UK study: League tables help predict children's success

Parents who choose secondary schools on the basis of league tables will ensure their children do better in their GCSEs than those who choose schools at random, according to a new study.

Academics from the University of Bristol's Centre for Market and Public Organisation (CMPO) working with colleagues from London’s Institute of Education (IOE), studied the GCSE results of half a million children who had to choose their school in 2003 and took their exams in 2009. They compared each child with similar local children who chose a different school.

“Parents should use GCSE performance information to choose schools,” conclude Dr Rebecca Allen of the IOE and Professor Simon Burgess of CMPO. “We find that using performance tables is better than choosing a local school at random.”

The findings will surprise sceptics who argue that raw exam league tables simply reflect the make-up of a school’s population and may not be useful in predicting a child’s success. For example, a school with good results might simply attract high ability pupils. 

Perhaps surprisingly, raw outcome performance tables outperform more sophisticated tables, such as those which attempt to show the “value added” to children’s scores by the school.

The findings are detailed in a CMPO paper entitled “Evaluating the provision of school performance information for school choice”.

Allen and Burgess found that league tables are most useful for students who have to choose among schools with very different levels of performance. When differences between local schools are minimal, league tables are not particularly helpful in predicting a child’s future academic performance, they say.

“Another surprise is that the best GCSE performance information is only slightly more useful in school choice than knowing the average ability of pupils entering the school,” say the authors. “We believe that this is because the demographic profile of pupils strongly influences the school’s ability to attract high quality teachers, headteachers, governing bodies, unpaid volunteers, teaching assistants, and other resources. 

“To be clear, our argument is not that school composition is all that matters directly and teaching quality not at all; rather, we argue that teaching quality matters a great deal, but that averaged over a number of years, this is strongly influenced by school composition.”

The authors conclude that the situation could be improved through policies which work harder to equalise school intakes, or which enable deprived schools to attract more resources.  Local authorities should have a mandate to publish exam performance data alongside admissions information in the school admissions brochures sent to parents of 10-year-old children, they say. This would improve the chances that more disadvantaged families use this performance information, and would make no difference to the choices of advantaged families who already incorporate this information into their decisions.

However, poor families will not benefit without reforms to the school admissions system “so that students from these disadvantaged families can actually access the schools that they might choose on the basis of the performance data”, Allen and Burgess conclude.