By Gary Eason
Education editor, BBC News website
The secondary school league tables for England include a new measure this year: "contextual value added". What is it and how does it work?
The new measure takes in many factors
Value added has been with us for some time - comparing each pupil's results with those of all pupils nationally with similar attainment when they left primary school.
The usual outcome has been ... tables dominated by independent and grammar schools.
Critics pointed out that the measure made no allowance for the very different circumstances in which schools operate, and over which they have no control.
Thus contextual value added (CVA) was conceived.
This factors into the progress measure, nine overlapping elements or "coefficients":
What CVA does is predict what a given child's attainment should be based on the attainment of other children with similar prior attainment and similar backgrounds.
- Special Educational Needs
- Eligibility for Free School Meals
- First Language
- In Care
- IDACI (a postcode-based deprivation measure)
The idea is in effect to create a level playing field so that how they actually performed - better or worse than the others - can be attributed to the school's influence.
It is important to grasp that this is entirely a statistical, mechanistic model. It is not as if someone in Whitehall sits down and decides what weighting to give to the results of, say, a Bangladeshi girl with behavioural difficulties who gets free school meals.
Instead, the formula looks at the actual prior attainment and current attainment of all pupils with such characteristics.
This gives a predicted outcome for any individual pupil.
Their CVA score is the difference between that predicted figure and their actual results.
To get the score for the whole school, all its pupils' CVA scores are averaged then a so-called "shrinkage factor" is applied, related to the number of pupils.
The final outcome is a number based around 1000.
The effects of the various coefficients can be, officials point out, counter-intuitive. Remember they are measures of progress.
For example, you might expect older pupils to attain better GCSE results, but the predictions show that younger pupils make more progress.
This makes sense because younger pupils tend to have been further behind at Key Stage 2 and close the gap as they move up through secondary school.
Likewise, those for whom English is not their first language tend to make more progress at each successive key stage as their grasp of it improves.
So, for instance, Chinese children on free school meals make more progress than anyone else - so their predicted grades are adjusted upwards more than anyone else's (by almost 40 points) - so they have to do really well for their schools to get any credit.
Conversely, travellers do extremely badly. The biggest loading - an adjustment of predicted GCSE points of more than 100 - is for travellers of Irish heritage.
There is a warning in the official explanations about all this, in bold, that CVA must not be used to set expectations for any pupils - they must all be treated equally.
But inevitably, critics point out that using CVA makes this likely.
And they say there are other things wrong with it. For example, no account is taken of different funding levels per pupil.
But schools in deprived areas or with certain types of children get extra funding to take account of such things - so why count them again in the progress measure?
There are no independent schools in the CVA tables, by the way. The government does not have the necessary data on them.
But why do grammar schools not give as good a showing under CVA as under the old, simpler value added system?
Officials at the Department for Education and Skills have suggested this is because they and other highly popular state schools in better off areas, with good headline results, often attract pupils who have done well in the past.
Consequently their scope for adding value is reduced.
Another factor is that each pupil's attainment is capped, at their best eight GCSE results.
Many of the high-flying schools enter pupils for more exams than that - so capping the score artificially holds down their CVA indicator. Not surprisingly, some are crying foul.
Others just say parents will be baffled.
Statisticians will be pleased to see that the government has included in its tables "confidence intervals" alongside the raw CVA scores.
These allow people to decide how much significance to attach to the basic indicator.
For most, the rule of thumb is that schools with a CVA of more than 1000 are doing better than expected, those below 1000 are doing worse.
How much better or worse?
Officials say a score of 1006 means that on average each of the school's pupils achieved the equivalent of one GCSE grade higher in one subject than the average attained by similar pupils.
So a score of 1048 (1000 plus 6 x 8) means that on average each pupil achieved one GCSE grade higher in each of their best eight subjects than the average attained by similar pupils.
And a score of 994 means that the school's pupils achieved one grade lower, and so on.