In Photos: The Prince and Princess of Wales Visit Boston and Harvard
‘Growing Pains’: Harvard Undergraduate Association’s First Semester Draws Mixed Reviews
Harvard Law Professor Asks Judge to Unseal Sidebars from Admissions Trial
Harvard FAS Releases Schedule for Previous-Term Course Registration
Nearly 250 Harvard Affiliates Sign ‘Free Speech’ Petition Addressed to University Presidential Search Committee
A paper published last month by Opportunity Insights, a Harvard-affiliated economic research group, shed new light on the ties between social connections and economic mobility, finding a strong link between cross-class Facebook friendships and economic outcomes.
The paper, titled “Social Capital II: Determinants of Economic Connectedness,” was the second of two papers led by Economics professor Nadarajan "Raj" Chetty investigating the effect of social ties across socioeconomic classes and economic mobility. The dataset used for the research — 72.2 million Facebook users and some Instagram accounts — represents by far the largest sample size of any study conducted on the topic.
Using a machine learning algorithm, the researchers estimated the socioeconomic status of each user in the dataset based on information from their Facebook profile, such as location, phone model, and education. The first paper found a strong link between the “economic connectedness” of an area — the “share of high-[socioeconomic status] friends among individuals with low [socioeconomic status]” — and the economic mobility of that area.
The second paper found two equally important barriers to economic connectedness: “exposure” to people of higher socioeconomic status and “friending bias,” defined as “the tendency for people with low [socioeconomic status] to befriend people with high [socioeconomic status] at lower rates even conditional on exposure.”
There are not “any ironclad explanations” for what causes “friending bias,” said Economics Ph.D. candidate Drew Johnston, a co-author of the paper, but researchers have identified some correlating factors.
“One of the things we found that was kind of striking was that smaller schools tended to have lower levels of bias, perhaps because there weren't enough people in there to allow it to segment into different groups,” Johnston said.
Florian Mudekereza, an Opportunity Insights pre-doctoral fellow who co-authored the paper, wrote in an email that “institutional structure” can also play a role in determining “friending bias.” He wrote that the portion of the second paper focusing on “friending bias” in high school stood out.
“What struck me the most is the fact that high schools with [a] history of conducting academic tracking tend to have relatively high friending bias,” Mudekereza wrote.
But the root cause of friending bias has yet to be found, Fluegge said.
“It’s a great area for future research because what we’ve documented is that it’s quantitatively really important,” he said.
—Staff writer Alexander I. Fung can be reached at email@example.com.
Want to keep up with breaking news? Subscribe to our email newsletter.