Study Finds Alarming Rate of Risky Behaviors by Boys
Risky behavior among teenage boys--involving sex, drug use and violence--may be far more prevalent than youths admit in standard paper-and-pencil questionnaires, according to the first national survey to test a computer-assisted interviewing approach.
The findings, reported today in the journal Science, present a “disturbing picture of the biological and social risks that confront young males in the United States at the end of the 20th century,” the researchers wrote. These risks may be “worse than anybody thought,” said primary author Charles F. Turner of the Research Triangle Institute in Washington, D.C.
Among the potentially underreported activities identified by researchers were homosexual contact, intravenous needle sharing, substance abuse during sex and carrying weapons.
The study confirms what survey scientists have long suspected--that youths’ answers to sensitive questions depend largely on how they are asked and how much privacy they are afforded. Yet it also suggests that, at least for teenagers, computers are far and away the most promising research probe, providing a measure of comfort and ease unmatched by other approaches.
Teenagers, the study indicates, will tell computers things they won’t write down, much less tell someone else. In this case, teenagers were much more likely to reveal illicit or stigmatized conduct if recorded questions were posed through headphones and they could respond by pressing a key on a laptop, committing their answers to an invisible hard drive--as opposed to paper.
Study Is Called Major Step Forward
From a public health standpoint, the findings are discouraging, because they suggest that thousands more youngsters may be at risk of harm or disease than previously estimated. But experts on youth, who have long been skeptical of self-reporting data, hailed the study as a major step toward defining teenagers’ problems with greater precision and targeting resources accordingly.
“From a prevention standpoint, it’s a blessing,” said Joseph Catania, an associate professor of medicine at UC San Francisco who does national AIDS research.
“Health educators, parents, physicians [and] school nurses all . . . will need to pay attention to these data,” said Dr. Mark Schuster, a RAND researcher and UCLA pediatrician. “Their work should have a major impact on how we do research and how we address at-risk behaviors.”
Some scientists suggested that the study had broad implications for survey research among adults as well, in fields ranging from anthropology to economics.
“There are serious questions about the accuracy and quality of data we have looked at [until now],” said David E. Bloom, a Harvard economist whose commentary accompanied the study in Science. “Normally we think of computers expanding the quantity of data; now we see computers expand the quality.”
The study--begun in 1995 by scientists at Research Triangle Institute, the Urban Institute and the University of Illinois--compared two methods of inquiry. One group of 15- to 19-year-old boys was asked about their experiences with sex, drugs and violence on a “self-administered” written questionnaire, which is the standard method in research for probing sensitive questions. Another group completed the survey via an “audio computer-assisted self-interviewing” technique.
The respondents--1,672 in all--were assigned to the two groups randomly.
Under the computer method, questions were administered orally through headphones and respondents answered by pushing labeled keys. The process was private, with no need for an interviewer.
Though the questionnaire method is generally more accurate than face to face inquiry because of its relative privacy, researchers speculated that respondents may remain suspicious because their identification number and answers are recorded on the same form.
The audio-computer method has other pluses: It does not require literacy and it makes it easier to follow instructions, the researchers reported.
Significantly, Turner said, the responses of the two groups were parallel on questions relating to what might be considered “normative” behavior--such as having sex with girls. They diverged substantially only when the questions addressed behavior that might be considered abnormal or illegal--when privacy became relevant.
Among the more striking results:
* Computer respondents were nearly 14 times as likely to report sex with an injection drug user (2.8% versus 0.2%). They were nearly 10 times as likely to admit sharing intravenous needles. If the needle-sharing numbers are extrapolated to the wider population, the small percentage difference translates to 102,000 young men versus 9,300, Turner said.
* Just 1.5% of questionnaire takers acknowledged homosexual contacts, compared to nearly four times as many who answered via computer.
* Among sexually experienced computer respondents, more than a third claimed to be drunk or high the last time they had heterosexual intercourse--double the percentage of questionnaire respondents. Intoxication during intercourse impairs judgment and is considered a risk factor for HIV.
* Computer respondents were significantly more likely to admit that they had carried a gun in the last 30 days (12.4%) compared to the questionnaire group (7.9%) and to report that they had carried a knife or razor in the same period (27.1% versus 19.4%).
Findings Checked and Rechecked
“I was very nervous about the results,” Turner conceded. “These differences were big.”
So the research team checked and rechecked the data, drawing some reassurance when a University of Wisconsin researcher, William S. Aquilino, independently came up with similar findings in a video computer survey on illegal drug use among adolescent boys and girls.
Might teenagers have some reason to exaggerate their behavior on the computer? Turner said it is not likely. Teenagers’ responses are consistent with what adult males have reported in studies when asked to recall their teenage behavior.
There is some debate over whether these findings suggest broad problems with past survey research on adult populations and on other topics. Turner said he was reluctant to generalize his findings beyond the adolescent population.
Aquilino said that in a drug use study of adults, he found no significant differences when comparing the video computer techniques with questionnaires. This suggests that the adolescent study may not be applicable to adults. Yet both researchers said more research is needed on the question.
