Diversity for Thee—But Not for Me
S E E D E D C O N T E N T
White progressives in America, as in Britain, avoid diverse neighborhoods and are more likely to leave diverse places than white conservatives. In effect, they don’t practice what they preach. These are the findings from large-scale quantitative research in the United States and Britain, recently published in my academic article, “ White flight from immigration?: Attitudes to diversity and white residential choice .”
Diversity is a core value for white progressives in America and other Western countries. Over 60 percent of them support increasing immigration. As the Manhattan Institute’s Zach Goldberg shows, they are the only major part of the population to feel warmer toward other racial groups than toward their own: they rate whites as more lazy and violent, and less intelligent, than blacks. Among white progressives, 87 percent say that having an “increasing number of people of many different races, ethnic groups and nationalities” makes the U.S. a better place to live, while virtually none says that it makes the country worse.
You would think that aversion to one’s racial group would prompt white progressives to flee disproportionately white areas for more diverse ones. Surveys do show that white progressives are more likely than white conservatives to indicate that they want to live in diverse places. Studies that present Americans with showcards of stylized houses as a proxy for race —some colored white, some black, with the proportion of the latter varied—find that those with progressive racial attitudes say that they prefer more diverse places than conservatives. These findings have been replicated in Britain and the Netherlands .
But when the rubber hits the road, white liberal attitudes don’t translate into behavior.
Getting a handle on actual migration is tricky. It requires a large survey because most people don’t move during a given year. Ideally, we need to track people over time, which demands repeated surveys of the same individuals. Finally, we must track both demographic factors, such as ethnicity and income, and political ones, such as voting or attitudes to immigration and diversity. Longitudinal surveys of this kind exist in Britain, but not in America. My strategy was therefore first to test my arguments with British data and then turn—for the U.S. study—to geotagged Twitter data from pro- and anti-Donald Trump individuals, as well as surveys that ask people for their past and present ZIP codes.
Results from Britain’s census confirm what we already know: minority people move toward areas of greater diversity than whites. My study is based on a massive representative census sample of 123,000 intercensal movers, and it controls for a wide range of material factors, including education, income, age, marital status, distance of move, and living in a mixed-ethnicity household. It also accounts for the affluence and population density of origin and destination neighborhoods. For data security reasons, I had to divide wards (averaging about 6,000 people) into five groups, each containing a fifth of the country’s minority population.
Figure 1, based on a 1 percent linked sample of the census of England and Wales, shows that, among people moving from the most diverse quintile of wards (averaging 66 percent minority), both whites and minorities move away from diversity—by necessity, because their only choice is similar or lower levels of diversity. However, the average white British person moves to an area fully three quintiles less diverse (averaging 13 percent minorities), while the typical minority moves to a neighborhood one to two quintiles less diverse (averaging between 27 percent and 43 percent minority). This ethnic difference holds regardless of where people start from. Thus, in the most homogeneous wards (averaging 2 percent minorities), everyone moves to more diverse places, but minorities move to more diverse areas than white Britons.
T he U.S. is somewhat trickier to study because it lacks longitudinal large-scale survey data with the right questions. I therefore collected a sample of 142 million Trump-related tweets between March and November 2016. Only a 1 percent subset had geotagging turned on, and of these only a small proportion moved during this period (residence is determined by matching morning and evening location). Sentiment analysis helped me identify strong pro- and anti-Trump accounts, which resulted in a sample of just over 6,000. These were then hand-coded to assign approximate age, gender, and race.
As in Britain, I found that whites tended to move to whiter neighborhoods than minorities. But did white Trump and anti-Trump Americans behave like British Brexit and Remain voters? Yes. The results of this analysis appear in Figure 4 and reinforce the British findings—namely, whites who strongly opposed Trump moved to places that were just as white as whites who strongly supported him. While the downward-sloping line shows that whites moving from diverse places chose less diverse destinations—as we would expect, given that those in diverse areas can only move to less diverse ones—there is no significant difference between the colored lines representing degrees of pro- and anti-Trump sentiment. The overlapping lines resemble the story in Figure 3.
Tags
Who is online
406 visitors
One of those things we all know from experience and the data follows.