Obesity is defined usually as a body-mass index of 30 or greater, where the body-mass index adjusts weight for height by dividing weight in kilograms by the square of height in meters. Obesity defined in any reasonable way has increased rapidly since 1980 in the United States, and to a lesser extent In Europe and Japan. About 30 per cent of all Americans have a body mass index of over 30, and about twice that number are considered to be merely overweight. While all economic levels, ages, racial groups, and both genders became heavier, obesity is much more common among the less educated, lower income persons, women, African-Americans, and Hispanics.
Several factors explain why the average weight of Americans (and those in other developed countries) increased a lot more rapidly after 1980 than it had before. The effective price of fatty foods began to decline rapidly at that time, in part due to the growth of fast food chains, like McDonald's. Also important, especially for teenagers, is the attraction of sedentary activities resulting from computers, email and instant messaging, and video games that replaced time at sports and other more physically challenging activities. (Television viewing by Americans did not increase, and may have declined, during the past 25 years.) The development of many drugs that combat high blood pressure, high cholesterol levels, and other ill health caused by being greatly overweight, and a reasonable expectation of further medical progress in the future, also contributed to the declining concern about being greatly overweight.
Social networks that influence eating and leisure activities has been recently suggested as a further factor in the spread of obesity. An article in the July 26 issue of The New England Journal of Medicine ("The Spread of Obesity in a Large Social Network Over 32 Years") analyzed the famous Framingham Heart Study for any evidence of social influences on obesity. This Study has followed about 5000 individuals and their children and grandchildren since 1948, with questionnaires on health, weight, friends, marital status, and many other variables. The recent exploration of these data between 1971-2003 for obesity social influences finds that a person becomes fatter when his or her friends, spouse, or siblings become fatter. The authors are aware that this does not necessarily mean causation from weight gain by friends and family to weight gain by this person. They probe further by adjusting for past weight, by looking at the timing of weight increases, by seeing if weight changes of neighbors are correlated (they are not), and develop a few other tests.
The results withstand all these tests, so they conclude that social influences of friends and family are important in the obesity "epidemic" of the past 25 years. Yet it is impossible with data of the kind in the Framingham Study to be sure that social influences rather than common changes in variables unobserved by the analyst explain why weight changes move together among friends or other social groups. To isolate the effects on behavior of group influences rather than the effects of changes in common forces, one needs evidence like what happened to their weights after college students are randomly assigned to each other as roommates (there is a study of the behavior of students who were so assigned), or evidence from other situations where one can more fully rule out friendships due to common interests and backgrounds. Still, it is plausible that eating habits and leisure and work activities are significantly influenced by what peers and family are doing. Indeed, considerable circumstantial evidence strongly suggests powerful social influences over many kinds of human behavior.
Nevertheless, however important is the impact of social influences on weight gains, they cannot explain why the trend toward obesity accelerated during the past several decades. For what social influences do is magnify, not start, the responses of groups to changes in prices, incomes, jobs, technologies available for games and communication, and other factors common to most members. For example, if a friend starts eating fast foods more frequently for whatever reasons, and I am influenced by his eating habits, then I would more frequently eat more fast foods as well. If he is also influenced by my eating habits, he eats even more of these foods than he did initially. But then I would eat still more as a result of his reaction to my eating. The end result could be a large increase in the fast food consumed by both of us, which could cause significant increases in our weights.
In effect, there is a multiplier response to any initial weight gains due to the presence of social interactions. To illustrate this "social multiplier" on weight changes, assume that if each time my peers gain a pound of weight, I gain 0.6 of a pound because I alter my behavior to correspond to theirs. If I influence their behavior to the same extent, then the social multiplier would be 2.5=1/1-0.6, which is the infinite sum of the series of back and forth interactions on weight changes among these peers. This means that if each member of a group on their own would gain 10 pounds as a result of say a decline in the price of fat, than each would end up gaining much more, 25 pounds, because of the social interactions among them.
This example shows how a sizable social multiplier on weight would transform moderate gains in weight when individuals are considered in isolation into possible epidemics after accounting for the interactions among members of a social network. The social multiplier on weight gains can be utilized to make it easier to lose weight. Individuals unhappy with being overweight can recognize that their chances of losing weight are greater if they acquire friends who are thin or who want to lose weight too. They could try to find such friends by attending exercise classes, or by joining online and other weight losing groups.
A company may want employees to lose weight in order to reduce absenteeism, and to save money on employee health insurance. It would be able to utilize the fact that employees' beliefs about their appropriate weights are influenced by how heavy their fellow employees are, especially when employees of the same firm are also friends. The company could heavily subsidize healthy meals, or offer bonuses to employees who lose weight, recognizing that these benefits might cause big reductions in weight because of the social multiplier.
Governments may try to reduce the incidence of obesity because of paternalism, or a desire to reduce public spending on ill health (I am skeptical about the value of such policies-see my blog post on October 8, 2006). It could take advantage of the social multiplier on weight by recognizing that even modest taxes on the fat in food or on fast food outlets would have large effects on the consumption of fat, and indirectly on weight loss, because of the social multiplier in eating patterns among friends, family, and others.