by Dean Karlan
Fortunately, in most developing countries, the highest cabinet position on health is not occupied by a Dr. Beetroot type; nonetheless, bubbling springs of misinformation abound. They are amorous potential partners, extension officers for rural clinics, parents, pastors, and teachers charged with delivering health curricula. Jake once attended a church service in Ghana where the preacher testified that the Devil himself, manifest now as AIDS, wishes to infect us all, but that we can protect ourselves—“slam the door in that Devil’s face” was the direct quote—by singing praises to God. There was no mention of sex, safe sex, or even abstinence. Those unlucky enough to be downstream of such contaminated springs usually drink foul water, and so it comes to pass that millions are confused or ignorant about the issues.
Adolescents, especially young girls just beginning their sexual lives, have the least personal experience to draw on and the longest road ahead. If anyone could benefit from a shot of the straight dope, it’s them.
Sugar Daddies
Girls in Busia, Kenya, grow up fast. They always have. By tradition, women become marriageable once they reach childbearing age—usually around fourteen—at which point many are snapped up by eager suitors, often in some version of an arranged marriage. In recent years, the tide has begun to turn. Girls are spending more time in school before entering domestic life, and some are continuing to university and starting careers. More marriages are voluntary, fewer arranged by parents. But the girls of Busia are still confronted early and often with decisions about sex.
They might look to their mothers for advice, but their mothers grew up in a simpler time. Vying for the girls’ affections today are not just men from the village, but also hotshot businessmen from town and boys from school. The sheer variety of partners available would be surprising enough to the older generation; the notion that a girl could choose freely among them, a real shock.
Of course, the limited number of women means competition between suitors, and this plays out in Busia much as it does in America, or in Archie comics. Those who can afford it buy the girls presents or take them for rides in their sports cars. Those who can’t find other ways to impress, like scoring the winning goal in a football match, or sending florid text messages.
In general the girls are not sharks; they’re teenagers. But they can see which side their bread is buttered on. In private they refer to older, wealthier suitors as “sugar daddies.” For an adolescent girl, choosing a sugar daddy as a partner is like buying reproductive insurance. Should she become pregnant, he is prepared—and expected—to marry her and support the child.
But there is a catch. Being older and wealthier (and, in many cases, more lecherous), sugar daddies have had more past sexual partners, and now they have more diseases—HIV in particular—to show for it. Among Kenyan men ages fifteen to fifty, the highest incidence of HIV is about 8.5 percent, in the thirty-five-to-forty-four group. By comparison, the girls’ peers in the fifteen-to-nineteen cohort are practically harmless, with only 0.4 percent infected. So, whether she knows it or not, a girl who chooses a sugar daddy is making a trade-off.
A strictly dollars-and-cents view of the situation is that financial security in case of pregnancy, along with all the sports car joyrides and the courtship gifts, is effectively compensation for two things: First, the girl is being paid for her companionship, a scarce and sought-after commodity. Second, she is being compensated for the risks she has to bear as a companion, like the possibility of getting pregnant or of contracting a sexually transmitted infection. If girls already know about the different rates of HIV infection for older and younger men, then standard economic theory would assume they are extracting a fair price from sugar daddies for taking one additional risk of infection.
And if this really is the case, then simply providing girls with information about HIV incidence by age group shouldn’t change anything about who they choose as partners or how they get compensated. It should be old news. But if they don’t already know, telling them might make a difference.
To see whether it would, IPA researcher Pascaline Dupas, then as part of her Ph.D. dissertation, set up an experiment in 328 government schools near Busia in 2004. Seventy-one of the schools were randomly chosen to participate in the Relative Risk Information Campaign, in which a program officer met once, for forty minutes, with the eighth-grade class. At the beginning of the meeting, students completed an anonymous survey to determine how much they knew about the prevalence of HIV among Kenyans. Then the program officer screened a short film about partnerships between girls and adult men, which led to an open discussion on the issue of cross-generational sex. During the discussion, the program officer presented a detailed breakdown of HIV incidence in Kenya by age and gender.
At the same time and in the same 328 schools, the Kenyan government was reviewing its own HIV education program. During the previous year, half of the schools had been randomly selected to receive additional teacher training on the national HIV curriculum (which they were supposed to be teaching already). The curriculum included information on a range of topics from biology and transmission, to caring for infected persons, to the impacts of the HIV/AIDS epidemic.
Where Dupas’s program presented information but stopped short of telling people what to do, the government’s curriculum did not hesitate to give advice. It also featured a module on prevention, the spirit of which was captured in its message to students: “Say NO to sex before marriage,” and, even more succinctly, “Avoid sex.”
By piloting in half the schools, the government hoped to determine whether training teachers would ultimately help students. It was a perfect opportunity for a horse race, pitting Dupas’s information campaign against teacher training on the standard HIV curriculum. Comparing results of follow-ups across all 328 schools, they could see which program had bigger impacts on students’ choices about sex.
