Why Lotteries Shouldn’t Harm the Poor

by Harry on March 4, 2007

Two sensible pieces, one by the BBC’s peerless Mike Baker (can we have an education journalist like him in the US, please?) and another by Fiona Millar on school places lottery flap. A very peculiar piece at the BBC site reporting on research that, as far as I can tell from the report, has nothing to do with the lotteries. The researcher is quoted as saying that:

Our research suggests that lotteries of over-subscribed school places would produce the worst of both worlds – greater educational polarisation and longer, more environmentally damaging car journeys to distant schools by middle-class parents. She said it was interesting that Labour-controlled Brighton was proposing it on the grounds of fairness and equality of opportunity, when this research suggested it might have exactly the opposite result.

I gather that this claim has gotten some play in the debate, so it’s worth refuting it. (As a bonus for my New Labour friends I include a criticism of David Willetts below the fold).

The research seems to replicate (using a tiny sample) the common and unremarkable finding that poor parents use quite different strategies in choosing schools from middle and upper class parents (no-one, I think, denies this). But the use of the lottery, in the marginal way that Brighton and Hove have implemented, has no bearing on who chooses what school. We can presume that middle class and poor parents will keep the same strategies as before. What the lottery does is just makes it harder for the most popular schools to choose the more popular students. In a pamphlet I wrote about this for the Social Market Foundation (Pdf) some years ago I fretted about the fact that lotteries give schools incentives to skew their applicant pool, which is true. But this isn’t the argument the researcher is making, and anyway the pre-existing choice system already contains those incentives, and I don’t see them being made worse by the introduction of lotteries. So I can’t see any reason to think that the lotteries will have the “opposite effect” from making things more fair and equal.

While I’m on the topic, perhaps it might be worth taking David two-brains Willetts to task. Here is Willetts on the Brighton lottery:

The root of this problem is that Labour has got bogged down in an obsession with how to allocate a fixed number of places in good schools, when they should be focusing on raising standards in all schools.

Now, I’m an admirer of Willetts, and although I realise that his IQ disqualifies him from the leadership of the Tory party I think he might be a good Education Secretary. But this seems to signal an odd conversion for a Tory. The theory behind school choice is that it deploys the tacit information that only parents have about schools, and harnesses their motivation to the benfit of all schools, by forcing worse schools to respond to the threat of exit. Lotteries mitigate a pretty awful market imperfection within the current system, which is that schools (producers) have the power to choose whom to serve. Lotteries make the school system more market-like, by reducing the power of the producer relative to the consumer, and should help to raise standards in all schools if the theory works. Me, I’m pretty sympathetic to the idea that choice might have some benefits in this context, and certainly think that if you are going to use market mechanisms you should make them less rather than more imperfect. Perhaps the left-ward shift of the Tory frontbench is more sincere and sharp than I had believed.



nick s 03.05.07 at 8:11 am

one by the BBC’s peerless Mike Baker

He’s come a long way since covering local stuff for Look North. Then again, Lara Logan worked for GMTV.

I still think George Walden’s central thesis on the British education system is irrefutable: that reform is hamstrung by the decision-makers on policy being precisely the kind of middle-class parents who’ll take advantage of the private system, move house to a better catchment area, or discover religious devotion and send the grant-maintained faith school across town.

Whats he doing now, by the way?


chris armstrong 03.05.07 at 11:24 am

It was interesting to read Mike Baker’s piece, if only to find out more of the details, which were entirely lost in the outraged media fanfare. Essentially most of the selection criteria remain the same, and selection is still limited to those in catchment, with the single change that the factor ‘closeness-to-school’ has been replaced by ‘random selection’ (still within catchment, and still balanced against other factors such as siblings’ school location, exceptional circumstances etc).

I’d have to agree wholeheartedly with this change, because ‘closeness-to-school’ is a factor that is by definition manipulable by parents, and more importantly, more easily manipulable by wealthier parents. A defensible choice agenda has to pay SOME attention to the ease and difficulty with which people can make the relevant choices, otherwise the resulting distribution looks likely be inegalitarian, right?

