In the early 1990s, you could tell a rich person by whether they bought Grey Poupon. Now, it’s an iPhone.

The US is so polarized that people hardly live in the same country any more. Beyond differing political views, they don’t watch the same TV shows, buy the same brands, or read the same media.

Right?

Wrong, and there is research to prove it. University of Chicago economists Marianne Bertrand and Emir Kamenica have produced a deeply researched study on whether people in the US have diverged culturally over the past half century. They haven’t, really.

Using survey data on the brands Americans buy, the beliefs they hold, the movies and TV they watch, and the ways they spend their time, the researchers tested how well they can predict someone’s income, race, and political views (pdf). Specifically, they were interested in whether it is easier to use this data to make predictions today than it was in decades past. If it is easier, they say, it means that Americans of different socioeconomic groups have grown further apart when it comes to how they experience the world.

Bertrand and Kamenica find that Americans are quite different, but that gulf is not widening. Knowing the TV shows someone watches or the shampoo they prefer tells you no more about their income or race than it used to. One exception is that cultural beliefs and attitudes—like whether premarital sex is wrong or whether the US is spending enough on defense—now better predict someone’s politics than in the past.

It’s significant and topical research, to be sure, but it’s also one of the more fun economics papers you will read. Beyond the usual methodological details, the authors have listed the products, opinions, and media that best predict someone’s background. It is funny, surprising, and sometimes sad. Here are some of our favorite nuggets from the research…

💰 Watching these TV shows was the most predictive of a high income in 2016*

*High income is defined as being in the top 25% of Americans. For this analysis, the researchers only tried to predict if someone was in the top 25% or bottom 25%—those in the middle 50% were left out.

** In the this table and the ones following, percent accuracy represents the accuracy with which you could predict someone belonging to a group only knowing this one piece of information and nothing else.

💰 Using these brands was the most predictive of a high income in 1992 and 2016

🙋‍♂️ Seeing or not seeing these movies was the most predictive of being male in 1998, 2007, and 2016

🙋‍♂️ These cultural traits, actions, or beliefs were the most predictive of being male in 2016

👨🏻 These TV shows and brands were the most predictive of being white in 2016

⬅ These magazines or cultural actions were the most predictive of being liberal in 2009