In population science, "You want to describe not only what you see, but also what you are missing; not only the appearance but also the reality," says M. Giovanna Merli, Associate Director of the Duke Population Research Institute (DuPRI). Merli has made a specialty of finding hidden populations and uncovering hidden truths about them, all in the world's most populous country, China.

To better understand the realities at the heart of health policy questions, Merli's approach often blends classical demographic methods with a range of modern tools, such as social network modeling. Whether she is simulating the behavior of actual people to gauge its potential to spark an epidemic, or tapping into real-world social networks to characterize an unknown population, Merli's findings frequently challenge widely held assumptions.

In recent work she explores, for instance, whether heterosexual transmission of HIV could lead to the kind of rapid disease spread seen in some other countries and forecast for China. Given that half of HIV infections in China are thought to happen through heterosexual sex, the great majority of those encounters paid, in early 2008 Merli obtained interviews with 550 female sex workers in Shanghai about the details of their clientele and frequency of their contacts.

The women ranged from "high-end" call girls to the lowest-status street walkers, and social network specialist James Moody, DNAC Director and fellow DuPRI Research Associate, used their responses to simulate their web of contacts and interactions. The picture the data painted of China's sex trade showed a network of sex workers, customers, and partners of customers still far too constricted to sustain HIV transmission on the scale it has occurred, for example, in Thailand or parts of Africa.

To Merli, the results illustrate the principle that predictions can't be based on past experience in other places, especially when it comes to China.

Her fascination with that country began with its language. As a teenager in her hometown of Milan, Italy, Merli had already mastered several European languages when she decided to learn Chinese in the Department of East Asian Studies at the University of Venice. China "was just beginning to open up," Merli says, and she ended up living in Shanghai for two years through one of the first student exchange programs between Italy and China.

She has returned many times as a scholar, and finds her time living in China helps in designing studies there. "If you know how the society is organized, you can better foresee potential problems," Merli says. "That's particularly true in China where the political bureaucracy had such a huge role in monitoring and determining outcomes, so I'm sensitive to the ways that political issues can creep in and interfere with the collection of statistical data."

SUBTLE AND PROFOUND SHIFTS

China's rising mobility and fast-forward social change can make Merli's work challenging, but not as difficult as some might think in light of her frequent focus on sexual behavior and fertility. China's political history actually makes surveying Chinese on their sexual and reproductive lives easier in some ways, Merli says.

State monitoring of daily life, particularly oversight of the national one-child policy, accustomed citizens to giving officials detailed answers about their intimate lives on a regular basis. As a result, population researchers conducting surveys in China could once routinely get response rates of 95 percent or more. Today, they're nearer to Western norms of about 65 percent. The breakdown of rigid social control in China, particularly in cities, has led its people to "acquire a new sense of privacy," Merli explains. "They don't feel they need to appease local authorities, so they refuse to answer questions. They're pretty much behaving like the U.S. population now."

Nonetheless, in 2007-2008 Merli and her collaborators at the Fudan University School of Public Health accomplished the first-ever survey of local sexual networks in China. The Shanghai Sexual Behavior and Sexual Networks Survey asked a population-representative sample of 1,600 men and women in that city of 19 million about their sexual activity and partners as well as their reproductive preferences.

Responses to the sexual behavior questions were fed into the network simulations Moody and Merli generated to model potential HIV spread. And Merli used the participants' answers about their reproductive ideals to examine what might happen if China lifted its controversial one-child policy established in 1979.

When the family-size restrictions were first announced, they were only supposed to last a generation. Now more than 30 years later, as scholars and politicians debate whether to rescind the decree, a generation of singleton kids are themselves getting married and making choices about having a family.

