IN MAY 1998 a group of social scientists announced in Science magazine that they had found much higher rates of stigmatized sexual behavior and drug use among high school age young men than anyone had previously reported.
The article "Adolescent Sexual Behavior, Drug Use, and Violence: Increased Reporting with Computer Survey Technology," reported a survey in which 1,672 youths, 15-19 years old, were divided into two groups. One group answered questions using a common written questionnaire. The other group heard and saw the questions asked by a computer and answered using the computer keyboard.
To the surprise of the researchers, the computer users admitted far more homosexuality and drug use than the questionnaire users.
In particular: Only 1.5 percent of those using the written questionnaire admitted any homosexual behavior, but 5.5 percent of those using a computer did so.
Some of the details were interesting.
In the more candid (less dishonest) computer wing of the survey, the most common homosexual behavior the youths admitted was being masturbated by another male (3.5 percent), followed by being fellated by another male (3.1 percent). Then followed masturbating another male (2.6 percent), fellating a man (2.3 percent), insertive anal sex with a man (1.9 percent) and finally receptive anal sex with a man (0.8 percent).
In each case, the written survey produced far lower estimates.
Not surprisingly, the more stigmatized the behavior the greater the difference between the two sets of answers. For instance, five times as many computer users admitted fellating someone, and eight times as many computer users admitted receptive anal sex. (Hardly anyone using the written survey admitted receptive anal sex.)
The researchers were amazed!
For years, social scientists tried to convince us that people give fairly honest answers on written questionnaires because that format assuages their concern about keeping their behavior secret. And they have doggedly, arrogantly, defended questionnaires against frequent criticism that they were showing absurdly low estimates for many activities, especially homosexuality.
So now all those earlier findings are shown to be palpable nonsense -- and by a factor of three or four.
"We had the analysis done and redone, but this is real," senior author Charles Turner confessed frankly to The New York Times. "It means everything you thought about the risks adolescents face is an underestimate, if you're deriving your perceptions from past surveys."
But trusting those previous surveys is exactly what Charles Turner told us to do in the past, with as much confidence in the old method as he now places in the new method.
The researchers wondered why the computer elicited greater honesty. They might better have wondered why written questionnaires do so poorly and wonder if the computer method is much better.
Pioneer sex researcher Alfred C. Kinsey scorned written questionnaires, calling them "an invitation to lie."
As Kinsey wrote: "People, even when guaranteed anonymity, will not commit themselves on socially taboo and legally punishable activities like? homosexual activity, when they have to put it down in black and white. Practically all previous questionnaire experience indicates that they come out with figures very much lower?"
But the computer survey results are subject to many of the same objections as a written survey.
Kinsey and his colleagues were able to get greater candor from people because they asked a large number of questions very rapidly, minimizing the opportunity to lie, and used numerous cross checks on accuracy.
In addition, Kinsey's small group learned to use whatever language people were most comfortable with, depending on their educational level (e.g., sometimes using "fuck" instead of "have insertive anal sex"). They also were able to explain the question if someone seemed confused.
Finally the Kinsey group could change the order of the questions, dropping a line of inquiry if a person seemed uncomfortable, coming back to the topic later when the person was more relaxed.
Standardized questionnaires, whether on paper or on computer, are able to do none of these things. In fact, modern researchers, including Turner, strangely continue to assert the superiority of the standardized wording and question order for everyone, despite evidence to the contrary, and examples of inconsistency, lying and misunderstanding in this study.
The researchers would have understood this if they had read Kinsey, but modern researchers seem determined to forget everything Kinsey ever taught, and then congratulate themselves on rediscovering any small piece of it. Much social science progress occurs this way.
Consider some of the methodological problems and difficulties in the newest survey.
Most obviously, many people refused to participate at all, many perhaps having something to hide. These people are conveniently ignored.
The researchers asked ambiguous questions. For instance, they asked about sex with a prostitute, but not whether the sex was for money. They did not ask the gender of the prostitute, though I know of youths who have paid for sex with men.
The researchers seem naive. Nearly 4 percent of the youths said they themselves had been paid for sex, and most (3.5 percent) said they had been paid by women (twice the number who had sex with a prostitute). However, that is not the way the world works. Women rarely pay for sex. (Kinsey found that about 0.l percent of his female subjects had done so.) If the researchers had any knowledge of the world they would know they were being lied to, yet they report this without comment as a "finding."
Third, can the researchers really believe that more than twice as many young men are being anally insertive with males as are being anally receptive?
Fourth, Kinsey pointed out that stigma and hypocrisy about homosexuality, like homosexuality itself, are more common among the less educated. Here too, none of the youths who were below normal grade level in school admitted any homosexual behavior on the written questionnaire. But a higher than average number of them using the computer admitted homosexual behavior (6.2 percent). Turner reports this as if no one knew it before.
Given such clear evidence of homophobia, some youths are no doubt still lying. So the question is: Given this and other evidence of lying, why should anyone think the new computer technique has eliminated cover-up and finally gotten accurate answers?