Photo by Lewis Wickes Hine via Shorpy.com
We tend to be very similar to our friends. We have similar interests and lifestyles. We engage in similar patterns of smoking and drinking, or are similar in choosing not to smoke or drink. But why? Did we come together from the start because of our similarities, say, by chatting in the shared smoking area at work? Do we actively encourage each other to drink responsibly, or not? Or are we all just responding similarly to a shared cultural milieu?
Examining data from over 12,000 adults within a densely connected social network, researchers found that if a person smokes, others up to three degrees of separation away from them are more likely to smoke. That is, if you smoke, it is 61% more likely that your friends smoke, 29% more likely that your friends-of-friends smoke, and 11% more likely that your friends-of-friends-of-friends smoke.
But the question of what exactly causes us to be so similar to our friends cannot be answered by simple patterns of association. To help disentangle the causes of these similarities, we turn to a twin study.
One possibility: Gene-environment correlation
A study of over 800 pairs of British teenage twins revealed that the proportion of best friends a twin had who drank was about 28% heritable. (See my previous post for a description of what heritability is and how we estimate it.) The extent to which the twins’ best friends drank was also a fairly good predictor of the twins’ own drinking patterns.
By jointly estimating the heritability of friends’ drinking patterns along with the heritability of the twins’ own drinking patterns, researchers established that this relationship was due to both overlapping genetic and environmental influences. Genetic influences on the twins’ and their friends’ drinking overlapped about 40%, while shared environmental influences (those that tend to make twins similar to one another, regardless of whether they are identical or fraternal) overlapped more than 80%.
The existence of overlapping genetic influences between our own drinking and our friends’ drinking is an example of gene-environment correlation (often labeled “rGE”). Gene-environment correlation occurs when genetic inﬂuences on a trait or behavior make it more likely that an individual is exposed to an environment that is also associated with that trait or behavior.
There are three potential sources of gene-environment correlation, and they can all be illustrated with a single example: You observe a child’s bedroom (let’s call them “Jessie”), and it’s full of books. Why?
- Passive gene-environment correlation: Jessie was raised by his/her biological parents, who are quite smart and well educated. The parents filled the whole house with books. Jessie’s parents have passed on both their genes (intelligence is about 80% heritable in adults and about 45% heritable in children) as well as an environment enriched with learning opportunities.
- Active gene-environment correlation: Jessie enjoys reading, so often requests books as birthday and holiday gifts. Here, his/her own heritable trait (intelligence) has lead Jessie to actively shape the surrounding environment (by asking for more books).
- Evocative gene-environment correlation: Everyone who meets Jessie knows that s/he is a very smart kid. Without asking, many people thought to give Jessie books as birthday gifts (even though Jessie really wanted some video games).
Most of the environments in which we find ourselves are not random. As we develop greater autonomy throughout the teenage years and into adulthood, it becomes increasingly likely that our environment is related to our own personal attributes and choices.
Social influence matters, too
Another powerful way to examine how friends’ behaviors relate to one another is to observe changes in behavior over a long period of time. The above-mentioned study of socially connected smokers and non-smokers showed that not only does smoking tend to aggregate, but so does quitting.
Over more than 30 years of observation, the rates of smoking within the sample dropped dramatically, from more than half of all participants smoking in the early 1970s to only around 20% smoking by the early 2000s. The authors included a neat little animation in their paper, illustrating the pattern of quitting smoking over the course of the study.
Rather than quitting occurring randomly, whole groups or clusters of people tended to quit smoking around the same time. The probability that a person smoked went down by 25% if one of their siblings quit smoking, 36% if a friend quit, and 67% if their spouse quit.
For comparison, a neighbor quitting had no effect. Further, greater geographical distance did not reduce the effect of a sibling/friend/spouse quitting. This suggests that social norms, or processes that involve communication and influence, are an important factor in the decision to quit smoking (rather than being due to strictly location-based influences, like the cost of cigarettes).
Friends aren't random, but they certainly are not genetically “determined” either. Knowledge of reciprocal interaction between our behaviors and our friends makes intervening in the cycle more possible.
If you’re looking to cut down or quit using any substance, having a social group who doesn’t use or uses responsibly won’t necessarily make it easy, but it should make it easier.
Christakis, N., & Fowler, J. (2008). The Collective Dynamics of Smoking in a Large Social Network. New England Journal of Medicine, 358 (21), 2249-2258 DOI: 10.1056/NEJMsa0706154
Fowler T, Shelton K, Lifford K, Rice F, McBride A, Nikolov I, Neale MC, Harold G, Thapar A, & van den Bree MB (2007). Genetic and environmental influences on the relationship between peer alcohol use and own alcohol use in adolescents. Addiction, 102 (6), 894-903 PMID: 17523983
Copyright 2012 Jaime Derringer