Evidence for peer effects

Danny Finkelstein argues that peer effects matter more than parenting in influencing how children turn out. In fact, there’s a ton of economic research which shows that peer effects are indeed important. For example, this paper shows that American teenagers whose friends take drugs are themselves much more likely to do so. And this shows that crime is strongly influenced by peer effects. These effects are not confined to impressionable youngsters. This paper shows that team-mates of Jose Canseco - a baseball player who took steroids - enjoyed a strange improvement in their performance when they started playing with him, suggesting they learned from his drug-taking.
Not that all peer effects are so bad. Shop workers become more productive when they work alongside more productive colleagues, a finding corroborated by this study (pdf) of fruit-pickers.
And then there’s a ton of evidence for peer effects in schooling. Some of it is summarized in this pdf. Other papers are here, here, here (doc) and here (pdf). The most notorious paper is Roland Fryer's claim (pdf) that negative peer pressure ("acting white") can hold back able black students in school.
You might object here that such peer effects are over-estimated, simply because like attracts like. Maybe teenagers who take drugs choose other friends who take drugs, so what looks like a peer effect is in fact a mundane selection effect. However, many of the papers overcome this problem by using random selection or other controls. And this paper finds that peer effects work even in laboratory settings.
So Danny’s right - peer effects matter.
This is not, however, entirely welcome for traditional conservatives, for three reasons. First, because it complicates the question of moral responsibility; if individuals‘ life outcomes are so heavily influenced by the accidental circumstances of whom they associate with, it‘s difficult to hold them responsible for what they become in later life - as I argued here.
Second, it’s an argument against selection in education - by ability or income, because such selection deprives less able children from poor homes of the opportunity to benefit from positive peer effects.
Thirdly, it’s an argument against banging up young criminals. After all, if peer effects matter, surrounding someone with criminals will create more crime, as this pdf shows.

Comments

[Bishop Hill]

But who one associates with is not a matter of accident, but a matter of choice, by and large. (Unless you are saying that all poor people are bad to associate with).

[Kit]

"because such selection deprives less able children from poor homes of the opportunity to benefit from positive peer effects"

This could be rewritten to defend Grammar Schools:

because non-selection deprives more able children from poor homes...

[chris]

Bishop - many of the papers I cite show that peer effects exist even in randomly assigned groups.
And you are surely plain wrong. It wasn't until I started woking in the City that I even met someone with a degree who worked in the private sector - that's segregation for you. And I don't think black kids in Tottennham fail to associate with Etonians out of choice, do they?

[Tim Worstall]

Children are now, as opposed to five or sex decades ago, much more influenced by their peer group than they used to be. Just the decline of family size would have seen to that: the influence of five or six siblings might be large relative to that of ones peers, or not large, but it would be greater than the influence of one or no such siblings.

Indeed, it's been said that the modern child is raised by its peer group, rather than the family.

[dearieme]

"This is not, however, entirely welcome for traditional conservatives, for three reasons. First, because it complicates the question..." Oh balls; pretty much the distinguishing mark of conservatives vs Lefties is that conservatives know that life is full of complications.

"..it’s an argument against selection in education ... because such selection deprives less able children from poor homes of the opportunity to benefit from positive peer effects": again, balls, because you are assuming that the quality of education is some sort of one-dimensional business, measured from good to bad. Whereas it's a more complex business of what suits whom.

[Recusant]

But of course 'peer effect' can't be the full answer, nor yet the major influence. For if it was, why is it that Joe Canseco, in your example, wasn't influenced by the preponderance of non-steroid takers amongst his peers? Or the criminal who first turns his peers criminal, etc., etc.. Clearly we are also affected by those who have charisma, authority, leadership or any other markers of a strong character.

[Peter]

"Thirdly, it’s an argument against banging up young criminals. After all, if peer effects matter, surrounding someone with criminals will create more crime"

I dunno - this actually sounds like a bloody good argument against allowing young hoodlums to roam the streets, having such a continuing bad influence on their hitherto law-abiding friends!

[john b]

"pretty much the distinguishing mark of conservatives vs Lefties is that conservatives know that life is full of complications"

No, the distinguishing mark of idiots vs non-idiots is that non-idiots know that life is full of complications. Someone who believes that distinction can sensibly be mapped to political axes right vs left definitely falls into the former category.

[Shuggy]

"Children are now, as opposed to five or sex decades ago, much more influenced by their peer group than they used to be. Just the decline of family size would have seen to that: the influence of five or six siblings might be large relative to that of ones peers, or not large, but it would be greater than the influence of one or no such siblings."

Did you have a Freudian moment in the first sentence there, Tim? And do you have any evidence for any of this? I'm not sure myself but I'd be inclined to think the influence of the peer group is in some ways *weaker* than it used to be. Don't you think that parents are *more* involved in their children's socialising and education than they used to be? This is my impression. It certainly seems different from when I were a lad, which admittedly wasn't yesterday - ferrying them in cars everywhere, then picking them up, in contact with them via mobile phones and shit.

