A new sampling method developed by a Cornell sociologist will be used by the Centers for Disease Control and Prevention to measure the HIV risk behavior of injection drug users in 25 U.S. cities with the most new number of AIDS cases. The sampling method, called respondent-driven sampling (RDS), uses “snowball sampling” — creating a network of individuals to refer their peers, who in turn refer those that they know and so on –in combination with a mathematical model that weighs the sample to counteract the fact that the data was not collected randomly.
Prof. Douglas Heckathorn, sociology, developed the RDS method in 1997 for a National Institute on Drug Abuse HIV-prevention research project. “RDS resolves the dilemma of sampling hard-to-reach populations,” Heckathorn explained.
Before the RDS method, it was difficult to statistically sample inaccessible populations that were geographically clustered because there is no consistent sampling group. For public health officials looking to sample high-risk individuals such as injection drug users, prostitutes and men who partner with men this was particularly problematic. Gathering data solely from users on the streets and from needle exchanges tends to over sample older men and under sample women, young people and those who recently started using drugs.
“The [individuals] you miss are not the same as the ones you capture,” Heckathorn said.
Another type of sampling, the network-based method of “snowball sampling,” allows the sample to expand from wave to wave using the principle of “six degrees of separation.” However, this method is prone to bias, as people tend to refer others with similar racial, ethnic and socio-economic demographics.
But RDS makes it possible for the first time to draw statistically sound data from hidden or hard to reach populations for which there is no list of population members by using the “snowballing” method in combination with a mathematical model. The new method enables researchers to draw statistics from previously inaccessible groups that are relevant to public health, public policy and art and culture.
Although used internationally to study injection-drug users by both the CDC’s Global AIDS program and Family Health International, a non-profit public health agency, the method will be applied on a large-scale for the first time through CDC’s National HIV Behavioral Surveillance System. The program will monitor risk levels for injection drug users, high-risk heterosexuals and men who partner with men in a three-year rotation. Looking at trends over time, the CDC hopes to investigate how risk behavior of these groups changes.
“Past estimates on risk behavior are inconsistent because there hasn’t been any national surveillance program,” Heckathorn said. “[Before now] there wasn’t any national integrated effort to find out what the patterns were.”
Heckathorn has already used RDS to study jazz musicians in major metropolitan cities and will continue to use the method in the future to study aging artists and storytellers. Currently on leave from Cornell, Heckathorn is completing a book about RDS as a visiting scholar at the Russell Sage Foundation in Manhattan. An article that he co-wrote with former Cornell grad student Matthew Salganik, which proves that RDS is unbiased, will appear in the December issue of Sociology Methodology.
The 25 cities that will be monitored by the CDC were chosen based on the number of new AIDS cases. Approximately 500 people will be monitored from each of the following cities: Atlanta, Baltimore, Boston, Chicago, Dallas, Denver, Detroit, Ft. Lauderdale, Houston, Las Vegas, Los Angeles, Miami, Nassau-Suffolk, New Haven, New Orleans, New York City, Newark, Norfolk, Philadelphia, San Diego, San Francisco, San Juan, Seattle, St. Louis and Washington, D.C.
Archived article by Olivia Oran
Sun Staff Writer