A study by the Institute of Education compared maths test scores of 14-year-olds with local youth conviction rates over three years.

It found higher conviction rates in areas with high proportions of pupils clustered at either end of the performance range.

Researchers said the data also revealed an increase in educational inequality between 1997 and 1999, the years studied. This accompanied an increase in the average local conviction rate for violent crime, from 1.8 to 2.2 per thousand.

Scaled up to a national level, the researchers said this equated to 222 additional juvenile convictions for violent crime and 51 for racially motivated convictions.

Dr Ricardo Sabates, from the Centre for Research on the Wider Benefits of Learning, said: “We do not know why educational inequality appears to be associated with violent crime and racially motivated offences but not with property crime. Nor do we know why there was a rise in educational inequality in the year groups we focused on.

“What we can say is that the increased conviction rates in educationally unequal areas cannot be explained by poverty or by lower average attainment levels.”

Dr Sabates emphasised that the results lent support to the Government’s drive to narrow the attainment gap, but more needed to be done to help those who are at the bottom progress faster than those at the top.

Schools have had to set 2009 targets based on the proportion of children progressing across two levels within four years.

Guidance sent to local authorities says the new progress targets will “help tackle the variance and uneven performance that are a feature of the current system”.

It says all pupils should make at least two levels of progress between Year 2 and Year 6. Previously, targets lead to schools focusing more closely on borderline pupils.

Full report available at www.learningbenefits.com.