I visualize earnings, employment, and demographic info for all college majors in the US, using the 2015, 2016, and 2017 American Community Survey 1-year Public Micro Use Samples (PUMS), obtained via IPUMS. This project was inspired by a 2014 FiveThirtyEight article by Ben Casselman, who used the 2010-12 3-year PUMS, covering a period when the US economy was still meekly recovering from the Great Recession. This data can be accessed via the fivethirtyeight package. The current project uses more recent data spanning the succeeding period of stronger economic expansion; also, it has pretty pictures.

Note: This is purely a descriptive exercise. No one should infer causality from these results. Rather, use them to get a general sense of student outcomes after graduation. The Handbook of the Economics of Education has an excellent, exhaustive chapter summarizing the research on the causal effect of college major on labor market outcomes ( earlier draft freely available via NBER).

Also, be aware that many majors don't have enough observations for one to be very confident in their estimates of earnings, unemployment, etc. Again, you should use these visuals to get a general sense of how majors compare (Library Science probably still earns less than Civil Engineering, on average, despite the uncertainty in the estimates) but don't get too caught up on the exact levels of each variable for the less popular majors. E.g., Petroleum Engineering has gigantic median earnings but very few people major in it.

Following Casselman, I consider someone employed "full-time, year round" if they reported working at least 50 weeks in the last 12 months, and at least 35 "usual" hours of work. Someone is considered a "recent graduate" if they have a Bachelor's degree and aged 27 or below.

I exclude individuals who reported working full-time, but reported total earnings from work which are lower than what they'd earn with the Federal minimum wage.

There are 173 majors (FO1P) in total in the ACS, but I dropped majors which had too few people (fewer than 30). Refer to ./src/R/process_ipums.R to see how I implemented these exclusions and decisions.

The processed datasets I used to make the following visualizations are available in ./data/processed/. Explore them and tell me what you find. My inbox is open to anyone with questions and/or feedback.

• Consider expanding the definition of "full-time, year round" workers to include those who report fewer than 50 weeks of work, but who work 35+ hours per week and have been employed at least 6 months.

• Consider excluding, from the visualizations, majors which have fewer than 100 respondents, instead of 30. HUD sets this cutoff for their SAFMR data.

• Look at home ownership rates.

Licensed under the General Public License version 3.0 (GPLv3.0). Some data I used cannot be provided.

IPUMS took the ACS data, cleaned it, recoded it, extensively documented it, and much more. Here is their full citation:

Steven Ruggles, Sarah Flood, Ronald Goeken, Josiah Grover, Erin Meyer, Jose Pacas, and Matthew Sobek. IPUMS USA: Version 8.0 [dataset]. Minneapolis, MN: IPUMS, 2018. https://doi.org/10.18128/D010.V8.0

This wouldn't have been possible for me without Ben Casselman's code. Reading and modifying it taught me R for the first time. Please check out his work.

I work in a crowded space. Here are other people and orgs doing similar work on college majors.

• The Labor Market for Recent College Graduates (2018) by the New York Fed.

• Lifetime Earnings By College Major (2018) by Dr. Doug Webber of Temple University. Here are the accompanying papers.

• Putting Your Major to Work: Career Paths after College (2017) by The Hamilton Project.

• The Economic Value of College Majors (2015) by the Center on Education and the Workforce at Georgetown University.

• The Analysis of Field Choice in College and Graduate School: Determinants and Wage Effects (2015) by Altonji et al. NBER. Access working paper version for free here.

• The Economic Guide to Picking a College Major (2014) by Ben Casselman. FiveThirtyEight.