Why Do Americans Wildly Overestimate the Size of Minority Groups?

A recent YouGov poll asked Americans to estimate the size of various groups. One finding stands out above the others. Taylor Orth of YouGov writes, “When people’s average perceptions of group sizes are compared to actual population estimates, an intriguing pattern emerges: [Americans] tend to vastly overestimate the size of minority groups.” Among many other groups, this was true of millionaires, transgenders, Jews, atheists, vegans, Asian Americans, and African Americans.

The survey respondents estimated that 41 percent of Americans are black, when in reality, only 12 percent are. In one sense, this question should be especially easy. If you had looked up the percentage of Americans who are black at any point in the past 50 years and just said that answer, you’d pretty much be correct. African Americans as a share of the overall population have held very steady over time, staying within the narrow band between 11 and 13 percent since the mid 1970s.

So why were the survey respondents so wrong? One reason may be social-desirability bias. It could be perceived as insensitive to a minority group to underestimate its size, so respondents would have incentive to say larger numbers. That’s especially true of estimating the black population, given the uniquely brutal history of slavery and racialized oppression against blacks in the U.S., and the past few years of protests and riots around racial issues.

What’s more, though, is that guessing the correct proportion requires lots of abstract thinking. Very few people live in a place where African Americans make up 12 percent of the population. While that is the national average, the distribution is very uneven geographically.

That phenomenon is not new in American history. President Lincoln, in his Second Inaugural Address, spoke of African Americans being “not distributed generally over the Union, but localized in the Southern part of it.” Even long after abolition, that remains somewhat true. According to Pew Research, 56 percent of African Americans live in the South, defined as states south of the Mason-Dixon Line or the Ohio River, plus Texas and Oklahoma.

But the Great Migration of the 20th century means there are many African Americans in the North as well, and they tend to be highly concentrated in cities (although that’s beginning to change). Here’s a graph of the 50 most populous cities in the U.S., sorted by the percentage of the population who identified as black or African American alone on the 2020 Census (the vast majority of respondents identify with only one race). You can float your cursor over the columns to see the exact percentages.

A few things stand out. First, many northern cities have black populations indistinguishable in proportion from southern cities. Chicago is roughly the same as Jacksonville, and Boston is roughly the same as Fort Worth. Second, some northern cities have black populations that are far larger proportionally than some southern cities. Milwaukee is way ahead of Nashville, and Philadelphia is way ahead of Virginia Beach.

Another interesting observation relates to the one yellow column. That’s the African-American proportion of the total U.S. population, 12.4 percent. Of our 50 largest cities, 32 of them have an above-average proportion of black residents. And the variability among those 32 is much higher than the variability of the 18 below the average. Even excluding Detroit, Memphis, and Baltimore, the range of the above-average cities is 26.7 percentage points higher than the range of the below-average cities.

Given these data, you can see how an attempt at statistical reasoning in estimating the overall African-American population could go awry. People likely started to answer the poll question by considering where they live. Let’s say someone living in Atlanta thinks the city’s proportion of black residents is about 50 percent (which is true). Then, he thinks most African Americans live in the South (which is also true). Therefore, Atlanta is probably a little above average, so let’s guess 40 percent (totally wrong).

Or, let’s say someone living in Chicago thinks the proportion of black residents in Chicago is roughly 30 percent (which is true). Then, he thinks most African Americans live in the South (which is also true). Therefore, Chicago is probably a little below average, so let’s guess 40 percent (totally wrong).

There is at least some logic behind this approach to answering the question, and it’s based on relatively accurate assumptions, but it completely fails to get the right answer. It fails to consider the West, where even very large cities that most Americans think of as “diverse,” such as Los Angeles, San Diego, and Phoenix, have very low black populations, with proportions in the single digits. It also fails to consider the vast swaths of the Midwest, Mountain West, and West where the proportion of residents who are black is around 1 or 2 percent. Those areas are sparsely populated, but there are many of them, and they add up.

Before you get on your high horse about how dumb the survey respondents were, consider this: Of the 50 largest cities, the one with the most nationally representative proportion of black residents — a perfect match at 12.4 percent exactly — is Omaha. Most people probably think of Nebraska as having an overwhelmingly white population, and it does: 78.4 percent of Nebraskans identified as white alone, and only 4.9 percent identified as black alone. But 62.4 percent of Nebraska’s black population lives within Omaha city limits.

Other large cities with nationally representative proportions of black residents also may seem unexpected: Las Vegas, Long Beach, and Wichita. Oklahoma City’s proportion of black residents is much closer to the national average than New York City’s is.

The better statistical reasoning would go like this: African Americans are highly likely to live in cities. Many cities have black populations around 20 or 30 percent. Therefore, let’s guess 10 or 15 percent for the national average. Alternatively, if you live in Omaha and know that about 12 percent of your city’s residents are black, that’s a solid guess for the national average because Nebraska has such a high proportion of white residents to begin with.

In the Internet age when supercomputers in our pockets can produce these statistics instantaneously, Americans should not be so wildly wrong when estimating basic demographics about their fellow countrymen. The results of the YouGov poll demonstrate, above all else, a failure of curiosity among people who are saturated in more information than ever before in human history. But a combination of social-desirability bias and well-intentioned but wrong statistical reasoning could have produced the large overestimate of the share of America’s population that is black.