Andrew Oswald: Wanna do well? Find a smart room-mate

The major advantage of going to a top university is the high standard of its teaching and research, isn't it? Um, no – think again. It is the psychological pressure induced by keeping up with your dazzling housemates. Peer pressure propels.

Fascinating new evidence from Bruce Sacerdote of Dartmouth College in the US proves that the quality of your student contemporaries is crucial to how you do at university. The subject is hard to study convincingly. Clever people tend to go to elite universities, so it is tough for a researcher to sort out whether their good results are merely because they had to be talented to get in. What you need, ideally, is an experiment in which some youngsters are randomly assigned to places at top universities, while others in a control group get sent elsewhere. But it is not possible to do that in a free society.

Sacerdote solves this simply and brilliantly. He notices that at Dartmouth, one of the Ivy League universities, all incoming students are randomly allocated to halls of residence (dorms, in American jargon). They share a bedroom with someone of their own sex, but don't choose who. These students are randomly matched and must live together for the whole of the first year. So we have a "natural experiment".

It turns out that how successful your room-mate is has a noticeable effect on your own academic performance in the first year. If you are paired with someone who does well, you tend to do well. If you are paired with a dud, you tend to become a dud. Because the accommodation service at Dartmouth draws room partners out of a hat, this is good scientific evidence for what might be called peer-pressure effects. The size of the sample, at 1,600 students, is reasonable, and the effect discovered by Sacerdote is a strong one statistically.

How extensive are these effects? Could a university, or a country for that matter, use this discovery to become more efficient as a producer of education?

Your room-mate's talent has quite serious consequences for you. If they start coming near to top of the class in the first-year exams, then, compared to having a room-mate who is average, you tend to go up (in British jargon) from a lower second to an upper second. This effect is probably a mixture of getting help from your talented friend, and having to compete with them.

A natural question to ask is whether social and personal behaviour also end up being moulded by the room-mate. Interestingly, it mostly does not. But which societies are joined, the author shows, do depend on the room-mate. How much beer you drink in the first year turns out not to be correlated with your enforced friend's, though it is affected by how much drinking there is in the entire hall of residence. Participation in athletic pursuits is not influenced in a statistically significant way by whether you live with a top athlete. Nor does your final choice of degree subject – Americans choose after their first year – turn out to be affected by your room-mate.

One finding by Sacerdote is that the size of peer effects depends on the exact mix of talents of the two individuals. If you are assigned a middle quality room-mate, you do no better than if you are assigned a bottom quality room-mate. Furthermore, middle students are not hurt or helped much by the quality of their room-mates. Ordinariness, it seems, is hard to overturn.

Yet being allocated someone who gets first class or very good upper-second marks helps two sorts of students. One type is those who enter with top grades from high school. They are more likely to get firsts in their early exams if they are made to live with someone who gets firsts. The second type are people who only just squeaked on to the course in the first place. The firsts pull up the potential thirds.

So the strange bottom line is this; get your housing administrators to drop first-class people into apartments with only very good or very bad students. The relaxed apartments where everyone is scraping an upper second cannot be galvanised into action. They probably just want to be friends.

The writer is Professor of Economics at Warwick University

education@independent.co.uk