Matthew Salganik tracks hidden populations to improve public health
Professor Matthew Salganik has helped develop statistical methods that unearth vital information about people such as drug users who elude public-health researchers.
Professor Matthew Salganik has helped develop statistical methods that unearth vital information about people such as drug users who elude public-health researchers.
Sameer A. Khan

In an age when many of our friends are only one click away, some people remain stubbornly out of reach. To public-health researchers, hidden populations such as drug users and sex workers are a longstanding Achilles’ heel: Their isolation conceals precious data, the lifeblood of epidemiology. In recent years, sociology professor Matthew Salganik has helped develop statistical methods that have unearthed vital information about these groups.

Those most at risk for HIV — a disease that afflicts more than 34 million people worldwide, according to the Centers for Disease Control and Prevention — are drug users who share needles, women who work in the sex industry, and men who have sex with men. Fearing social stigma or legal repercussion, members of these groups may decline to take part in surveys, or may be beyond the reach of researchers altogether. Consequently, efforts to prevent the spread of HIV have been hampered by lack of information on the size, composition, and behavior of populations most at risk. “This problem has existed for a long time,” Salganik says.

To conduct surveys, researchers used to canvass an area of known heroin users, for example, to identify those who had shared a needle. Tiny samples often were used to make rough generalizations, which some in the field considered inaccurate. Salganik, who has an undergraduate degree in mathematics and a doctorate in sociology, proposed viewing these vulnerable people “not as individual units, but as people embedded in social networks.” He championed the adoption of a study method that draws on the power of relationships. The method, known as respondent-driven sampling (RDS), asks a small number of survey subjects to recruit friends by handing out coupons that request their participation in the survey. When a new participant arrives with a coupon, the friend who gave it out receives a modest cash reward, as does the new participant, who leaves with more coupons to distribute.

Salganik analyzes the data from these surveys, which now are used throughout the United States and the developing world. Recently, he analyzed data from an RDS study of drug users in Curitiba, Brazil, where 303 people answered questions about their drug use and HIV-related behavior — starting from an initial sample of five. “Matt has been part of a very small group of people who advanced the math,” says Keith Sabin, a senior adviser on epidemiology at UNAIDS, an HIV/AIDS organization run by the United Nations. RDS “brings us into contact with people who have assiduously avoided any contact with government agencies.”

Salganik has received two grants from the National Institutes of Health for nearly $1 million over seven years to further develop RDS and refine his statistical methods. “When people think about social networks, they think about Facebook and Twitter,” he says. But valuable data can be gained from social networks that, he points out, “have been around for thousands of years.”