Lecture_14_Zimmerman.Rmd

---
title: "Lecture 14 Zimmerman 2014"
author: "Nick Huntington-Klein"
date: "`r Sys.Date()`"
output:   
  revealjs::revealjs_presentation:
    theme: solarized
    transition: slide
    self_contained: true
    smart: true
    fig_caption: true
    reveal_options:
      slideNumber: true
    
---

```{r setup, include=FALSE}
knitr::opts_chunk$set(echo = FALSE, warning=FALSE, message=FALSE)
library(tidyverse)
theme_set(theme_gray(base_size = 15))
```

## Zimmerman 2014

- Zimmerman 2014 uses a cutoff in the admissions process to estimate the returns to education for academically marginal students
- Today we will be discussing that paper

## Zimmerman 2014

First off:

- What does he look for and what does he find?
- Why might we be particularly interested in the returns to education for marginal students?
- How do we know that RDD gives us the return for *those* students?
- What kind of RDD is this?
- How can we characterize his results and any strengths/weaknesses?

## Zimmerman 2014

- Why does he check for *manipulation of the running variable* in Section V.A?
- Why might this be important?
- What does manipulation mean and why might it mess up an RDD result?
- How does he do this check?

## Running Variable Notes

- We can do these sorts of tests ourselves for manipulation using the `rddensity()` and `rdplotdensity()` functions in the **rddensity** package 
- Other potential issues with running variables: *granularity*
- Why might it be difficult to do an RDD if the running variable is very *coarsely defined*?

## Zimmerman 2014

- What other tests does he do? 
- What does Figure 3 show us?
- How can we get the results from the graphs and from the regression tables?
- Is there anything we might want to do differently in this study?