CASA Study Linking Teen Social Media, TV Watching, and Drug/Alcohol Use May Have Been Rigged
August 23, 2011
The National Center on Addiction and Substance Abuse (CASA) at Columbia University has released yet another simplistic, sensational study on youth, this one essentially blaming social media and TV for teenage smoking, drinking, and drug use. The National Survey of American Attitudes on Substance Abuse XVI: Teens and Parents is the latest in a long history of CASA “studies” that damage reasoned drug and alcohol policy by employing blatantly spurious methods to exaggerate teen problems and misattribute them to popular influences.
The study’s conclusion that teens age 12-17 who use social media, view images of drinking and drug use, and watch explicit television shows like “Jersey Shore” and “Skins” are more likely to have tried tobacco, alcohol, and drugs than teens not exposed to such media is typical of a genre of uncontrolled studies that make sensational claims touting media influences without assessing their importance in relation to a host of far more important family, community, and individual influences.
It is possible that CASA’s study is simply invalid on its face. My reading of the lengthy report found no evidence that the authors controlled for age, a fatal flaw in a report that includes ages as diverse as 12 to 17. To have a glimmer of validity, the study would have to compare the behaviors of youths of identical age. That 12 year-olds both would be far less likely to use alcohol, tobacco, and drugs than 17 year-olds, and also much less likely to access social media, would mean that all the survey found its that older teens use both media and substances more than the youngest teens. That is, it found nothing unusual and certainly nothing about media influences.
Our effort to obtain details from CASA on its data base and method was rebuffed. We received the following reply from Lauren Duran, CASA’s Director of Communications: “I am sorry to say that we do not share our raw data. You are welcome to use anything that is in the survey itself in the questionnaires.”
CASA also did not reveal responses by age for the key behaviors it reports, provided no public information on its data analysis, and refused scholarly requests to provide more details. Therefore, I believe the most reasonably conclusion is that CASA crudely rigged this study by employing a clearly biased method to exaggerate the negative influences of social media and television. If so, this subterfuge should cost CASA and its co-surveyors, QEV Analytics and Knowledge Networks, serious credibility.*
The CASA study shows other important reasons why broadening the analysis is vital. Its survey finds that teens are 3 times more likely to use tobacco, 2.5 times more likely to use pot and alcohol, and twice as likely to use prescription drugs (including being 5 times more likely to get them quickly) if parents use these substances. More than half of parents drink, and half of parental drinkers admit they make no effort to control their alcohol use around their kids.
Yet, in hyping social media and TV influences, CASA not only did not control for parental habits as substantiated predictors of teen drug/alcohol use, it does not even include other crucial factors such as living in divorced or separated families, histories of violence and abuse, adult drug and alcohol use within communities, etc. For example, the National Household Survey shows teens in North Dakota are nearly 3 times more likely to drink than those in Utah (paralleling adult patterns).
These powerful patterns, in which teen use of drugs, alcohol, and tobacco closely parallels those of nearby adults directly challenges CASA’s mantra that teenage substance use is the root cause of addiction and other problems. The reality is that teen and adult drug/alcohol issues are closely interrelated, as cultures with healthier attitudes than ours shows.
Can teens’ social media use really explain the wide divergences in teenage drug/alcohol use by family and community? Not likely. The new CASA survey actually shows media influences are quite small. For example, while 70% of teens use social media sites, just 10% of them have ever used tobacco (even once)—that is, 7% of all teens. Even accepting the unfounded notion that social media use caused all of these 7% to smoke, the other 93% are not so influenced. Likewise, even if watching “suggestive teen programming” causes all teenage marijuana use, fewer than 5% of teens would be affected. In many cases, findings are based on responses from a few dozen of the 1,037 teens surveyed.
One indicator that social media are not particularly important influences can be seen in an ignored finding in the CASA survey: “Our analysis showed no significant difference in substance use among teens spending 1 to 30 minutes, 31 to 90 minutes or more than 90 minutes on a social networking site in a typical day.”
If a stimulus is influential in producing a behavior, there nearly always will be a classic dose-response curve. That is, if Facebook use causes drug use, we would expect that a teen who spends 2+ hours on Facebook every day would drink and use drugs significantly more than one who spends 45 minutes a day, who in turn would be more likely to use substances than a teen who spends only 10 minutes a day on Facebook. Yet, CASA found this is not the case. This adds to the suspicion that a multi-factorial analysis would show social media use in and of itself is a trivial factor and that parents should not be stricken by “Facebook fear.”
Some CASA findings are quite startling despite efforts to inflate its findings. For example, the survey asks:
Have you ever had someone write or post mean or embarrassing things about you online, like on Facebook, Myspace or other social networking site?”—a phenomenon we refer to in this report as “cyber bullying.” Almost one in five teens (19 percent) responded, “Yes.” Girls are almost twice as likely as boys to be cyber bullied (25 percent vs. 14 percent).
What? Three-fourths of girls and six in seven boys have never had anyone write or post anything mean or embarrassing about them online ever? That’s a remarkable level of civility! It also renders meaningless CASA’s findings that teens who report being “cyberbullied” are, say, “more than twice as likely to have used alcohol (40 percent versus 17 percent)”—in other words, even if cyberbullying is the only cause of drinking, 11 in 12 teens are not affected by it.
Instead of moderating its findings in the face of nuances, CASA simply presents the singular, strongly implied conclusion that teen use of social media like Facebook and exposure to suggestive programming like Jersey Shore and Skins causes teens to use alcohol, tobacco, and drugs, when in fact media may be a nonexistent or one of the most trivial influences.
Without strict multi-factor controls—especially for age, parental drug/alcohol use, family factors, and individual histories—that are absent from this study, CASA should not even be implying causality. The study could even more justifiably led with the finding that if parents do not want their teens to use tobacco, alcohol, or drugs, the most useful strategy is for parents not to use these substances either–a behavior that is 100% within parents’ ability to control. That is, the key is for parents to control their own behaviors and create healthy family environments, not control their teens’ social and media behaviors.
But that’s a disagreeable conclusion the news media would probably not feature, and CASA went with the sensational headline-grabber instead. The larger problem is that CASA’s survey is not sufficiently controlled to draw the inflammatory, emotionally-stated conclusions it presents. And that fits into CASA’s and director Joseph Califano’s long history of issuing junk “studies” that hype crowd-pleasing malarkey that hampers addressing America’s serious drug and alcohol issues among all ages–to the shame of their Columbia University sponsor.
*Note: standardizing to make the survey sample representative of the demographics of youth nationally, which CASA did do, is not the same as adjusting for age differences in behavior. To assess social media influences, the substance use of social-media-using 12 year-olds must be compared to those of non-social-media-using 12 year-olds of comparable age, gender, and demographics; 17 year-olds with 17 year-olds, and so forth. I can find nothing in CASA’s method to indicate they made such adjustments. Indeed, the comparisons seem founded in crude comparisons of raw rates, an invalid method.