I suspect that the response (quoted by Harry) that this policy change will increase carbon emissions is disingenuous. I guess it shouldn’t be ignored, but it looks like an argument for school buses rather than for preserving the status quo.


Alex 03.05.07 at 11:35 am

Willetts is being twattishly stupid here. Gains from selection are zero-sum – collecting the kids with the best life chances in school X will certainly improve it, but only as far as collecting the kids with the worst who’ve been crowded out in school Y worsens it.

If you improve “all the schools” (and hey, what about a pony whilst we’re at it?) but don’t tackle selection, you’ll at best get a highly unequal distribution around a rather higher median value.

Further, I suspect education, like everything else, has diminishing returns, so putting someone in a worse school does them more harm than putting someone in a better school does them good.


soru 03.05.07 at 12:10 pm

You know, I think you could fix the issue with a certain type of middle-class person having disproportionate electoral influence (the root cause of New Labour) by assigning voters to constitutencies by some kind of similar random method.


Alex 03.05.07 at 1:02 pm

simpler version of 4: move the constituencies, not the voters.

Or use proportional representation.


Luis Alegria 03.05.07 at 7:11 pm

Mr. Harry,

This is a pretty universal situation. We have much the same sorts of issues in the US, including, in some places, such lottery systems. They are all subject to gaming.

The important point being missed is on the nature of a “good school”. As far as I can tell, this is not a condition that exists independently of its intake – it is rather defined entirely by its intake – a school is its students, more than anything else.

There is no question that some (measurable but fairly marginal) improvement in educational outcomes for students with relatively poor prospects can be achieved by placing them in “good schools”, defined as those with a critical mass of positive peer groups. There is a great deal of empirical reasearch backing this, from Coleman to Hoxby and on.

If one accepts that as a starting point, one can move on to matters of engineering and asset-allocation, where the assets in limited supply are the “good students”. One needs to arrange to concentrate these in such a way as to provide the maximum number of schools with a critical mass.

There is probably a lot of research still needed to pin down the parameters of such a critical mass, but in US ethnic terms, as in the US race and ethnicity are everything, this seems to be @60%-80% white+Asian (the US type of Asian), to which one can assign a minority of the remainder. Diluting the “good students” further seems to remove their value as peers.

Also notably, I have seen research where the value of such peer effects are greatest among the poor students (black mainly) with the greatest potential. Lacking a positive peer environment – being in a majority-black school for instance – seems to affect them the most. Black educational achievement in majority-black schools is truncated at the right side of the curve.

So the best bang for the buck – making the most of the limited resource of “good students” – seems to call for a strategy of concentrating these in a two-tier system to generate the maximum number of schools with a positive critical mass, and mixing in the most likely educational prospects out of the remainder.

There is still the problem of what to do with the rest. All I can say is that there seems to be only so much that the peer effect can do. It helps, and helps much more than increased resources, but it is not in itself a cure for the educational gaps.


SamChevre 03.06.07 at 3:34 pm

Hear, Hear!

Luis Alegria makes the point I wanted to, better than I could. Good students are to a significant extent producers, not just consumers, of good schools. One further detail is that really undesirable students are extremely disruptive to eductation. (19 good students, and one student who is severely disruptive, can produce a classroom where no learning takes place pretty easily.)


harry b 03.06.07 at 6:15 pm

As you know, samchevre, I agree with that. But it is orthogonal to the point here, UNLESS people think of choice or lotteries or whatever as some sort of magic bullet.

Though I disagree with what luis says about “in the US race and ethnicity are everything”. Not according to the evidence. Class is nearly everything, and race a little. When you look at those black students carefully, it turns out that they have remarkably positive attitudes to school, and that their class background is less affluent than you might have thought..


SamChevre 03.06.07 at 7:32 pm


I knew you agreed, but thought others might not.

My thought is that replacing selection with lotteries is intended to reduce schools’ ability to choose the more desirable students. It seems to me that this is likely to reduce the number of good schools.


Luis Alegria 03.06.07 at 7:54 pm

Mr. Harryb,

I beg to differ. In the matter of school assignment, selection, choice, lotteries, related research and that whole kettle of fish, in the US it is indeed all about race. Everything in this area from Colemans 1960’s research on was based on addressing the problem of racial integration. If there was no racial issue there wouldn’t be 1/10th of the effort put into these things.

