Pubdate: Fri, 07 Oct 2005
Source: Wall Street Journal (US)
Copyright: 2005 Dow Jones & Company, Inc.
Contact:  http://www.wsj.com/
Details: http://www.mapinc.org/media/487
Author: Carl Bialik

THE LINK BETWEEN DINNER AND DRUGS

If you missed Family Day last Monday, you may have missed an important
opportunity to prevent your kids from smoking, drinking and using drugs.

Local governments and a national public-service ad campaign headlined
by Jamie Lee Curtis and Barbara Bush told parents to eat dinner with
their children on Sept. 26, based on a study that showed frequent
family dining reduced the risk of substance abuse in kids by 50%.

"There is no more important thing a parent can do" to reduce the risk
of their children using drugs, said Joseph A. Califano, Jr., chairman
and president of Columbia University's National Center on Addiction
and Substance Abuse, or CASA, the group behind the study. The report's
claim was repeated in local media reports in Arizona, Florida and
Indiana, among other places.

But before the government takes its war on drugs to the family
kitchen, a closer look at the study is warranted. The study, based on
a phone survey of 1,000 kids between 12 and 17 years old, and 829
parents, didn't show that family dinners cause a reduction in
substance abuse. Instead, it found that teenagers who dined frequently
with parents scored 50% lower on a substance-abuse risk assessment
than did teenagers who didn't. That alone doesn't prove anything, and
other factors could explain that correlation. What's more, the study
found that other behavior actually had a greater influence on a teen's
decision to use drugs, though those findings didn't fit as easily into
a TV commercial.

Finally, the claim that frequent dinners reduce risk by 50% doesn't
account for age -- a key failing. You might be unsurprised to learn
that 17-year-olds are more likely to use drugs than 12-year-olds.
Older teens are also the ones most likely to eat dinner away from
their families.

Survey Trouble

I sent the study -- which was self-published by CASA, rather than
vetted by a peer-reviewed journal -- to several statisticians, and
they pointed out some drawbacks.

The problems start with the teenagers excluded in the phone survey.
Researchers started with a pool of more than 37,000 phone numbers.
One-third weren't included because there was no answer, a language
barrier or other operational problems; the study suggests that more
than 1,000 of these excluded households probably had eligible teens.
Of the rest, 13,000 households didn't include children of the targeted
age; 9,000 declined to provide information; and about 1,000 interviews
didn't count because they were cut off, or the parent denied
permission. The high level of excluded households may help explain why
black and Latino teenagers were underrepresented in the final sample.

"It is doubtful that the resulting 1,000 completed interviews
represent the U.S. teen population with respect to substance abuse
risk and dinners with the family," Cindy Christiansen, associate
professor of health services at Boston University and a fellow with
the American Statistical Association, told me in an email. Frank
Potter, a senior fellow at Princeton, N.J., research company
Mathematica Policy Research Inc., said the survey had "a very low
response rate."

Steven Wagner, president of QEV Analytics Ltd., a public-opinion
research firm that worked on the survey and analysis for CASA, says
that the study's institutional review board required it to get
approval from both parents and teen respondents, and the need for two
gatekeepers had a "dampening effect" on response rate. He added that
"we have no way of knowing what effect that might have on the overall
results, but we certainly must acknowledge that there is a possible
effect."

Elizabeth Planet, project manager of the CASA study, says the group
will "oversample" black and Latino households next year to ensure that
the final sample is more representative of the general population.

Another concern is the method used to calculate the likelihood a child
would use drugs. Rather than directly correlate family dining with
drug abuse, researchers assigned a risk score to each respondent based
on whether they had used drugs in the past, and their answers to other
questions -- like whether their friends use drugs. Those children who
said they dined with their parents five or more times a week had an
average risk score of 0.79, compared with 1.49 for those who dined
with parents two or fewer times a week. (The overall average risk
score for respondents was set to 1.)

