This research has been knocking around for years. Everywhere you go, at education and crime conferences, you will hear it quoted. It knocks the breath out of people who have never seen it before. It knocks the socks off most other causes-of-crime research because it is such a thorough piece of sociology.
The High/Scope Perry Preschool Study took a group of three- and four- year-olds from the poorest, most crime- and drug-ridden neighbourhoods in Michigan. It divided them in two. One lot got two years' intensive nursery education, the other didn't. Researchers then tracked all of the original group through the rest of their lives. (They are now nearly 40.) The difference between those who had the two years' special nursery education and the group that didn't is phenomenal.
The project started in 1962 in Ypsilanti, Michigan. The latest check when the group was 27 found these results: the High/Scope children have half as many criminal arrests as the control group; they earn far more; nearly three times more own their own homes, marry and stay married longer; 20 per cent fewer have ever been on social security. They will be traced and monitored again when they reach 40 - if the project's founder, David Weikart, can raise the funds, because all along he has struggled to get the money to keep monitoring these children. Everyone likes the results, no one wants to pay. What's new? Sociology is a pitifully poor relation of, say, health research and yet we want to know as many social answers.
David Weikart was in London this week visiting a British version of the High/Scope project, largely funded by Barnados and National Children's Homes, who use the methods for their nursery schemes in deprived areas. Of course we are suspicious of American gurus pedalling magic systems. But Weikart is a modest and moderate man who eschews jargon. An educational psychologist by training, he devised the High/Scope approach for children who have had too little attention at home. He says his method makes little long-term, measurable difference to children with good parents.
Starting from the premise that these children come from chaotic, unpredictable households, High/Scope teaches them how to think, analyse and structure their time - "Plan, Do, Review". Weikart is fighting a rear-guard action against the current reactionary fashion for more Chalk and Talk, more Sit Up and Shut Up teaching. What they learn hardly matters at all, he says. It does not last and it makes little impact on their future. Learning to think about what they are doing is what makes the difference to the rest of their lives.
Weikart's studies show that children who have a heavily academic nursery education emphasising the three Rs do far less well in the long run than those who have been taught ways of thinking rather than content. (Though any nursery schooling helps.) The trick is to make each child plan, think about and describe each activity they choose to do. The success of the scheme also depends on bringing parents into the project, with at least 20 home visits a year.
Weikart is scathing about the importance of IQ tests. High/Scope improves the children's IQ score by some 25 points, as they enter primary school. But by the age of 10, the High/Scopers and those with no nursery education all have the same average scores. Yet the High/Scopers go on to do spectacularly better. Whatever IQ is, he does not find it a useful predictor of the qualities needed for social stability in later life.
So for several decades we have had these results to mull over. What effect has it had? Weikart smiles wryly. Intensive nursery education is still only patchily provided. "But Michigan is building two new prisons every year instead, and has plans to keep building them indefinitely." Prison building is eating up all other budgets progressively, as it will in Britain, even though every dollar spent on High/Scope children saved $7 later in their lives on crime and welfare. In Michigan, which funded the initial programme, 3,000 children are in High/Scope when, he estimates, some 25,000 severely deprived children a year are in need of it.
Governments have no idea of economic planning when it comes to social projects, Weikart says. "If a company wants to build a new hotel, they don't wait until they have saved that money. They go out and borrow it, knowing they will make profits to pay the money back. The same is true of nurseries and other preventative schemes. Borrow to invest now and reap the profits later. But they refuse to think that way."
Social problems are never treated like health problems. If some new medical cure arrives, even at great expense for a small number of sufferers, the NHS gives in to the clamour to provide it. But when the solution is not medical but social, the policy-makers simply ignore it. That is partly because there is rarely anything like the same social evidence expensively gathered to bolster arguments for new medical treatments. Most good social schemes only just manage to scrape together funds for the project itself, living hand to mouth from year to year, with no extra money for long-term monitoring of results.
But if we were to redefine crime as a disease, we would think about it in a more constructive, problem-solving way. Imagine if mighty Royal Colleges with highly paid and distinguished consultants were in charge of curing various social sicknesses. Imagine a whole great establishment devoted to rooting out the causes of crime, researching the epidemiology and the cost effectiveness of various treatments with all the grandeur, status and funding we give physical illness - then we might make some progress.
Since people are as worried about crime as they are about health, a gigantic crime-busting social package should be Labour's priority for its manifesto. And the Tories, too, for that matter. Politicians have made nursery pledges in the past but not focusing on the need for expensive, intensive programmes for those children who are most likely to cost society dear in the future. But to do that would take borrowing or taxing now to invest wisely for long-term savings.