The week of January 18, 2015

The weird racial politics of online dating

By Aaron Sankin

We’re in the midst of a cultural sea change to one of the most central institutions in the life of the nation.

American attitudes on interracial relationships have taken an enormous step forward in the last two decades. As recently as 1995, fewer than half of all Gallup survey respondents favored interracial marriage—and only 4 percent did in 1985. Now such sentiments are relegated to shadowy Internet message boards and corners of right-wing talk radio.

And yet, while the actual number of interracial relationships in the United States is certainly climbing, the overwhelming majority of Americans are in relationships with another person of their same race. In 2010, only about 15 percent of new marriages were interracial—bringing the total number up to 8.4 percent from 3.2 percent in 1980. Based on random matching alone, the expected proportion of interracial relationships in the United States should be as high as 44 percent.

This disparity indicates there’s still a considerable disconnect between what people think is “acceptable” when it comes to dating versus what they actually do themselves.

There are two possible explanations for this gap. On one hand, it may be that people tend to pick mates from their real-life social groups—people with whom they live, work, socialize, and go to school—and in the U.S., those are still largely structured by race. The other option, of course, is that most people, when given the choice, still prefer to be in relationships with someone who looks a lot like them, regardless of what they may tell a pollster.

Online dating sites like OkCupid and Tinder have given researchers a new window into how people conceptualize what they want (or don’t want) in a romantic partner. As it turns out, race is a huge factor when it comes to making romantic connections online, one that puts certain groups at persistent, structural disadvantages.

There’s a good chance, however, the growing prevalence of online dating may actually be having the effect of breaking down racial barriers instead of erecting new ones.

The Tinder effect

People of the same race are inevitably going to have at least some shared experiences, simply because, in many ways, they are treated the same by the culture at large. That’s just how racial identities are formed. So it would make sense that, outside of physical characteristics like skin color and eye shape, Asian people would have significantly more in common with Asian people, and black people would be more compatible with black people, and so on.

According to Christian Rudder, the Harvard-educated data whiz who founded OkCupid, that’s not actually how it works. It’s just how people think it works.

“OkCupid users are certainly no more open-minded than they used to be.” —OkCupid founder Christian Rudder

In a 2009 post on the dating site’s OkTrends dating research blog, Rudder noted that there’s very little variation in how people of different races match up with each other based on the site’s algorithm, which analyzes their interests and spits out a score showing their compatibility. There is a tight correlation between how well two people match each other and how likely they are to message each other back and forth—the best sign the site’s operators have that a relationship is blossoming.

Using the signs of the Zodiac as an example, Rudder found that every single Zodiac sign matches every other Zodiac sign at exactly the same 60 percent compatibility rate—save for a one point drop for inter-Aquarius pairings. The rate at which members of each Zodiac group respond to messages from other Zodiac signs is basically identical for every possible match. This distribution of compatibility across the astrological spectrum should probably both tell you about how little astrological signs actually matter and also serve as a useful control group for looking at how OkCupid users deal with race.

Rudder found that people of different races tend to match each other at roughly even rates. The matching rates of each group to all the others spanned only a small range of 56 to 62 percent comparability. In some cases, certain groups had higher compatibility scores outside of their races—for example, Hispanic/Latin men paired up one point better with black and Middle Eastern women than they did with women of their own ethnicity—but the margins weren’t statistically significant. The major takeaway, judging from the numbers, is that almost all groups should be about equally compatible with each other.

That is not even remotely close to how it works in real life.

Response rates across racial lines diverge wildly. Black men and women get far fewer responses to their initial inquiries then virtually any other group across the board. White men get the most responses. White women strongly prefer men of their own race to all other races or ethnicities. Asian and Hispanic women are actually more likely to respond to white guys than Asian or Hispanic men. Despite being the most likely to respond to messages themselves, black women tend to have the lowest rate of messages received—from any race, including black guys.

Fast-forward five years, and Rudder looked at the same question again using fresh data. Had anything changed? No. In fact, things may have actually gotten worse. “OkCupid users are certainly no more open-minded than they used to be,” he wrote in a blog post. “If anything, racial bias has intensified a bit.”

The one thing that had changed was users’ willingness to proclaim they had no racial preference while still clearly acting on the same racial prejudices. Welcome to post-racial America, everybody.

This difference between people’s stated racial preferences in online dating and how they actually behave has been replicated in other research. A study of a large online dating site conducted by researchers at Stanford and Harvard in 2009 found that people on the right side of the political spectrum were far more likely than liberals to explicitly state that they were exclusively seeking partners of the same race, but both parties ended up displaying similar preferences.

Welcome to post-racial America, everybody.

“Both men and women of all political persuasions act as if they prefer same-race relationships even when they claim not to,” the researchers wrote. “As a result, the gap between conservatives and liberals in revealed same-race preferences, while still substantial, is not as pronounced as their stated attitudes would suggest.”

The researchers suggest that what’s going on here isn’t overt racism. Instead, it’s an interplay of social issues operating just below the surface. “Individuals often have inaccurate beliefs about their own preferences,” they explain.

When someone sees another person’s face, their brain, subconsciously and automatically, runs through a litany of images and associations, each coming with their own individual value judgment, that help that person make sense of the world. One method psychologists use to isolate these judgments is through a process called the Implicit Association Test.