Even todays continuing turmoil over school assignment is entirely racial. In the San Francisco School District for instance we have seen about five school assignment schemes over the last twenty years, all of them in response to civil rights lawsuits and court decisions of one sort or another in every direction based on the flux of response and reaction to anti-discrimination laws. And this is just a continuation of the same turmoil thats been going on since @1967-68. This is typical of whats been happening in big-city school districts over that period.

As for class as such, this is not a strong driver in the US with respect to public perception and preference. White people do not typically walk away from mostly-white public schools on class issues. The biggest providers of private education in the US are the Catholic church through its parochial schools and the various Evangelical churches through theirs, and these have no class-cachet, often the opposite. The premium private schools in the US have a very limited following.

As for class and black school performance, that is another matter. If you are interested in pursuing this very interesting topic, I highly recommend John Ogbu’s “Black American Students in an Affluent Suburb: A Study of Academic Disengagement”.

As for black students attitudes, surveyed attitudes seem to be of little value in predicting behavior (see Ogbu), and of course these do not match measured performance. Actual behavior observed through field research does not square with such surveys.

As for magic bullets, of course there are none. Interestingly, the outcome of all schemes, after some turmoil (often a lot of turmoil), seems to settle into something approaching what I have described as parents adjust to the changed rules, often by simply moving elsewhere, but often also by learning to game the system. School choice, where available, seems to allow the sorting process to work with the minimum sturm und drang.


harry b 03.06.07 at 8:53 pm

Well, I have looked pretty carefully at the black-white test score gap literature, and it is pretty clear (Ogbu’s ethnography notwithstanding) that socioeconomic status explains most of the gap, given the data we have.

As for race being the whole explanation of all the fuss, this is sort of true, but that is because there is no countervailing pressure to socioeconomic segregation. There is no class equivalent of Brown, and no class-based social movement equivalent to the civl rights movement. The only poor kids who wealthy white parents are forced to have their children in schools with are non-white kids, because it is only those that the courts do anything to try and integrate. If we tried to integrate by socioeconomic class that would be where the fuss would be.

Diocesan Catholic schools, btw, which are indeed far more socioeconomically integrated than public schools, have all sorts of advantages in this respect over public schools like the religioous commitment to a certain vision of education not only from the schools but also many parents. They also can, and do, get rid of the most disruptive kids (of whatever social class) and thus are not as vulnerable to the problem that samchevre raises (and everyone knows this, so that plays a big assurance role).

samchevre — I get it, thanks — I didn’t understand from previous comment. Whether you are right depends on very hard to make calculations (eg, my guess, though its impossible to be precise about this, is that the move from one to two disrupters is an exponential decline, but that at a certain threshold the next one doesn’t make things worse at all!). SOmeone must be able to model these things but it ain’t me (as you know!)

I doubt the effect is negative in this context given this (very small) rule change, but I can’t say that with any certainty. (I’ll be more confident after I ask my dad! if that doesn’t sound too feeble….)


Luis Alegria 03.06.07 at 9:45 pm

Mr. Harryb,

Definitely off-topic, but again I disagree about the value of race as a predictive variable in educational performance in the US. The consensus I know of is precisely the opposite to what you perceive.

It is clear, for instance, that a pure socioeconomic preference scheme for college admissions would very disadvantageous towards blacks and hispanics, and would disproportionately benefit poor whites and, especially, Asians.

This was also seen in the SFUSD when such schemes were tried as part of the admissions processes for public elementary and high schools.

“there is no countervailing pressure to socioeconomic segregation”

Precisely my point – there is no such pressure. It is not a pressing political or social issue in the US.

“The only poor kids who wealthy white parents are forced to have their children in schools with are non-white kids”

That is not correct. US high end schools are full of “poor kids” that got in purely on their merits, and this is a very old tradition to boot. High-end schools are full of poor, upwardly mobile immigrant kids with outstanding academic abilities. In the past these were ethnic whites and especially Jews, these days they are Asians.


harry b 03.06.07 at 10:08 pm

luis — well we are off-topic. I know there is a consensus of sorts on this, but it flies in the face of the evidence. Look at the Jencks/Phillips book, especially the introduction. We might have to agree to disagree about this, but I think there is an emerging counter consensus among statisticians and economists.

