Can we really ‘catch’ healthy behaviors?

Research suggesting that certain behaviors affecting health are “contagious”—that people who quit smoking are likely to influence friends to do the same, or that happiness spreads from one person to another—has generated a great deal of media buzz.

But the statistical techniques used to establish these so-called network effects are prone to “large biases that might produce effects where none exist,” says Jason M. Fletcher, Ph.D., M.S., assistant professor of public health.

Fletcher and a colleague studied headaches, acne and height, conditions for which network effects are implausible. In the December 4 issue of the British Medical Journal, the team reports “surprisingly high” network effects for these conditions that disappeared with statistical refinements.

“Our results suggest caution in attributing similarities of health outcomes between friends to a ‘contagious’ process,” Fletcher explains.