In 2006, a pair of New York University neurologists conducted a study using the Implicit Association Test to determine racial bases among 150 white college students. The students were shown a series of faces of black and white people and then were immediately asked to categorize words as either positive or negative as quickly as possible after looking at a face. The researchers found that people were significantly faster at sorting unpleasant words after being shown a black face and, conversely, faster at sorting pleasant words after looking at a white face. Whether liberal or conservative, the biases were present among pretty much everyone.

White meant good, black meant bad. It worked at a speed faster than that of rational thought—on a level that people weren’t even conscious of.

“If you take an IAT, you may notice a familiar feeling,” recounted writer Zach Stafford in a recent opinion piece in the Daily Dot. “The test feels like being on Tinder.”

A sea of data

When BuzzFeed writer Anne Helen Petersen built her own tiny version of Tinder for a study she conducted on bias in online dating last year, her conclusion was a good bit more interesting than “people are racist.”

Petersen, who got her doctorate in media studies from the University of Texas at Austin, took about 60 stock photos of individuals (30 men and 30 women), ran them through Instagram-like filters for authenticity, and nestled them in the middle of Tinder frames. She then circulated the experiment on social media, letting participants swipe left or right based on attractiveness, just like real Tinder. But then she also asked them a number of questions about their judgments of each person based on appearance of the people in the images. Petersen didn’t just want to see that someone said yes or no; she also wanted to know why.

Petersen’s argument is that people’s primary issue is class, and they use race as a marker, consciously or not, to determine it.

For both men and women, the best performing stock photo models were black. These results don’t jibe with the findings from OkCupid until you start to look at what assumptions the participants in Petersen’s experiment made about the two people who performed the best. They both read as college-educated and middle-class. Nothing in their clothes or in the background of their pictures carried signifiers of African-American culture. Petersen’s argument is that people’s primary issue is class, and they use race as a marker, consciously or not, to determine it.

The strength of the best-performing woman in Petersen’s study, whom she named “Yasmin,” is that while she read as black, she didn’t necessarily read as exclusively black. Forty-eight percent of the people who looked at Yasmin’s picture said she looked “mixed race.”

Research has shown that people who appear multiracial on dating sites are typically viewed as the most attractive potential partners. A study based on data from an unnamed online dating site conducted by scientists at the University of Massachusetts Amherst and the University of Texas at Austin found that individuals from certain mixed-race groups outperformed even whites, who tend to the do the best.

To confound things further, an analysis of data from Facebook-linked dating app Are You Interested found that men of each racial group preferred women from another race over their own. Other studies have shown that the more attractive someone is, the less likely they are to be concerned with the race of their potential partners. Hot people, as it turns out, just like other hot people.

Basically, we’re swimming in a sea of data on people’s racial preferences that shows hierarchies where certain groups get preferential treatment based solely on the color of their skin, despite actual levels of compatibility. But nothing about it is particularly simple or precise, and preferences aren’t necessarily segregated into homogenous racial silos.

The tipping point

Sometimes things that happen on a dating site can change the way someone looks at race and relationships.

A study from University of California, San Diego sociologist Kevin Lewis published by the journal Proceedings of the National Academy of Sciences in 2013 found something fascinating buried inside a mound of data from OkCupid. Lewis noted a lot of the same things as other researchers, but he also noticed how people’s preferences changed over time.

After an OkCupid user received a message from a person of a different race or ethnicity, their interactions with other people of that race or ethnicity had a tendency to skyrocket. After that first interracial contact, a person would, on average, increase their interactions with people of that race by 115 percent. There was no halo effect. If a white woman was messaged by a black guy, her interactions would only increase with black guys with no marked change on Hispanic or Asian men.

Hot people, as it turns out, just like other hot people.

In some instances, the effect was super-charged. After Asian women were contacted by someone of a difference race, their rates of messaging people of that race jumped by nearly 238 percent. For Asian men, the increase was almost as high.

Lewis suspects that what’s happening is that a lot of people don’t send messages to people of certain races or ethnicities out of fears about a lack of shared experiences or a disinclination toward future rejection. Getting that initial message effectively tells them there may be nothing to worry about. Suddenly, that person’s perceived pool of potential mates expands considerably. Since OkCupid’s own data shows actual compatibility has little to do with race, getting people past that first step of deciding to send an initial message is huge.

Lewis argues that determining whether online dating puts people in a better position to make these jumps is tricky because, to his knowledge, no one has ever directly studied the difference between the online and offline arenas.

“There are arguments on both sides: The Internet dissolves boundaries because it makes identity more fluid/less salient,” he told the Kernel. “The Internet recreates boundaries because it makes it so much easier to be biased without personal/social penalty.

“Personally, I believe the differences are probably much weaker than we would probably expect. At the end of the day, I am the same person online and offline, and I am interested in the same things. … Certain interfaces just make these goals easier or harder to realize. “

However, it seems that just by throwing so many people into one giant pot, online dating has the potential to put more of them into a shared virtual space they wouldn’t otherwise inhabit. It’s a world where preexisting, overlapping social networks don’t matter quite as much.

For the most part, people will probably continue to represent themselves online as the same jumble of oft-contradictory prejudices that they do in the real world. That will inevitably make the rejections felt by people endlessly rebuffed or ignored simply for the color of their skin continue to sting. But if the act of using an online dating site is an expression of hope over past disappointment, maybe holding the institution of online dating to a similar standard may not be so foolhardy.

llustration by J. Longo