According to Kotkin, the common belief is that college graduates have been flocking to the coasts. However, according to data provided by Wendell Cox, the largest gains in college-educated citizens as a percentage of the adult population has occurred in the South and Midwest:
"It turns out that none of the top 10 gainers were large Northeastern cities, but largely Southern or Midwestern. New Orleans; Raleigh, N.C.; Austin, Texas; Nashville; Birmingham, Ala.; Kansas City, Mo.-Kan.; and Columbus, Ohio, all scored high marks. Only one California city, San Diego, made the top 10. Perennial "brain gainers" Denver, Colo., and Seattle round out the top 10."
Comments
Is this the right measure?
Is there anybody looking at this with a more comprehensive analysis, rather than simply grinding an ideological ax?
There are some obvious questions about their measure. The increase in numerical number of bachelor's degree as a percentage of the population may not be a meaningful measure of "brain magnets" for a number of reasons.
First, there is an inherent bias against metropolitan areas that already have well-educated populations. Droves of college-educated workers flocking into traditionally well-educated metro areas, especially large well-educated metro areas, could easily be masked by the large numbers of existing college-educated residents, as well as some gradual attrition among older intellectuals.
Second, a bachelor's degree may not be the most indicative measure for a "brain magnet." Would a different picture emerge by using advanced degrees instead?
Third, the scale may not be meaningful. Separate trends could easily be occurring in different parts of the same metro area. There could be huge numbers of college grads moving into the urban cores, without any increases or even modest declines in the sprawling suburban areas, yielding a low percentage for the whole metro area. This possibility, with entire neighborhoods of energetic college grads, would entirely contradict the claims of the article. A meaningful analysis would need to address this possibility.
It is worth noting that Kotkin's article conflates total college grads with recent college grads. The discussion clearly suggests conclusions about the choices of recent graduates, but for all we know from their analysis, the numbers might be driven by trends among educated retirees. Kotkin also makes claims about migration that cannot be supported with the method he describes (increases could be due to maturing youth populations graduating from local institutions instead of choosing to move).
Educational density
In looking at simple data there is often a bias toward percentage calculations when density might be as or more important. Thus I agree that density of college degrees and density of advanced graduate degrees per square mile of city area may be a much better indicator than percentage of college grads, especially if considering the synergies of bumping into people with interesting ideas. (Of course not all interesting ideas come from people with college degrees.) I live in Somerville MA which went from 16% college grads in 1980 to 41% in 2000. Big change in ad hoc grass roots capacity. With nearly 20,000 people per square mile, there is no conventional suburb in MA, no matter how income wealthy, with close to the same educational density. Plus we have artists, real coffee and real beer. And lots of hard working, interesting immigrants - over 50 languages in the public school population.
Cheers, Wig Zamore
Educational attainment density may not be very useful either
I have even more skepticism about educational attainment density:
http://www.theatlantic.com/business/archive/2010/05/the-density-of-smart...
It seems to reflect a built-in assumption that density begets intelligence. There are pitfalls with this measure.
First, it seems too sensitive to general population density. It would be critical to select a sufficiently high level of educational attainment for this measure to have any meaning. For example, selecting completion of an eighth grade education (a high level by historical standards) would reduce American cities to a mere measure of their population density. It is unclear whether an undergraduate degree is sufficiently high.
Second, it probably can't support its premise that it indicates a greater likelihood of interactions among smart people. That claim is probably little shaky to start with, since it assumes that a confined geography is inherently more significant than social networks. More importantly, it can yield patently absurd results. Consider two cities each covering one square mile:
1) Has 1,500 people with a college education among a population of 75,000 people.
2) Has 1,000 people with a college education among a population of 7,500 people.
The measure of educational attainment density would suggest that City One had a greater "cluster of smart people" at 1,500 per square mile, as opposed to 1,000 in City Two. Half-again as smart, right?
Of course, the overall educational attainment in City Two is obviously far higher. The people in City Two are far more likely to interact with other highly educated people than people in City One, either as part of their social network or as a chance encounter.