A study in May’s Science showed that pen-and-paper queries have gone the way of Catholic confessionals as reliable measurements of teen behavior.

The National Institutes of Health electronically quizzed 1,729 African-American, Caucasian and Latino boys nationwide, ages 15 to 19, on such touchy topics as homosexuality and intravenous drug use. Questions were fed through headphones and respondents punched in numbered keys to answer. The technique was meant to eliminate the literacy problems that have marred previous teen studies.

Judging by what they keyed in, American boys engage in much riskier behavior than previously believed, and are more at ease ’fessing up to a computer than to a person. Typing teens were almost four times as likely to report male-male sex than those filling out paper questionnaires. Also, reports of injection drug use tripled with the new techno system.

The jump in admitting behaviors that might foster unsafe sex shocked the survey’s lead author, Charles Turner of Washington, DC’s Research Triangle Institute, who said, “We have always known written surveys don’t reflect reality. But nobody thought the discrepancy was this large.” How the findings will filter into HIV prevention has yet to be determined, but Turner called the method a research coup: “Better tools provide better data, and better data provides better programs.”

Clinical psychologist Walt Odets called the quest for numbers misguided. “Honing down the data to percentage points isn’t a prevention tool. We know boys are exposing themselves to HIV. What difference does it make if it’s 49 percent or 60 percent? There’s this idea that if we get the figures a little more accurate, we’d get closer to the ‘truth.’ But the notion that prevention is done scientifically is a masquerade.”