I think, too, that “US high end schools are full of “poor kids” that got in purely on their merits” is an exaggeration. Most high end private schools also go to great lengths to recruit children of colour, too, btw. And most white parents of kids in those schools are happy with that as long as they are the right kind of child of colour.

Maybe we’re talking past each other about race and class being the cause of all the fuss. Sure, because race is the only trigger for integration that is where the action is. And I know that some white middle class folk really really don’t want their kids around black kids. BUt most don’t want their kids around poor kids, black or white, and the point is that they are able to act on this preference without anything getting in the way, so there are no flashpoints around it.


harry b 03.06.07 at 11:01 pm

I should add that when I said “maybe we are talking pst each other” that was meant not as a veiled criticism of you but a (well) veiled self-criticism. I was making the point that class would be a big issue in the politics around the allocation of kids to schools if there were any pressure to integrate, but of course you’re right that there’s none so it is not an issue in that politics.


Luis Alegria 03.07.07 at 12:54 am

Mr. Brighouse,

I will look for the Jencks-Phillips book, thank you for the suggestion (is this the one you mean? http://www.amazon.com/Black-White-Test-Score-Christopher-Jencks/dp/0815746091)

But if the basis of the book are the ideas expressed here –


It seems to me that Mr. Jencks understands the situation exactly as I do –

“The number of affluent black parents has grown substantially since the 1960s, but their children’s test scores still lag far behind those of white children from equally affluent families. Income inequality between blacks and whites appears to play some role in the test score gap, but it is quite small. ”

In fact I have no complaint at all about this very fine article.

As for the acceptable sort of “child of color”, from my experience that child is usually the kind that gets in on merit. The high-end schools I am thinking of are the academic public high schools like Lowell in SF and Bronx Science/Stuyvesant in NY, as well as the best of the suburban districts like Palo Alto, in all of which the proportion of Asian students is very large and growing, and often already a majority.


harry b 03.07.07 at 1:15 am

I’ll read the propsect version when I have time. If you dig, it turns out that when you control for parental wealth you get very different results than when you control for income, and when you control for grandparental wealth the gap gets very small indeed. This is clear in the long intro to the book. Further studies confirm this, and close the gap even more (I’ll get the cites later if you liek — maybe its worth a whole post).

Not quite sure what you mean by merit. I assume you mean something like “good scores and the probablistic judgment that the child will not be disruptive”. Sure, no middle class person minds their kids being around kids of color, or poor kids, like that. Even black kids. That’s acceptable. As long as we can select them away from all the unacceptable poor and working class kids, as the schools you mention do. So the class bias is, indeed, overridable by the right sort of kid.


Katherine 03.07.07 at 5:24 pm

The very definition of “good school” continues to be a problem in the UK because of the blunt instrument of league tables. They tell you output, without bothering to tell you anything about input.

At the risk of getting all anecdotal here, I went to a secondary school that was, on a league table basis (although they didn’t quite exist at the time) dreadful. Nearly the worst in the county. I did very well, as did a few of my peers. Overall, I’d say that the school, given the socioeconomic status of its intake, didn’t do a bad job at all. But it is labelled now as a bad school.

What is always missing from the definition of “good school” as it currently stands is any kind of added-value assessment – ie how does a school do, relative to what you’d expect from its intake.


harry b 03.07.07 at 5:53 pm

Katherine — my guess is that I went to school a fair while before you (and before league tables etc) but I know that the school I went to was the last choice for anyone in that city, and my experience was, well obviously, rather like yours. (The school is now in special measures, and unlikely to survive I suspect).

Don’t underestimate how incredibly hard it is to come up with meaningful value-added information though — there are some powerful technical barriers to be overcome.

All in all what that means is that “good” school is very very hard to define, I agree.


bzonkers 03.07.07 at 5:56 pm

As for American education journalists, I like Joanne Jacobs, Samuel Freedman, and this guy.

Comments on this entry are closed.