"Second, it’s an argument against selection in education - by ability or income, because such selection deprives less able children from poor homes of the opportunity to benefit from positive peer effects."

Agreed. The thing is, unlike childless bloggers, parents understand this 'peer effect' very well - which is why they want segregated education for their children. They want them to do well - but if they can't do this, they want them to avoid getting into trouble - falling in with the 'wrong crowd' or getting bullied. Fair enough but the solution isn't grammars or any of that nonsense - it's dealing with the minority that are allowed, via their apologists in the social work department and 'pastoral care' teachers', to set the agenda in schools and wider civil society.

Clearly more press-ups are the solution here.

[Tim Worstall]

I have all sorts of Freudian moments Shuggy. Which specific one are you referring to?

As to direct evidence about peer groups now rather than then....I'm using peer group here to mean specifically children of the same age. And yes, I do think that smaller family sizes mean that those that children are at school with, in the same class as, have a greater effect upon them than did the larger familial influence of decades ago.

Not a blind bit of proof of this, of course.

[ad]

"it complicates the question of moral responsibility; if individuals‘ life outcomes are so heavily influenced by the accidental circumstances of whom they associate with, it‘s difficult to hold them responsible for what they become in later life"

The same argument can be made about any form of causality. So what has changed?

[Bob B]

"Second, it’s an argument against selection in education - by ability or income, because such selection deprives less able children from poor homes of the opportunity to benefit from positive peer effects."

C'mon. Why the implicit assumption there that peer group pressures are necessarily benign or that benign influences prevail in the competition for influence?

What of the "bad company" syndrome, the gangs and the current fashion for carrying knives or guns?

What of the differentially high drop out rate from training and education of 16 year-olds in Britain compared with almost all other OECD countries?

What of record levels of binge drinking compared with the rest of Europe and the record rates of teenage pregnancies?

What are the life chances of bright, academically inclined boys or girls in great cities - like Bristol or Manchester - where pupils in less than half the secondary schools achieve the government's target of gaining 5 GCSEs at A*-C grades including maths and English?

"Almost a fifth of England's state secondary schools do not yet meet the government's new 'floor target' for GCSE attainment, league tables show. In 638 schools, less than 30% of pupils got five good GCSEs including English and maths, the target for 2012."
http://news.bbc.co.uk/1/hi/education/7179479.stm

[Andrew]

"Second, it’s an argument against selection in education - by ability or income, because such selection deprives less able children from poor homes of the opportunity to benefit from positive peer effects."

Sorry but like other posters, I am unconvinced by this assertion.

1. Have you considered that it won't be the academically bright kids in big city comprehensives creating the peer effects, but there rougher, harder more street wise peers. This at least was my own pupil experience in a big comprehensive, where smart kids tended to pretend not to be academic as a survival mechanism.

2. You are also assuming all children would be selected using the same criteria. What if you allowed a variety of selection? Then you could use the positive peer effects of selecting talented athletes and sports people together and the positive peer effects of selecting more traditionally academically able pupils to study together; or arts and crafts/designers/tradesmen etc. Hey, a bit like the original idea, which was never fully achieved, of the tripartite 1944 Butlers Act. An argument to bring back Grammars maybe...

[Andy Cooke]

"Second, it’s an argument against selection in education - by ability or income, because such selection deprives less able children from poor homes of the opportunity to benefit from positive peer effects."

And contrarywise, the lack of such selection deprives the more able children from poorer homes the opportunity of escaping negative peer effects.

Who's to say that the "positive peer effects" will prove stronger than the "negative peer effects"? In my experience, the geeks didn't tend to win the day on the peer-pressure front. Rather the reverse, actually.

[Shuggy]

"And yes, I do think that smaller family sizes mean that those that children are at school with, in the same class as, have a greater effect upon them than did the larger familial influence of decades ago."

Hmmm - I wouldn't rule this out either. I'd be interested in how this fits with your enthusiasm for triumphant capitalism? All the evidence from economic history points towards smaller family size as a function of greater economic prosperity. Probably don't have to rehearse the reason for you but just in case: a) there are alternative 'entertainments' to having children - i.e. cars, houses, holidays and shit which people today feel the need to consume before they have children, b) greater economic prosperity means children are no longer needed as a cushion to protect against the poverty that may ensue with the on-set of old age or injury, c) the emancipation of women, which I would agree is compatible with capitalist individualism, means smaller families - largely because they tend to be postponed (see first point), d) smaller families are now technologically possible - partly through birth control, partly because prosperity and advances in medical science mean that one no longer has to ensure against the possibility of infant death in the way people did until the 20th century. So what say you?

[Marcin Tustin]

Perhaps young criminals should be sentenced to a term at the local public school (and not allowed to leave during that time). This is likely to be relatively cost effective, if nothing else.