Ms. Planet explains that while researchers did ask if respondents had
used drugs, children may be less likely to respond truthfully to
direct questioning about their own use, so researchers didn't include
it in risk scores.

The risk scores mentioned above weren't adjusted for age, but those
respondents who dine frequently with parents were, on average, younger
than those who don't. I asked QEV's Mr. Wagner to evaluated the
correlation of risk score with age, and also with family dinners.
After adjusting to separate the overlap between the two factors, he
found that age correlated more strongly with risk than did family dinners.

The study also found several other factors that more strongly
corresponded with substance-abuse risk, including how often
respondents watched R-rated movies and whether their friends were
sexually active. (Researchers didn't ask respondents if they were
sexually active, out of ethical concerns raised by an institutional
review board that approved the study.)

"We teach even in [introductory] statistics that when you have two
variables -- like frequency of family dinners and [drug] use in
children -- that look very associated, there may be other things
causing the association," Stanley Wasserman, professor of sociology,
psychology and statistics at Indiana University, told me. "Most
statisticians would be really cautious interpreting a study such as
this one."

Those risk scores also didn't take into account a family's
socioeconomic status. Researchers reasonably concluded that they
couldn't rely on children to report their families' incomes, so they
asked parents. But they only interviewed parents of about 28% of the
teens surveyed. (Based on interviews with parents, researchers
concluded there was little correlation between income and substance
abuse.)

Overstated Benefits

CASA's initial study measuring the relationship between various
factors and substance abuse, published in August, carefully couched
the results, as did a separate, shorter report published last month.

The press release is less restrained, however. In addition to Mr.
Califano's quote, it includes a statement from Larry W. Jones,
president of Family Day's corporate sponsors, TV Land and Nick at
Nite, saying: "The benefits that come from family dining are endless."
(From 6 p.m. to 7 p.m. on Family Day, the TV stations played dinner
music and showed pictures of families dining together.)

The TV spots for Family Day -- viewable from this page -- eschewed
nuance for overly optimistic messages. Jamie Lee Curtis says, "Want to
protect your kids from drugs and alcohol? Pick up a knife. And a fork
and spoon. And sit down to dinner together." And Barbara Bush, in a
separate spot, adds, "We know the more often children have dinner with
their families, the less likely they are to smoke, drink and use
drugs. So simply having dinner together can help your children, forever."

The statisticians I spoke to had a different take. "The report can
conclude that for its 1,000 teen respondents, who probably do not
represent the American teen population, those who ate dinner with
their parent(s) five or more times per week also were less likely to
have a high substance abuse risk as defined by the study," Dr.
Christiansen said. "This doesn't make a good headline, though."

CASA staffers say they have studied this issue for nearly a decade and
have consistently found the same link. "We do believe from our
research that . there is a strong association between family dinners
and substance abuse risk," Ms. Planet says. "And it's a finding that
we've seen consistently over the years."

Why It Matters

Even if there isn't a cause and effect relationship between family
dining and reduced drug use, what's the harm in encouraging kids to
sit down for a meal with their parents? On the surface, not much,
beyond the potential for a dent in fast-food industry revenue. Common
sense, and other studies, suggest that greater parental engagement is
associated with a lower incidence of substance abuse. As I've written
before, many numbers go unchallenged simply because no one has an
incentive to challenge them.

But by trumpeting a quick fix based on flawed numbers, this national
campaign risks papering over the deeper issues behind young people's
use of harmful substances. It's a lot easer for governments to back
Family Day than to reduce availability of drugs to children -- and in
fact, the study found that drug availability was more strongly
associated with drug use than was a child's dining habits.

And in many families, dinners may well be an indicator of familial
relations, rather than a motivator for better relations. "Family
dinners is an excellent surrogate question for how well the family is
functioning," says Mr. Wagner of QEV Analytics. "If as a family you
did everything wrong but you did get together for dinner, it would not
create kids who are safe from substance abuse." 
- ---
MAP posted-by: Matt Elrod