Column: Do you live in a bubble? These 100 communities do
Editor’s note: Charles Murray is a conservative thinker and author who first gained notoriety in 1994 as co-author of “The Bell Curve.” Murray, who was widely denounced at the time, argued that intelligence, as measured by IQ, is a major influence on economic success and most controversially, that IQ is significantly influenced by genetic factors — including race. In recent years, Murray’s work has focused on white America and the growing class division within it. In 2012, Making Sen$e devoted a broadcast segment to Murray and his bestseller “Coming Apart” and published his “Do You Live in a Bubble?” quiz online. Economics correspondent Paul Solman interviewed Murray again during the 2016 presidential primary campaign, when Murray denounced Donald Trump and related his success to the alienation of white working America. He updated his quiz at the time, and we republished it, asking those who took it online to provide their ZIP codes.
This year it became obvious to just about everyone that America really does have a new elite that is isolated from and often ignorant about white mainstream America.
It seems like an appropriate moment to present some final results from the NewsHour’s database of scores on the bubble quiz. I posted preliminary results last spring. Now I’ve got many more scores and have had enough time to vet and analyze the data more thoroughly.
A little background for those who are new to the Bubble quiz: In “Coming Apart,” a book I published in 2012, I asked my readers to score themselves on a 25-item test titled “How Thick is Your Bubble?” Scores could range from 0 to 100. The lower the score, the thicker one’s elite cultural bubble. Last spring, Paul Solman interviewed me for the NewsHour’s regular Thursday feature, Making Sen$e, for a segment on why economic anxiety is driving working class voters to support Trump.
The Making Sen$e staff also created a version of the bubble quiz that could be taken online. As I write, more than 142,000 people have posted their scores. Along with those scores, participants were asked to report their ages, current ZIP codes and the ZIP codes where they lived when they were 10 years old.
The bubble quiz had the specific purpose of testing your familiarity with mainstream white America, not with all of mainstream America. Here’s how I explained it in “Coming Apart”:
Some of the questions are ones that whites will get right more often than minorities, and that people who do not live in metropolises will get right more often than people who do. That’s because I am writing about the problems of the new upper class, the new upper class is overwhelmingly white and urban, and the readers of this book are overwhelmingly white and urban.
Today I’ll give you some basic results and a list of the 100 ZIP codes with the lowest mean bubble scores — by my logic, the ZIP codes with the thickest bubbles.
Some Basic Stats
In preparing the data, I deleted cases without valid and geographically-defined ZIP codes and respondents who were very young or suspiciously old. I also cleaned the database of cases in which the same person appeared to have taken the test a few minutes apart. This left me with 136,677 respondents ages 20 to 99 with valid bubble scores and current ZIP codes. For multivariate analyses, I further restricted the sample to the 130,919 who had valid ZIP codes for age 10 as well.
It is not a nationally representative sample — not surprising, considering that the first 30,000 or 40,000 people who took the quiz were mostly people who watch PBS, which has a famously well-educated, sophisticated audience. Even after the first few months, the people who took the quiz online were necessarily people who take online quizzes — not your average American. And so it came to pass that the typical quiz taker came from a ZIP code that was 28 percent richer than that of the average American, with 55 percent more people with college educations and 34 percent fewer people without high school diplomas. The ZIP codes of the quiz takers were whiter and more Asian than those of the average American and less black and Latino. They were more urban, with the typical quiz taker living in a city with more than twice the median population of the city where the average American lives. The mean bubble score was 41.4 with a standard deviation of 16.3. Correcting for the skew in the sample produced an estimated current national mean of 45 and a standard deviation of 16.1.
More or less by accident (I devised the quiz as a teaching device), the bubble quiz is a pretty darn good test. Its scores have the desirable psychometric property of forming an almost perfect normal distribution (a bell curve). More importantly, the scores were, as I hoped, related to the socioeconomic status of the ZIP code where people lived, with socioeconomic status measured by a combination of median family income and the percentage of people age 25 and older who have a college education. The data for assessing ZIP codes were taken from the 2010 to 2014 combined American Community Surveys, conducted by the Census Bureau.
Here are the basic correlations, on a scale of –1 (a perfect negative relationship) to +1 (a perfect positive relationship) of the bubble score with the socioeconomic status index and its component parts:
Correlations for the ZIP code at age 10
Socioeconomic status index –.38
Percent of adults with a Bachelor of Arts degree –.41
Median family income –.36
Correlations for the current ZIP code
Socioeconomic status index –.29
Percent of adults with a Bachelor of Arts degree –.34
Median family income –.25
In the social sciences, these are respectable correlations. But the really interesting finding emerged when I moved from these simple correlations to a multivariate look at the data. When I controlled for the age of the respondent and the urbanization of the ZIP code, it turned out that virtually all the effect on the bubble score is driven by the percentage of adults with a college degree in the ZIP code where the respondent lived. The ZIP code’s median family income had almost no independent effect. Another interesting finding: The ZIP code where people lived at age 10 had a modestly larger effect on their bubble scores than their current ZIP code. It’s not an implausible result, but it’s also not one I would have confidently predicted ahead of time.
