In honor of the new summer, which has made the east coast something of a sauna lately, and prompted many here in San Francisco to quote Mark Twain, perhaps incorrectly, the optim.al data team decided to take a look at what happens when throngs of newly minted college graduates head out into the big bad world, or, as the case may be, move back in with their parents or follow a girlfriend across the country. Our aim here is to learn what cities on Facebook have the highest proportion of college graduates, and in doing so to shed light on what makes these cities unique and what they can teach us about the way people in America cluster. We’ll give recent graduates an idea of where it will be easiest to extend their collegiate experience, at least until the prospect of being an adult becomes real enough and scary enough to send them back to graduate school.

So, what cities on Facebook have the highest proportion of college graduates?

We included in this study only cities that have more than 300,000 Facebook users—62 cities, to be precise, from behemoths like New York and Los Angeles, to smaller places like Birmingham, Alabama, and Colorado Springs. The most educated large city on Facebook is Washington, DC, followed closely by San Francisco, Portland, and Boston.  None of this should be surprising: these are cities with academic reputations whose job markets and lifestyles tend to attract young graduates from across the country.

It would be easy to think that college graduates tend to cluster in the country’s biggest cities.  But, proportionally at least, that does not appear to be the case.  The size of the Facebook audience in each city has almost nothing to do with how educated that city is.

Are cities where college graduates congregate safer than other cities?  Unfortunately for proud and protective mothers everywhere, there is no relationship between a city’s crime rate (as compiled by the FBI) and its proportion of college graduates.

So what does define the cities where college graduates make up an abnormally large slice of the population?  And what does it mean about America?  Luckily for us, Facebook is more than just a frequently updated repository of census-style data.  It is the single largest database of personality ever assembled.  So let’s dig deeper.

Interests by City

For each city in our study, we determined popularity indexes for the top 1,000 interests on Facebook.  This gave us about 60 interest lists, each of which, when read from strongest to weakest, forms something like a round of Taboo.  See how many words it takes for you to guess the city:

Mad Men, The National, In N Out, Gay Marriage, The Cure, Wine, Community Service…

Entertainment, The Beach, Relaxing, Roller Coasters, Spring Break, Shakira

Hillary Clinton, International Relations, Martin Lawrence, Politics, Network, Kat Williams

Let’s call each of these lists a City Psyche.  They give you a pretty good idea of what the residents of each city care about in aggregate, of the entertainment they consume, the food they eat and the activities they enjoy.  It is nothing less than a measure of culture, of what it is like to live in a specific city.  And because each keyword has a strength value attached to it, we can do some fun math to them!

Here is a chart showing similarity indexes between selected City Psyches.  The value in each box is the cosine distance between the 1,000-length keyword vectors associated with each city.

So from this sample, we can say that the City Psyches of Baltimore, Birmingham, and Atlanta are all relatively similar; and that Albuquerque’s and Austin’s are quite similar also.  Boston’s is the least like the others in this small sample.

Just as we built a City Psyche for each city in our study, we crafted a National Psyche representing the top interests of Facebook users across United States.  We then evaluated the similarity between each City Psyche and the National Psyche.  This leads us to a measurement of how culturally different each city is from the national mainstream.

We found an incredibly strong relationship (r^2 = 0.69) between the level of college graduates in a city and the distance of that City Psyche from the National Psyche.   The level of college graduates in a city explains far more of that city’s ‘uniqueness’ than any other demographic factor we looked at, including race, average age, even income.  Higher education creates difference, not sameness; it draws people away from a national average of tastes and values.  Whether that is a good thing or bad thing we leave to the ideologues.

To paraphrase: the volume of the message that a city broadcasts about itself, the extent to which a city has been able to individuate from a national consensus of tastes and values, is most strongly explained not by race and not by income, but by how educated its citizens are.