Deciding how to track those choices was a thorny issue in itself. The real item of interest was HIV status, easily determined by a blood test—but screening each student was not feasible. A first alternative was to ask students about their sex lives directly. About eight months after the programs had run, field teams administered a short survey on sexual activity, condom use, and demographics of sexual partners. There were legitimate doubts about the value of girls’ responses to the survey. They knew they weren’t supposed to be having sex—let alone unprotected sex. Why should they be expected to tell the truth on a questionnaire?
Another strategy was to monitor pregnancies as an indicator. Granted, it is an imperfect measure—only a fraction of unprotected sexual encounters lead to pregnancies—but the number of pregnancies is at least a lower bound for the amount of unsafe sex girls were having. Student pregnancies were also easier to observe because they were talked about by classmates, and, unlike surveys, were not susceptible to reporting biases. A big lump is difficult to hide.
The results on pregnancies were clear: The Relative Risk Information Campaign worked. Dupas’s program cut the incidence of childbearing by about a third. This suggested a significant decrease in the overall amount of unprotected sex, which was encouraging. Most of the reduction in childbearing, it turned out, was due to a 60 percent decrease in out-of-wedlock pregnancies. On the whole, that meant girls in the Relative Risk group who did get pregnant were much more likely to get married too.
Meanwhile, training teachers on the national HIV curriculum had no significant impact on overall rates of childbearing or on the likelihood of marriage for pregnant girls. Which is not to say it didn’t accomplish anything—quite to the contrary. There were big academic impacts. Teachers taught more about HIV, and students learned more, as measured by their scores on HIV knowledge tests. But the impacts on real-life choices about sex, captured in the follow-up survey, were harder to tease out. Girls in the teacher training schools reported having about 25 percent less sex overall, and about a third less unprotected sex too—but this is hard to square with the persistence of pregnancy rates. The girls in th
ese schools might just have been underreporting the extent of their sexual activity to the surveyors.
For students in Dupas’s Relative Risk program, it was just the opposite. In the follow-up survey, more girls reported being sexually active. That’s right: more sex with fewer pregnancies, and presumably fewer infections too.
The girls achieved this satisfying result by changing their partners and their habits. They began to choose younger men over sugar daddies, a solution that allowed them to lower their risk of HIV infection without giving up sex altogether. But how does the increase in the girls’ sexual activity jibe with the significant drop in childbearing we saw above? The key is protection. Most of the students who had more sex were using condoms. This is perhaps the most encouraging result of all, because the Relative Risk program didn’t have anything to say about condom use. Students were making the right choices on their own.
Just as Dupas had suspected, girls, armed with the relevant information, made better decisions. Either they extracted more compensation from sugar daddies in return for the higher risk of contracting HIV (as evidenced by the drop in out-of-wedlock pregnancies), or they found ways to lower the risk of infection—by choosing younger partners or practicing safe sex.
Paying for Testing
The Relative Risk program succeeded not just because it gave the girls of Busia valuable and relevant information about their prospective sexual partners, but because the girls took that information and put it to use. That’s an important distinction. The fact is that you can lead a horse to water, but you can’t make it drink.
Some portion of the bad decisions about sex can rightly be attributed to misinformation—thanks to scourges like Dr. Beetroot and the Ghanaian preacher Jake encountered—and in these cases, information-based approaches like the Relative Risk program can be the answer. But a big part of the problem is that we’re just stubborn horses. Most people around the world (and virtually everyone in developed countries) actually know enough about HIV and other sexually transmitted infections to protect themselves and their partners—if they want to. In that sense, information is not the issue. The issue is that, even though we know we should use protection, we don’t.
From a public health perspective, that’s just not good enough. People who know better but continue to make bad decisions don’t just endanger themselves; they pose a risk to the wider community. In that sense, HIV and other sexually transmitted infections are like the worm and malaria infections we talked about in chapters 9 and 10. Because individual protection redounds to the good of all, there are strong arguments for governments to step in and actively promote it.
Now, the most effective thing would be to have public health officials on the scene at the moment of truth, tearing open the foil wrapper and handing you a condom. Thankfully, that’s out of the question. Until we invite them into our bedrooms, they will accede to the demands of propriety, keep a respectful distance, and try to influence our decisions indirectly.
How? The sex ed class you took in high school was one way. Giving away condoms at the health clinics of many U.S. colleges is another. A third is advertising. A few of our favorites: (a) A billboard in Accra, Ghana, where naked cartoon silhouettes, coupling in an impressive variety of positions, cavort across the width to spell out “WEAR A CONDOM.” (b) Another billboard in Accra, Ghana, where there is a picture of a smiling couple and a slogan in big print below: “Just because you can’t feel it, doesn’t mean it is not there.” (What is “it”?!? I still wonder whether the advertisers intended this to be a triple entendre, double entendre, or no entendre at all.) (c) A billboard in El Salvador that translates to “Be faithful to your wife, or wear a condom.” (d) A video that shows a scene in a grocery store with a father and his child, who is throwing the worst temper tantrum you’ve ever seen, tearing cans off of the shelves and screaming at the top of his lungs. And then a simple tagline: “Use Condoms.” And a fourth way to influence decisions is to pay people in exchange for learning their HIV status.