It is possible that teenagers are more comfortable with computers than adults, who sometimes struggle with the technology and have their own doubts about security of computer databases.
But several scientists said computers may show stigmatized behaviors among adults--such as child abuse, marital infidelity, abortion frequency--are as underreported as teenage risk-taking.
In his commentary, Bloom says that the Turner study “calls into question the validity” of a range of past survey research, including much-cited household surveys that are foundations for public policy decisions. He said Turner’s team “only scratched the surface of what is possible,” and their work should spawn a range of follow-up experiments.
Other researchers cautioned that the computer technology is expensive and difficult to implement, requiring, as Catania put it, “a lot of high-tech personnel with large salaries doing the programming.”
“That is not to downplay their success,” Catania said of the Turner team. “It’s a slick improvement” over the past.
(BEGIN TEXT OF INFOBOX / INFOGRAPHIC)
Electronic Openness
Youths ages 15 to 19 were questioned about their behavior through an audio computer technique were much more likely to report illegal or stigmatized behavior than those who filled out written questionnaires, according to the 1995 National Survey of Adolescent Males. There was little difference when questions were asked about non-stigmatized behaviors. Computers may have made the process simpler and conveyed an enhanced sense of privacy, researchers say.
STIGMATIZED BEHAVIOR: Ever had sex with a prostitute
% YES, ON PAPER: 0.7%
% YES, ON COMPUTER: 2.5%
****
STIGMATIZED BEHAVIOR: Ever been paid for sex
% YES, ON PAPER: 1.6%
% YES, ON COMPUTER: 3.8%
****
STIGMATIZED BEHAVIOR: Any male-male sex
% YES, ON PAPER: 1.5%
% YES, ON COMPUTER: 5.5%
****
STIGMATIZED BEHAVIOR: Had sex with someone who shoots drugs*
% YES, ON PAPER: 0.2%
% YES, ON COMPUTER: 2.8%
****
STIGMATIZED BEHAVIOR: You/your partner drunk or high during last heterosexual intercourse*
% YES, ON PAPER: 15.3%
% YES, ON COMPUTER: 34.8%
****
STIGMATIZED BEHAVIOR: Always/often drunk or high during heterosexual intercourse last year*
% YES, ON PAPER: 2.2%
% YES, ON COMPUTER: 10.8%
****
STIGMATIZED BEHAVIOR: You/your partner had been drinking at time of last heterosexual intercourse*
% YES, ON PAPER: 13.9%
% YES, ON COMPUTER: 25.4%
****
STIGMATIZED BEHAVIOR: You/your partner used drugs at time of last heterosexual intercourse*
% YES, ON PAPER: 9.7%
% YES, ON COMPUTER: 15.8%
****
STIGMATIZED BEHAVIOR: Ever taken street drugs using a needle
% YES, ON PAPER: 1.4%
% YES, ON COMPUTER: 5.2%
****
STIGMATIZED BEHAVIOR: Used crack/cocaine within last year
% YES, ON PAPER: 3.3%
% YES, ON COMPUTER: 6.0
****
STIGMATIZED BEHAVIOR: Someone threatened to hurt you in past year
% YES, ON PAPER: 25.7%
% YES, ON COMPUTER: 34.4%
****
STIGMATIZED BEHAVIOR: You threatened to hurt someone in past year
% YES, ON PAPER: 17.1%
% YES, ON COMPUTER: 26.1%
****
STIGMATIZED BEHAVIOR: Carried a gun in past 30 days
% YES, ON PAPER: 7.9%
% YES, ON COMPUTER: 12.4%
****
STIGMATIZED BEHAVIOR: You pulled knife or gun on someone in past year
% YES, ON PAPER: 6.2%
% YES, ON COMPUTER: 8.9%
****
STIGMATIZED BEHAVIOR: Someone pulled knife or gun on you in past year
% YES, ON PAPER: 16.9%
% YES, ON COMPUTER: 21.1%
****
STIGMATIZED BEHAVIOR: Carried a knife or razor in past 30 days
% YES, ON PAPER: 19.4%
% YES, ON COMPUTER: 27.1%
****
NON-STIGMATIZED BEHAVIOR: Sexual intercourse with female within last year
% YES, ON PAPER: 49.6%
% YES, ON COMPUTER: 47.8%
****
NON-STIGMATIZED BEHAVIOR: Five or more lifetime female partners
% YES, ON PAPER: 15.8%
% YES, ON COMPUTER: 18.8%
****
NON-STIGMATIZED BEHAVIOR: Sexual intercourse with female within last year
% YES, ON PAPER: 49.6%
% YES, ON COMPUTER: 47.8%
****
NON-STIGMATIZED BEHAVIOR: Condom use during last sex*
% YES, ON PAPER: 64.4%
% YES, ON COMPUTER: 64.0%
****
NON-STIGMATIZED BEHAVIOR: Drank alcohol last year
% YES, ON PAPER: 65.9%
% YES, ON COMPUTER: 69.2%
* Among males reporting sexual experience
