Sex on television makes teens pregnant

The front page of today’s Washington Post blares a new and troubling finding, Teen Pregnancies Tied to TV Sex. According to the article:

Teenagers who watch a lot of television featuring flirting, necking, discussion of sex and sex scenes are much more likely than their peers to get pregnant or get a partner pregnant, according to the first study to directly link steamy programming to teen pregnancy.

The study, which tracked more than 700 12-to-17-year-olds for three years, found that those who viewed the most sexual content on TV were about twice as likely to be involved in a pregnancy as those who saw the least.

“Watching this kind of sexual content on television is a powerful factor in increasing the likelihood of a teen pregnancy,” said lead researcher Anita Chandra. “We found a strong association.” The study is being published today in Pediatrics, the journal of the American Academy of Pediatrics.

That claim sounded so improbable to me that I sought out a copy of the paper to read the full study. Does TV watching lead to teen pregnancy? No, and the authors of the study are irresponsible to suggest that it does.

What does the study show? Here’s how the authors of the study, Does Watching Sex on Television Predict Teen Pregnancy, describe their findings:

Data from a national longitudinal survey of teens (12–17 years of age, monitored to 15–20 years of age) were used to assess whether exposure to televised sexual content predicted subsequent pregnancy for girls or responsibility for pregnancy for boys… We measured experience of a teen pregnancy during a 3-year period.

RESULTS. Exposure to sexual content on television predicted teen pregnancy … Teens who were exposed to high levels of television sexual content (90th percentile) were twice as likely to experience a pregnancy in the subsequent 3 years, compared with those with lower levels of exposure (10th percentile).

CONCLUSIONS. This is the first study to demonstrate a prospective link between exposure to sexual content on television and the experience of a pregnancy before the age of 20. Limiting adolescent exposure to the sexual content on television and balancing portrayals of sex in the media with information about possible negative consequences might reduce the risk of teen pregnancy. Parents may be able to mitigate the influence of this sexual content by viewing with their children and discussing these depictions of sex.

The study suffers from so many problems that it is difficult to know where to begin. First of all, the study does not, indeed cannot, show that watching sexual content on TV leads to teen pregnancy. The study only looks at whether there is a correlation between sexual content of TV viewing and subsequent teen pregnancy. As anyone who has studied statistics knows, correlation is not the same thing as causation. Two independent events might appear to be correlated, but that does not mean that one caused the other.

For example, over the past 100 years, recreational use of marijuana among teens has increased dramatically. In the same time period, teen death from infectious disease has decreased equally dramatically. We could draw a graph and perform statistical analyses that show that increased marijuana use is associated with decreased death from infectious disease, but that does not mean that smoking marijuana protects teens from death. Simply demonstrating a correlation does not tell us anything about causation.

Even if the authors had shown a causal relationship between viewing sexual content on television and teen pregnancy, that would not tell us which was cause and which was effect. The authors claimed that viewing sexual content led to pregnancy, but it is equally likely being sexually active led teenagers to prefer sexual content compared to their sexually abstinent peer.

The study suffers from major technical problems. The study was based telephone interviews of teenagers and relied solely on their honesty, accuracy and recall; teenagers are not noted for their honesty, accuracy and recall. Three separate interviews were conducted at predetermined intervals over 3 years. More than ¼ of the participants dropped out of the study, and the authors simply ignored them. However, those who dropped out of the study might have differed in significant ways from those who remained in the study, and their absence may have led to erroneous findings.

Of the 1461 teens who remained in the study, 146 (fully 10%) refused to divulge whether they were sexually active. The authors simply ignored them. Of the remaining 1315 teens, 571 (43%, or almost half) were not sexually active at all. The authors simply ignored them, too. That is a very bizarre way to handle the data. At a minimum, the 571 teens who were not sexually active should have served as a control group for the sexually active group. How much sexual content did the abstinent teenagers watch? Was it the same amount as the sexually active teenagers? We don’t know, although the authors do know and chose to keep that information to themselves.

In the end, the authors looked at the television viewing habits of those teens who did not drop out of the study, were willing to divulge their sexual status, and claimed that they were sexually active. After eliminating those who refused to divulge whether or not they had been involved in a pregnancy, only 718 teens were left of an initial group of 2003 that had started the study. In other words, the authors only looked at 36% of study participants (all of whom were sexually active) and ignored the other 64% (including everyone who was not sexually active).

By deliberately excluding teens who were not sexually active, the authors severely damaged the credibility of their study. Unless they could show that abstinent teens were much less likely to watch sexual content on television, they cannot claim any relationship between television viewing and teen pregnancy. The fact that the authors deliberately left this data out of their study strongly suggests that it does not show that abstinent teens watch less sexual content on television. Indeed, I would not be surprised to find that the authors set out to investigate a connection between watching sexual content and likelihood of sexual activity, but found that there was no relationship at all. Instead, they were reduced to concocting a spurious relationship between sexual content on television and the likelihood of pregnancy among teens who were already sexually active.

Does watching sexual content on television lead to teen pregnancy? We don’t know, and this study certainly does not tell us.