But it's not as simple as preference. Young people, especially African-Americans, are also more likely to be exposed to those in their own ethnic communities, making segregated school communities all the more inevitable.
A team of economists at Stanford and the Universities of Venice and Siena used data from the U.S National Longitudinal Study of Adolescent Health (Add Health), which collected information from students in grades seven through 12 at 84 schools across the country.
In examining data on nationwide friendship networks, the team determined that students in particular tended to form friendships based on common ethnicity to a greater extent than individuals in the general population did. Their findings were published in the Proceedings of the National Academy of Science.
Racial self-segregation is part of a phenomenon called "homophily," the tendency of individuals to bond with those who share similar characteristics. In this case, the team determined that the selective bonding has two sources: Students can be biased in whom they befriend, or might be bound by inherent bias in the opportunities they have to meet people in different ethnic communities.
Of course, polling students on racial sentiments, or simply observing their friendship groupings, isn't an accurate way to determine how ethnicity affects relationships. So the team used a well-known economic method, called revealed preference theory, which examines individual preferences within the context of opportunities.
In economic discussions, that might mean choosing an orange over an apple when there's an opportunity to pick one of the two. Within a high school, the theory looks at choice of friendship, given the size and ethnic demographics at play.
The team found that every racial group exhibited some degree of bias toward segregated friendships, as well as having a greater chance of meeting students of their own ethnicity in the first place.
They used the data to create "estimated biases." For example, black students valued friendships with non-black students 55 percent as much as those with other blacks. On the other hand, Asians valued friendships with other races 90 percent as much as those with other Asians. Hispanic and white students fell somewhere in between.
The team also created an estimated bias evaluation for the opportunity students had to meet individuals not of their ethnicity. White students didn't have any bias, meaning they had ample opportunity to interact with those of diverse races. Asians and blacks exhibited biases of 7 to 7.5, and Hispanics ranked 2.5. An interaction bias of 7 means that over 90 percent of people one meets are those of the same ethnicity, so long as that ethnicity makes up 50 percent of the population.
In their analysis, the researchers conclude that the significant homophily of black groups is likely due, in part, to their bias of opportunity. Whites, on the other hand, exhibit segregationist tendencies exclusively because of racial preference.
The biases were more extreme at bigger schools, which offer an increased opportunity for students to self-segregate.
Prudence L. Carter, a sociologist at Stanford who specializes in racial and ethnic relations in education, said the team's data reflect what she's found in her own research. She thinks the greater bias among African-Americans comes down to support.
"Because black kids tend to be such a minority, they segregate for a sense of community," she told AOL News. "They're trying to get that connection."
Carter adds that academic tracking (where students are separated by academic ability into different classes) and extracurricular activities often exacerbate self-selected segregation at high schools. White and Asian teens are more likely to participate in advanced placement classes, and different ethnic groups often partake in extracurricular activities that reflect their history, cultural preferences and established norms.
"We've achieved spatial integration in schools," Carter said. "But sitting next to somebody in a classroom does not mean the same thing as meaningful integration.
"Educators need to be aware of the academic environment, who's participating in what, and how financial opportunities can impact that participation and that level of achievement," she adds. "This remains a class issue."
The original Add Health data used by the team was collected between 1994 and 1995, with a follow-up in 2008. Because such a large, widespread survey has yet to be repeated, Stanford economist Matthew Jackson said his team relied on assumptions that its data remain valid today.
But he also acknowledges that contemporary forums, like Facebook, might be transforming how teens forge relationships. "They facilitate meeting people who have similar characteristics and similar interests; they can help in seeing who your friends' friends are as well as monitoring what your friends are doing," he told AOL News. "It is hard to know what the ultimate consequences of this will be, but it certainly has the potential to have a significant impact on friendship patterns."
Carter agrees that online forums can facilitate relationships, but notes that they're unlikely to be meaningful. "We all have hundreds of friends on Facebook," she said. "But who are kids close to, dating, inviting to their houses? Where are meaningful connections taking place?"
No matter the impact of 21st century media, understanding teen homophily has implications for beyond the classroom. Jackson's main area of study is the economic impact of social networks. The roots of homophily help explain everything, he says, from "how we form opinions to which jobs we end up with and how likely we are to be employed."