Using the ACS: Income by group Choices and income

I have chosen to analyze income by groups that people have some control over, like their level of education, rather than groups that we didn't get a say in, like gender. This choice is not intended to downplay the very real and very problematic institutional prejudices, like sexism and racism, that reduce income among very large groups of Americans, like women and people of color. That effect is also glaringly obvious in the ACS data. However, this analysis is geared toward answering the question "How can I make more money?".

This analysis was done without using the weights provided in the ACS that account for the different frequencies with which different kinds of people respond to surveys. Naively assuming that our survey sample is representative has two main effects. First, our data will look more like your typical survey responder than your typical American. Second, we will have smaller standard errors and be more confident in our conclusions than is warranted.

Income is not even remotely normally distributed (shock!), so I used non-parametric tests to test for significant differences, specifically the Wilcoxon and Kruskal-Wallis rank sum tests.


Income by education level

Stay in school

Each increase in education level brings a significant increase in income, statistically if not necessarily practically. We're considering professional and doctoral degrees to be the same "level" but different types.

If you only have a high school education, a diploma is worth $3,000/year more than a GED, giving a median income of $18,004/year versus $15,000. Similarly, some college is better than high school alone, giving an increase in median income from $18,000 to $20,000, but it's much better to finish the job, even with an Associate's ($30,000) rather than a Bachelor's ($41,000).


Income by field of degree

Conventional wisdom sadly true

Unsurprisingly to your guidance counsellor, people who earned engineering degrees have a significantly higher median income ($75,000) than people with any other type of degree, even math and science ($60,000). This analysis grouped computer science degrees under math and science. People with industry specific, career-focused degrees, like accounting or hotel management, earned still less ($50,000)

People in the social sciences ($43,000) and humanities ($38,000) earned the least. Sorry, liberal arts. The economy does not value you as highly as bridges and microprocessors.


Income by industry

A study in categorization

I used the ACS industry categories for this one, and honestly it's giving an argument for doing machine learning. These groupings generate categories with huge variance, since, for example, investment bankers are lumped in the same categories as people working customer service at a car rental.

Perhaps the most notable conclusion from this analysis is that public servants, like senators or mayors, make pretty good money. But, if you want a shot at making truly ludicrous amounts of money, be a banker (Banking, real estate and rentals) or a management consultant (Professional services).