The link between higher education and smoking rates emerges well before students step onto a college campus.

Though previous research has demonstrated that adults with college degrees are less likely to smoke than those without, it was unclear what contributed to this split. To investigate these mechanisms, study author and Yale sociology professor Vida Maralani analyzed an existing data set tracking a group of youths from adolescence to adulthood. Maralani found that the relationship between education and smoking in later life can be predicted as early as age 12.

“If I take a randomly selected person and I give them more education, what about them will change?” Maralani asked. “The reason you smoke has nothing to do with your education.”

To investigate the link between smoking and education, Maralani analyzed Add Health, a federally funded national longitudinal study of children who were in grades 7-12 in 1994. It tracks the children’s health and risk behaviors over 14 years, and has been used as a data resource for over 10,000 researchers.

The study found that most adult smokers started smoking before age 20, and among daily adult smokers 71 percent began smoking daily by age 18. Maralani concluded that school policies, peer groups and health expectations from teen years are among the characteristics that can serve as early indicators of future smoking.

The study also pinpointed a number of adolescent risk factors for subsequent smoking. Having a best friend who used marijuana in grades seven through nine increases the likelihood of adulthood smoking by 30 percent, while having a parent who has ever smoked increases the odds of smoking in adulthood by 80 percent. Students at schools that ban faculty from smoking on campus have 33 percent lower chance of smoking in adulthood than those attending schools without this regulation.

For Maralani, this large group of risk factors that emerge before higher education disproves a causal relationship between education and smoking.

“People who go on to get a college degree smoke less than people who didn’t finish high school,” Maralani said. “But all the difference happens before these kids are 19.”

Suchitra Krishnan-Sarin, a professor of psychiatry at the Yale School of Medicine who was not involved in the study, said the study’s use of the well-known Add Health database helps validate its conclusions. Yet, as with all longitudinal databases, she said Add Health cannot control for all potentially relevant factors. For instance, the study did not consider genetic factors, which are known to contribute to subjects’ risk for substance abuse.

Krishnan-Sarin said that future research on this topic lies in developing programs that help children avoid the known risk factors for adulthood smoking.

The research shifts the focus from educational checkpoints like high school and college diplomas to factors earlier in development, said Fred Pampel, a professor of sociology at the University of Colorado Boulder who was not involved in the research.

“Should we force a kid to sit in school for another year and hope he doesn’t smoke? That I don’t think is going to work,” Maralani said. “Should we instead think of a school policy at age ten that teaches kids better decision-making skills? That I think might work.”

Beyond decision-making skills, schools should try to teach the skills that families current do, such as self-esteem, optimism and self-control, Maralani said.

According to a 2011 report from the Centers for Disease Control and Prevention, approximately 443,000 Americans die prematurely each year from smoking or the effects of secondhand smoke.