3 ms·

> This is subject to some major sampling bias. Drinkers cluster with drinkers. Non-drinkers with non-drinkers. Or they don't, and virtually in every community

• retrac 4y ago
  It's certainly no hard binary! But drinking behaviours are socially determined, so unsurprisingly it does tend to cluster: > People were 50 percent more likely to be heavy drinkers if one of their friends or relatives drank heavily. However, if a friend-of-a-friend drank heavily, people were only 36 percent more likely to do so, and if a friend-of-a-friend-of-a-friend drank heavily, they were 15 percent more likely to do so. > The pattern worked in reverse, too, although it wasn't as pronounced. People with a friend or relative who didn't drink were 29 percent more likely to be teetotalers themselves. > The quantity of relationships mattered as well. The more connections to drinkers--or nondrinkers--a person had, the more powerful the effect was. People "surrounded by" heavy drinkers were 70 percent more likely to drink heavily themselves, while those with many abstinent friends and relatives were 50 percent more likely to also abstain, according to the study. http://www.cnn.com/2010/HEALTH/04/06/drinking.alcohol.family.friends/index.html http://www.cnn.com/2010/HEALTH/04/06/drinking.alcohol.family... Other studies mostly point in the same direction.