Sunday, May 8, 2016

The effect size from their perspective is that much smaller than from yours

Somewhere among these many posts, I have speculated about the connection between where news media outlets are located and the representation of America in the media. This is more than an issue of "fly-over" country. Specifically, my speculation was that the fact that most national news media organs are located in a handful of major cities (New York, Boston, Philadelphia, Chicago, Los Angeles, San Francisco, and the ilk) predisposes reporters/journalists to assume that the rest of America looks like the urban America with which they are familiar.

Cities have historically been unrepresentative of the population of the countries in which they are located. They tend to be more socially liberal and more accommodating of personal privacy. Among the top 25 cities in the US, many have significantly disproportionate LGBT populations, criminal activities, Jews, Muslims (and other religious minorities), and ethnic minorities.

I speculate that the experience of reporters in their residential demographics filters into their assumptions of their reporting and that this accounts for the dramatic misestimation on the part of the public of minority populations. For example, 17% of the population thinks that African-Americans are in the majority and the average estimate is that African-Americans are 33% of the population as opposed to their actual representation of 14% (see Gallup). Similarly but to an even greater degree, the public dramatically overestimates the share of the LGBT population in the nation. Whereas LGBT are between 1-5% of the population (depending on definitions and surveying mechanism), the public estimates that 25% of the population is LGBT.

These overestimations are of such a magnitude as to boggle the mind. Statistically it is impossible for people to be sampling their own social networks and be deriving these elevated estimations. Instead, they have to be generating these estimates from other impressions. Hence my guess as to the outsize influence of unrepresentative cities as the basis for misimpressions transmitted via the media. Reporters (and talking heads) live in cities that are disproportionately foreign born, African-American, LGBT, etc. When they communicate in articles and reports, that is the frame of assumptions which they are communicating and which everyone else picks up on. Hence the overestimations.

Even for cities these are dramatic overestimates but not quite as wide of the mark as for the country as a whole.

All this is brought to mind by a Charles Murray post following up on his Bubble Quizz (the degree to which people can and do create unrepresentative cognitive environments which lead them astray). His post is Where White America Lives. Murray often focuses on class, culture and behavior and therefore focuses only on the white population in order to avoid complicating the issue with other races/ethnic groups. In other words, he can say things about whites that cannot be said about others.
The Bubble Quiz has gotten a lot of attention since I started posting results from the people who took it on the NewsHour’s website. Some have recently complained that the Bubble Quiz ignores the real America, which is urban and racially diverse. The website FiveThirtyEight, bringing its formidable quantitative skills to bear, has determined that the most “normal” American setting is a place like Tampa.

None of this is relevant to the Bubble Quiz, which is explicitly intended to illustrate how isolated the new upper class (overwhelmingly white) is from mainstream white America, but the complaints made me curious: Where does the typical non-Latino white American live?

If you’re asking about the mean population of the places where white Americans live, the answer is a city of 647,700 people — a big city. The mean is meaningless, however. If you have a sample of 10 Americans, one from New York City, population 8.5 million, and 9 from villages of 100 people, the mean population of the ten is 850,090 people. That doesn’t tell you much.

If you’re asking about the median population of the places where white Americans live, the answer is 45,200. That’s not Tampa. That’s the size of Wallingford, Connecticut, a town between New Haven and Hartford and home to Choate, the famous prep school.

Let’s be clear about what that median represents, because it’s a pretty astonishing result: Fully half of white Americans live in places smaller than 45,200 people.

It’s so astonishingly low that you should be suspicious. Specifically, a place can be called a “city” with its own mayor and city council, but for practical purposes it is part of a dense urban area. Cambridge, Massachusetts, for example, is technically a city of 107,000 people, but it is contiguous to Boston, and you definitely feel like you’re in an urban area. Belmont, only three miles to the west, is listed at 24,700. It is a residential community with a small-town shopping center and doesn’t have the feel of a city, but it is a suburb of Boston.
Murray then goes through the exercise of segmenting town/cities/suburbs and comes up with:
That leaves the twenty largest urban areas, all of which are larger than 1.2 million, including all the cities that come first to mind when we think of urban America: Greater Los Angeles, New York, Houston, Chicago, Miami, Seattle, Las Vegas, Phoenix, Denver, Philadelphia, San Diego, San Francisco, Dallas, and Atlanta.

Sprinkled among the top twenty are other places that may surprise you: Minneapolis, Fort Lauderdale, San Antonio, Sacramento, Orlando, and San Jose.

Here’s how the breakdown for where white Americans live works out:
Rural areas–Newton IA (0–15,000) - 26.4%

Newton–Wallingford CT (15,000–45,000) - 18.9%

Wallingford–Des Moines (45,000–370,000) - 20.9%

Des Moines–Indianapolis (370,000–1.2 million) - 11.8%

Suburbs of cities - 11.6%

The top twenty cities 10.5%
More than a quarter of all white Americans live in small towns (or no town at all) and almost two-thirds live in cities smaller than Des Moines. Only 10.5% of white Americans live within the top twenty cities. So the lesson for today is that white America is still, by a substantial majority, an America of rural areas, small towns and small cities.

Since 17% of the entire American population lives in the 20 largest cities compared with less than 11% of white Americans, isn’t the implication that the residential profile for nonwhite America is radically different from the one for white America? It is indeed.
I suspect this sheds some light on several things. There is clearly an elite bubble issue based on class and income and exacerbated by the fact that elites tend to live in the big cities.

I suspect that this information on where people live also shapes assumptions in ways that are not explicit. America is whiter, straighter, more community oriented and more native born than you would understand if you take your reference only from big cities. That is difficult for elites to process along with their own cognitive bubbles.

Similarly, advocacy groups tend also to be highly concentrated in big urban environments. It would make sense that their sense of injustice might be inflated by the anchoring effect of their own urban environs compared to the country at large.

Say you are an LGBT advocate and based on your neighborhood in your city, you have the perspective that 25% of the population are LGBT. Whatever the issue you might be focused on, your estimate of its importance is going to be an order of magnitude greater than everyone else in the country because your population estimate is off by an order of magnitude (2.5% versus 25%). So the passion you bring to your advocacy seems out of proportion to everyone else, not because they disagree or because they are intolerant or because they are bigoted - simply because the effect size from their perspective is that much smaller than from yours.

Very interesting.

No comments:

Post a Comment