By reputation at least, London and New York City are rich and cosmopolitan, and the average one-stop-light town in the middle of Iowa is relatively poor and insular. But is a greater diversity of social connections really correlated with greater prosperity within a community?
Yes, says a team of researchers who studied the phone records—properly anonymized—of just about everyone who lives in the United Kingdom. Employing the mathematics of network analysis, the team found that communities in which people have more diverse social networks, at least as gauged by their phone calls, also tend to be more prosperous.
Longstanding social theory suggests that the extent of relationships among members of a community affects their well-being. And previous research has hinted that individuals with more diverse social networks tend to find jobs more easily, earn higher salaries, and have more success as entrepreneurs. But is there a similar link between the diversity of social connections and the prosperity of a whole community? That question has been tough to address because researchers lacked the data to measure a whole community’s level of social diversity, says Nathan Eagle, a computer engineer at the Santa Fe Institute (SFI) in New Mexico.
To find the answer, Eagle, telecommunications engineer Rob Claxton of SFI, and sociologist Michael Macy of the Brookings Institution in Washington, D.C., combined data from two disparate sources in the United Kingdom. First, to get a handle on a community's level of prosperity, they consulted the 2004 Index of Multiple Deprivation (IMD), in which the U.K. government evaluated the relative prosperity of 32,482 communities throughout the country using standards such as income, education, employment, health, crime, and housing. They first noted IMD for the region defined by each telephone exchange.
Second, the researchers tackled the harder problem: quantifying the diversity of a given community's social contacts. To pin down that slippery concept, they analyzed U.K. phone records for August 2005 that contained over 90% of the cellular phone calls and over 99% of the landline calls made in the country that month. Noting which phone called which, they developed a network with some 65 million point, or “nodes,” connected by some 368 million lines, or "edges," with each node denoting a phone and each edge denoting phone contact.
For each person’s phone within an exchange, the researchers calculated two measures of diversity of contacts using a mathematical concept from information theory called the Shannon entropy. In one, if a subject called, say, 11 different people during that month and spent roughly the same amount of time talking to all of them, the program assigned a social diversity index close to 1, denoting a high diversity of contacts. If, on the other hand, the caller spent almost all of the time on the phone with just one of those people, the index would be close to 0, denoting low social diversity. The researchers defined a similar measure of geographical diversity depending on whether a subject tended to call several different exchanges equally or just one or two exchanges.
Finally, the team looked for a correlation between their measures of prosperity and diversity. The team found that communities registering more social diversity also tended to be more prosperous. Curiously, diverse calling patterns correlated with prosperity more strongly than the average number of contacts per person or an exchange's total call volume.
Eagle notes that the team has seen a correlation only between greater diversity in social contacts and prosperity. So the researchers cannot tell whether social diversity promotes prosperity or the other way around. "The causality probably works both ways," Eagle says. He hopes the results will influence public policy officials to consider not just giving aid to poor, insular communities but also finding ways to help people foster relationships with others outside their communities.
The paper represents "an early illustration of the promise of computational social science," says behavioral scientist Noshir Contractor of Northwestern University in Evanston, Illinois. This team's efforts show how such vast data can help to tease out longstanding puzzles in social science.