In a 2000 study based on early-1990s data, Merli found that birth rates among rural Chinese families were low, but slightly higher than official reports claimed—in part due to underreporting of female babies born to families trying for a son. Merli argued that these girls were likely hidden with the knowledge of local officials charged with enforcing the one-child policy and motivated to meet their quotas. Still, a subsequent examination of the same data set in 2002 showed that in some rural areas, the one-child policy had not only succeeded overall in changing reproductive behavior, but had also started shifting personal preferences toward having fewer children.

In the December 2011 issue of Population, Merli makes the case that based on her more recent Shanghai survey sample, the shift in fertility preferences, at least in Chinese cities, is sincere and grounded in a new individualism and consumerism that have little to do with official fertility policy. Her respondents professed in many cases to be content with a single child or two at most. They often cited reasons like the expense, lifestyle constraints, and career costs of large families that will ring familiar in other nations where fertility rates have fallen naturally as prosperity rose.

TOOLS OF THE TRADE

Though the Shanghai residents in the larger sexual-networks survey were selected through standard census-based probability-driven sampling techniques, Merli's sex-worker survey used a recruitment approach gaining popularity for reaching people in the shadowy margins of society. Respondent Driven Sampling relies on an initial group of survey respondents to invite select associates to participate, with the chain of referrals revealing an ever-larger sample of the target population. The U.S. Centers for Disease Control and Prevention has used this method to survey intravenous drug users and men who have sex with men. Other researchers are employing it to assess health risks among street youths and the homeless.

"RDS is widely used now, in fact it's too widely used," says Merli. "It's a very efficient way to quickly and easily sample hidden populations," she explains, "but it also makes claims about population representation based on assumptions that are only now starting to be tested." Indeed, because the technique is becoming so popular in public health research, the National Institutes of Health is sponsoring several studies, including one by Merli and Moody, to evaluate its validity.

The team will use the Shanghai sex-worker survey data set, as well as data collected for this purpose in the southern Chinese city of Liuzhou, to test assumptions about the respondents' actual underlying social networks and improve estimations in those populations. "What's unique is we're really grounded empirically," Merli says. "We really want to understand how it works on the ground."

Population scientists are getting better answers and asking harder questions, Merli says, in large part thanks to new tools—such as network modeling, econometric methods of determining causality, and new, richer, more complex data sets—borrowed from other disciplines to enhance traditional demography. In turn, she adds, population scientists contribute their own tools and insights to other fields—for instance, by refining how populations examined in public health studies are chosen and how the results are interpreted.

As head of the new DuPRI Development Core, Merli's goal is to offer students the entire toolbox available for studying populations and to train them to take advantage of DuPRI's unique interdisciplinary opportunities. "It's unusual to be so diverse and to have such a welcoming and collaborative group," Merli says of DuPRI. "We are in public policy, economics, sociology, neurosciences, global health." Indeed, that "dynamism" is what Merli says attracted her to Duke four years ago. "This is a place where you can take risks in terms of the kinds of studies you take on," she says.

RISKS AND REWARDS

The next project Merli has planned is all about risk. This spring, she hopes to launch a pilot study of Chinese migrant workers in Tanzania. "Of particular interest to me is the negotiation of the Chinese migrants with a very new and challenging disease environment, and the risk [they] take to go to Africa," Merli says. Some are sent by the Chinese government for temporary assignments, but many are migrating on their own. No one knows how many, or whether their willingness to take such a leap will translate into riskier behaviors in their new setting.

The project brings together many threads from Merli's previous work. She'll conduct RDS surveys among the workers, model their social networks, and even use network modeling as a tool to get an estimate of how many Chinese workers might already be in Africa. "It'll be an opportunity to test the claims about RDS in a very different population," she notes. And, of course, if workers' risk tolerance and behavior are changed by their experiences abroad, it would be interesting to see how those who return home contribute to the ongoing social changes there.

Merli considers herself an original scholar whose topic, ultimately, is China. It is a rich "laboratory" for population science, she says. "The very rapid pace of change means you can observe processes that have been seen unfolding elsewhere over decades or centuries, there you can observe them sometimes in just a few years."