A third interesting finding is one that I presented in the earlier posts, but bears showing again with the larger sample sizes: The relationship between the bubble score and the ZIP code’s socioeconomic status is nonlinear. Take a look at this graphic:
For people who live in ZIP codes in the bottom quartile of socioeconomic status, increases in socioeconomic status had little association with the bubble score. In the middle two socioeconomic status quartiles, scores gradually declined. In the top socioeconomic status quartile, increases were associated with accelerating declines in the bubble score, becoming especially steep for those in the top few percentiles.
The sample sizes are large enough to have some confidence that this drop-off at the top of the socioeconomic status range is real: 41,565 of the bubble scores were registered by people who are now living in ZIP codes in the top socioeconomic status decile, and 26,923 of them come from people who lived in such ZIP codes at age 10.
The important point about the graph is that the top few percentiles are crucial for understanding our cultural divide. The people living in ZIP codes in the top two percentiles include almost all of those who run the nation’s culture, economy and politics. And that’s where the bubble scores plunge.
With that as preface, it’s time to look at America’s 100 bubbliest ZIP codes. Of the 2,439 ZIP codes that had 15 or more respondents, here are the 100 ZIP codes with the lowest mean bubble scores in the country. The means are based on the ZIP codes where people live now, not when they were age 10.
The New York City area dominates, accounting for 34 ZIP codes in the top 100. Manhattan alone has 18 ZIP codes in the top 100, including all the ZIP codes in the Upper East Side and Upper West Side. The ZIP codes surrounding San Francisco Bay constituted the next largest contribution, amounting to 29 ZIP codes. I have divided them somewhat arbitrarily between those associated with San Francisco and those of Silicon Valley. The Boston area has 15 ZIP codes, and Los Angeles has 8. That adds up to 86 out of 100. The entire rest of the country had just 14. The lowest mean score for the top 100 was 23 (Fremont, Calif.), and the highest was 29.8 (the Rittenhouse Square area of Philadelphia).
The dog that didn’t bark was Washington, D.C. The ZIP codes in the Washington area where the city’s movers and shakers live are extremely elite, but only one made it into the Top 100 on the bubble quiz. That doesn’t mean Washington isn’t in a bubble. A total of 2,633 people living in the Washington area’s ZIP codes that are in the 99th socioeconomic status percentile submitted bubble scores. The mean was 33.8, more than 11 points below the national average. But the elite Washington ZIP codes aren’t nearly as extreme as the elite ZIP codes in New York, Boston, Los Angeles and the Silicon Valley/San Francisco region.
Along with the top 100, I must add some other ZIP codes that are conspicuously bubbly even though they didn’t meet my minimum requirement of 15 respondents. Fourteen ZIP codes had 10 to 14 respondents and scores lower than 27, meaning that they were virtually a lock to have qualified for the top 100 if they had added just a few more respondents. In order of scores from low to high, they were Westborough, Massachusetts (with an incredibly low mean of 18.7); Rockville Centre, New York; Briarcliff Manor, New York; Old Greenwich, Connecticut; Del Mar, California; Mount Kisco, New York; Malibu, California; Chestnut Hill, Pennsylvania; Stanford, California’s second ZIP code (94304); Lincoln, Massachusetts; Tarzana, California; Rockport, Massachusetts; Greenwich, Connecticut; and Manhasset, New York.
How well did the bubble quiz database cover the people who live in elite ZIP codes? In the entire United States, 414 ZIP codes were in the top two socioeconomic status percentiles, with a total population of 6.28 million people. Of those, 212 had 15 or more respondents to the bubble quiz — just a little more than half. But the ones with that many respondents also tended to be the most populous elite ZIP codes, holding a total of 5.02 million people. In other words, ZIP codes with 15 or more respondents amounted to 80 percent of Americans who live in those elite ZIP codes.
I also just identified the ZIP codes with 10 to 14 respondents who would have qualified for the top 100, which brings us up to 88 percent of the population in elite ZIP codes. Scanning the ZIP codes that make up the remaining 12 percent indicates that almost all the well-known elite towns have low scores, albeit based on fewer than 10 respondents. ZIP codes with means lower than 27 and at least three respondents were Pound Ridge, New York; Saddle River, New Jersey; Glen Rock, New Jersey; Gladwyne, Pennsylvania; Southport, Connecticut; a downtown Chicago ZIP code (60603); and the lower Manhattan ZIP code (10282) just to the west of Tribeca.