A majority of white Americans have few, if any, close confidants who are non-white, according to a recent study. But some media outlets went further than that, claiming that the study found most white Americans don't have any non-white friends at all. That conclusion is not warranted -- although we should still be concerned about what the study does show.
According to the survey, conducted by the Public Religion Research Institute in 2013, 91 percent of people in the close social networks of white Americans, or the people they most often talk to about important matters, are also white. Similarly, 83 percent of those in the close social networks of black Americans are black.
Dan Cox, research director at PRRI, said the purpose of the study was to identify the demographics of "individuals who are important to the person and have an influence over them." The characteristics of those people matter. For example, surveys have shown that people who have a close gay friend or family member are far more likely to support same-sex marriage than people who don't.
Having little interaction with non-white friends might have a similar impact. White people who have few or no non-white confidants may be missing out on certain perspectives on important issues -- for example, when tensions flare up over racially charged situations like the recent shooting of black teenager Michael Brown by a white police officer in Ferguson, Missouri.
The PRRI survey determined the demographics of close social networks by asking each respondent to name up to seven individuals who are "the people with whom you discussed matters that are important to you" over the past six months and then following up with questions about the people named. By analyzing the information about many specific individuals, PRRI reached overall conclusions about whom Americans are closest to.
That approach was chosen over less reliable methods like asking respondents themselves to estimate the average characteristics of the people they usually interact with. As Cox said, "the reason we do it that way is that it's really challenging for respondents to provide that information accurately otherwise."
As the survey found, few black or white Americans tend to talk about critical matters with people in a racial group other than their own. Only 1 percent of the people white Americans said they talk to about important issues are black, while only 8 percent of the people black Americans said they talk to about important issues are white.
That adds up to not a lot of conversations happening between the two groups, which could lead to a lack of understanding about each other's concerns.
But some media outlets took the conclusions of the survey still further. "Three quarters of whites don't have any non-white friends," a Washington Post headline claimed. The Post article also declared that if every person is assumed to have 100 friends, 91 of the average white person's friends are white, and only one is black.
The Huffington Post got in on the action too, earlier claiming that "one of the most glaring statistics from the study revealed that 75 percent of white Americans are exclusively friends with those of the same race."
Here's the problem with these more sweeping conclusions: They are based on the assumptions that when the survey respondents chose up to seven friends or family members to describe, they chose seven people who are demographically identical, on average, to their entire friend-and-family network -- in the Post's example, the person's 100 closest friends -- and that if a person didn't list any non-white people among those seven possible choices, that person has no non-white friends at all.
The PRRI survey does not support those conclusions.
First of all, though respondents were allowed to name up to seven people, they listed on average only 3.4 individuals. That would suggest they were naming an inner circle of confidants, not all the people they might count as friends.
Second, "people with whom you discussed matters that are important to you" is not synonymous with "your friends." In fact, 1.8 of the 3.4 confidants the average person listed -- more than half -- were relatives, spouses or partners.
It remains true in early 21st century America that a person and his or her own parents, grandparents and siblings are highly likely to be of the same race.
Couples, whether married or established partners, are also highly likely to be of the same race. According to a 2012 Pew Research Center report based on data from the Census Bureau's American Community Survey, interracial marriages account for only about 15 percent of new marriages and only 9 percent of new marriages where at least one partner is white. Interracial marriages account for just 8.4 percent of all married couples, according to Pew.
The fact that marital relationships remain so segregated is concerning. But the factors involved in choosing a spouse are not identical to the factors involved in choosing friends. One's spouse is not necessarily like one's friends demographically or otherwise.
In PRRI's survey, less than half of the confidants listed by the respondents, or only 1.5 of the 3.4 people, were friends. (To be fair, PRRI's discussion of its survey results sometimes described that set of close confidants more broadly as a person's social network.)
The data about those friends still shows strikingly little diversity. According to Cox, 87 percent of the people listed by white respondents specifically as friends, not family members, were white.
But even there, it's easy to imagine how a network of up to 100 friends might be different from one or two very close friends, and it's certainly not obvious that every person with one or two white best friends has no non-white friends at all. In another PRRI survey, conducted in 2012, 71 percent of white Americans said they have a close friend or family member of another ethnicity -- although Cox cautioned against reading too much into those findings.
"Respondents may not want to admit to having no members of their social group who are of a different race or ethnicity," Cox said of the older survey. "The definition of a close friend may be a little elastic in this case."
Of the newer survey, he acknowledged, "It doesn't get to how diverse their entire network is," but he added, "Those more distant relationships also have less of an impact."
We have no idea, in other words, how many non-white friends the average white American has. But we do know that the vast majority of the people whom the average white American trusts most are also white -- and we don't need to make the data sound more dramatic than that to be troubled by the results.