23andMe Looks At Regional Traits

23andMe mapped geographic patterns for more than 1,500 traits using aggregated and anonymous customer data to produce American tapestries of everything from personality to food preferences.

The data, compiled by researcher Emma Pierson, show that there are statistically significant regional differences that go beyond red states and blue states or whether you call a soft drink a “soda” or a “pop.” Eating habits show statistically significant discrepancies by region, for example:

All visualizations in this post can be viewed here. Note the maps appear all black, but if you roll over the image, the states and percentages will appear.

All visualizations in this post can be viewed here. Note the maps appear all black, but if you roll over the image, the states and percentages will appear.

And, somewhat unsurprisingly, the states where people eat the most fast food are also the states where they exercise the least. Dog owners and cat owners also favor different states. That’s probably a good thing.

Fast Food:Pets

Illnesses also show regional differences. People catch more colds in the northern states, particularly the northeast, while people appear to get frizzy hair in humidity more often in the East.  Rashes from poison oak, ivy, and sumac are most common where the plants are common. Kidney infections are more common in the South. People who report that they suffered from altitude sickness tend to favor the flatter Great Plains states, and coastal states like California and Hawaii.

Emma also found regional differences in the prevalence of certain genetic variants. This is likely tied to the concentration of people with similar ancestry in certain states. For example, genetic variants in the GBA gene that cause Parkinson’s and Gaucher disease are most often found in people of Ashkenazi Jewish ancestry. Concentrations of people with these variants are found in the Northeast where some of the highest concentrations of people with Ashkenazi ancestry live. Other genetic traits common in people with Ashkenazi ancestry, like lactose intolerance, also peak in the Northeast. G6PD mutations, which can cause anemia and are more common in people of African ancestry, peak in states with large African populations.

While these maps highlight regional differences, they also speak to what unifies us. In some traits, especially personality traits, it’s very hard to pick out a regional pattern.

Emma, who also has a blog about statistics called Obsession with Regression, then looked at how traits correlated with how people moved from place to place.

For instance, she found that — as essayist Joan Didion said 50 years ago about New York City being “for the very young” — young people tend to flock to Gotham. Inspired by this wind map, she produced maps that track rivers of movement of people. These lines follow the path of older Americans migrating from New York to Florida, and in general out of the northeast and to the south and to California. The line means that customers who move from New York to Florida are older than those who move from Florida to New York.  The redder the line, the bigger the difference and the thicker the line, the more people moving.

The maps also track Latinos moving out of California and Florida and into other parts of the country, as well as showing people with Ashkenazi Jewish ancestry moving out from the Northeast.

The maps also track the many different types of people moving either to the Northeast or California, regions that attract college graduates, people with higher incomes, people attracted to those of the same sex, and the physically attractive. Emma was even able to see political conservatives migrating out of the Northeast to places like Texas, Arizona and Florida.

While these are fun bits of seemingly random bits of data, they also have the potential to shed some light into other parts of 23andMe’s research. We ask our customer any number of things from the odd — like whether they can do a cartwheel or curl their tongue — to more serious questions about reactions to medication and whether they have had cancer. Taken together this data will help our researchers gain insight into both the genetic and non-genetic influences on various diseases and conditions.






  • croftraider

    Very cool post. But, the link to “All visualizations of this post…” is currently broken.

    • Scott23H

      Thanks for giving us a heads up. It will be corrected shortly.

  • Jean_Brocklebank

    I enjoy sharing information on my 23andme account. I am fascinated by data from the surveys. I haven’t finished all surveys yet but hope to do soon.

  • Alicia

    Love the data visualization. However, it might be worth comparing with nation-wide BRFSS data on some of the health measures. BRFSS is a random-sample and 23andMe data is not. Not sure the statewide sample size of 23andMe vs. BRFSS, but would have to imagine that 23andMe is predominantly accessing views of middle- to high-income Americans, likely mostly urban, etc., so the sample is not representative of “America.” Would love to know more about the 23andMe sample. Despite the limitations, I definitely apprecaite the maps. Data visualization is so in right now.

Return to top