA games between 1994 and 2009. Will you get the same results as they did? Let's find out!

The outcome variable we are interested in is called home.team.final.margin, which is the final margin of victory (or loss) for the home team. The running variable in this RDD is called home.team.halftime.margin, which records the scoring difference between the home and visiting teams at halftime. Let's keep things simple and take a look at when the home team is winning at halftime, and define our treatment variable as the home team being ahead at halftime, i.e. when home.team.halftime.margin > 0.

In this exercise, you will compute the treatment effect of being ahead at halftime on the final margin of victory. You will use regression methods as well as nonparametric methods to assess how robust the effect is.

Instructions
100 XP
  • 1) Use OLS regression to estimate the treatment effect of being behind at halftime on the final margin of victory under two different parametric scenarios
  • 2) Estimate the treatment effect using non-parametric methods