Many HIV researchers and policymakers, both in and outside government, believe that people who learn their status will act accordingly. That is, those who are infected will protect others, and those who aren’t will protect themselves. This belief is the rationale for programs that emphasize testing. If it is true, then convincing people to learn their status is a viable, if indirect, route to the bedroom, a way to influence the important decisions that would be made in private. But that’s a big “if.”
Rebecca Thornton, an economist at the University of Michigan and IPA researcher, wanted to know for sure whether learning HIV status really did cause people to behave differently. So in 2004 she went to Malawi to find out. Thornton designed an RCT that offered participants free HIV tests and an opportunity to buy condoms. If participants who learned their status decided to buy more than those who hadn’t, there would be some hard evidence in favor of the connection between testing and decisions about sex.
The question is: How do you randomly select, for the sake of an evaluation, which individuals learn their status? Barring some people from testing altogether, or forcing tests on others, was unrealistic and, more important, unethical. But Thornton saw an opportunity in people’s general tendency to drag their feet about getting HIV test results. To be fair, there are good reasons for avoidance, not the least of which is fear, plain and simple. Jake remembers literally shaking in his seat at the Columbia University Health Services office while he waited to hear the result of an HIV test, even though he was quite certain it would be negative. (It was.) People in the Malawian context, where 12 percent of adults are HIV-positive (compared with 0.6 percent in the United States), had much more reason to fear bad news; but they could also have been inclined to stay in the dark for reasons far more concrete than existential trembling. The prospect of losing a day’s work, or a fivemile round-trip walk to the clinic, might have been enough.
Being an economist, Thornton had incentives on the brain. (It’s true—economists think about them a lot.) Figuring that more people would take action if it was worth their while, she built a reward into the HIV testing program, giving people money in exchange for learning their test results. And in order to see just how much encouragement was needed, she randomly varied the reward amounts, from nothing to three dollars. In a country where the average daily wage in 2004 was about a dollar, these were substantial sums, presumably big enough to impact people’s choices. That’s precisely why randomly assigning the rewards was so clever: The reward amount acted as a proxy for the likelihood that an individual would learn his or her status. Voilà—randomization without forcing anybody into or out of testing.
In the end, the program worked as follows: Health workers went door-to-door in 120 villages, offering participants free HIV tests. Each person who agreed to be tested gave a saliva sample on the spot and received a randomly chosen voucher redeemable at a mobile health clinic. If an individual went to the mobile clinic to learn his test results, he received the voucher amount in cash, and researchers recorded that he had learned his HIV status. About two months after results became available, participants were visited at home by surveyors who had no part in the testing portion of the program. They administered a short questionnaire about recent sexual behavior, gave the respondent about thirty cents in appreciation for his time, and then offered condoms for sale at a highly subsidized price. A package of three cost five cents; a single was two cents.
The study’s findings shed light on two questions. First, are incentives a good way to get people to learn their HIV status in the first place? Second, and more important, does learning one’s status really lead to better choices about sex?
There could be no doubt about the first: The incentives worked. Participants who received vouchers for any positive amount were more than twice as likely to get their results as those who got vouchers for nothing. Interestingly, the size of the reward—as long as it was greater than zero—seemed to matter less. For the nonzero vouchers, each additional dolla
r did increase the likelihood of learning results, but only marginally so. In fact, a voucher worth ten cents produced more than three-quarters the effect of one worth ten times as much. This is an important finding in its own right, as it gives us valuable information for designing policies and programs. If each additional dollar generates an increasingly weak responses, then there is a limit to the amount we can boost participation simply by throwing more money in the pot.
Incentives worked well enough in Malawi that 69 percent of people who were tested went to learn their results. But the critical question was whether those 69 percent would act differently once they knew—and here the results were mixed. On one hand, learning status had a substantial impact on HIV-positives who were sexually active. (It is worth nothing that these were only about 4 percent of all the participants.) Those who got their results were more than twice as likely to buy condoms as those who didn’t, which was a step in the right direction.
But, on average, those individuals bought only two more condoms than their counterparts who hadn’t learned their results. That’s not a whole lot of extra protection. For HIV-negatives—who made up about 94 percent of the participants—the results were dishearteningly bland. Those who learned their test results were no more likely to buy condoms than those who didn’t. The follow-up questionnaire about sexual behavior, administered when the condoms were offered, was also a wash. It found no difference in behavior among sexually active participants, whether they had learned their test results or not.
Thornton had to conclude that the program didn’t stack up. Door-to-door testing is expensive, and despite the boost from incentives, it appeared to generate only a small change in behavior—and only for HIV-positives, who made up just 4 percent of the participants. Resources could be better spent on other programs proven to have bigger impacts.
Beyond a verdict on this cash-for-test-results program, there are some general lessons to take away from Thornton’s study.
