In Lab 2, we documented differences in income and poverty rates among White and Black US residents. There are various reasons for what causes these gaps (racism, discrimination, and intergenerational mobility, just to name a few). We will investigate whether education can explain these gaps.

We’ll be working with data from the American Community Survey (ACS) which you can download on the course Wordpress site.

The unit of observation in the ACS is an individual.

Distribution of Education and Income Gaps

We will first summarize the gaps in college and income between White an Black individuals. The goal is to present a visualization showing these inequities. Let’s go through the following steps.

1. Generate Dataframes

Starting from the ACS dataset (individual observations), let’s generate two dataframes. The first will consist of state averages for college completion rates and income for Black individuals over the age of 25. The second will consist of the same for White individuals.

So we are going from a dataset where the unit of observation is the individual level (ACS) to two separate datasets where the unit of observation is the state level.

Hints:

  • Use filter, group_by, and summarise

  • Your state-level datasets should have 51 observations (DC + 50 states)

  • The variable educ_college_4yr is known as an indicator variable and takes a value of “1” if the individual has completed college

2. Join the 2 Dataframes

We haven’t covered this yet in labs, but we’ll do it here. We will join or merge these two dataframes that you created in part 1. How do you do this? See this 5-min video

3. Create two new columns/variables that measure the gap

Using your newly joined/merged dataset, create two variables/columns that take the gap or difference in college completion and avg income for each state.

Hint: -Use mutate

4. Visualize the Distributions

Present the distribution (histogram) for the college gap and the income gap.
If college was accessible by everyone, how would you expect the college gap distribution to look like? How does it compare to what we see in the data?

Based on the distributions for the college and income gaps, are there any patterns between the distributions?

5. Rank all states by the Black/White gap in college in decending order. Juxtapose this with states ranked by Black/White income gaps. Is there evidence that reducing racial inequalities in education leads to reductions in income inequality? What else might be leading to the patterns that we see in this analysis?
6. Closing the Gap

If we want to close the Black/White gap in college completion, we might want to improve access via the supply-side. For example, reducing tuition and making college more affordable is an obvious policy to pursue, however, there might be more at play. We discussed some possible demand-side constraints, such as students knowing the true distribution of wages/earnings for those with a college degree.

What if there’s another constraint? Peer pressure can certainly affect the decisions that people make. Leonardo Bursztyn, a behavior economist, has explored these issues with his co-authors. Bursztyn did an experiment in Los Angeles high schools and wanted to see who would sign up for a $200 SAT prep course when it was offered for free. Surprisingly not everyone signed up for it. Why?

Read more about it in the following articles.

-Interview - easy read

-Academic Paper

The academic paper is a little more difficult. I tend to focus on the abstract and on the first 4-5 page introduction, and then skim the rest.

Based on these readings, what sort of policy intervention would you propose if taking SAT prep classes and applying to college had some social stigma tied to it?