Events | 2009.12.02
Michael White's political briefing | When the statistics are flattering, the headlines are more modest
It is a notorious fact of life, not confined to whipping boys of the public sector, that what can easily be measured is what gets measured and what is harder to measure tends to be neglected. So deaths from hospital-acquired infections routinely generate headlines while citizens walking round in good health because the NHS has given them high blood pressure pills or tests for bowel cancer do not.
Yesterday it was the turn of the education system to take another kicking, several actually, though most adverse media comment focused on a report from the Office for National Statistics. It concluded that £30bn worth of extra spending has not generated the hoped for increase in schools productivity as measured by ONS yardsticks of output - GCSE results included - against inputs such as money, staff and equipment.
As the parliamentary standards watchdog, Sir Christopher Kelly, noted at a conference this week, greater transparency and higher standards do not always translate into higher public confidence. Alan Johnson was insisting yesterday that crime is steadily falling. At the weekend the Care Quality Commission denied sweeping charges of lax standards in top hospitals.
Do voters believe them? Often not. "Reports which are positive are deemed unbelievable, those which are negative are believed," one government official noted sorrowfully last night. The approach of an election and the cyclical decline in the government\'s authority make for open season attacks.
It is not as if Labour\'s top-down drive to raise educational standards is beyond criticism. Controversies such as the widely perceived grade inflation in A-level and GCSE results – itself pre-dating 1997 – have also contributed to public scepticism about statistical measurement of its policies and achievements. A decade in office makes waste and duplication more likely; only yesterday David Law, the Lib Dem education spokesman, stumbled on what looks like a costly duplication of an online service for teachers which the private sector already offers them free.
But as seasoned politicians should know – George Osborne\'s shadow Treasury sidekick, Phil Hammond, seems in danger of forgetting – is that the hardest time to demonstrate higher productivity is when extra funds are pouring in and extra staff are being hired; ditto new schools and hospitals.
As education ministers muttered yesterday, wider outcomes than GCSE results, such as the cost of universal childcare, are harder to measure. And the quickest way to improve productivity is to sack better paid teachers and expand classroom sizes, not what parents want at all. In fact the ONS\'s chart suggests that output has broadly risen in line with inputs; what has not risen is what the system gets for £1 worth of taxpayers\' investment, the bigger bang for the buck: elusive productivity.
Last week Hammond claimed to have detected £60bn worth of lost public sector productivity compared with the private sector since 1997. His hopes of rectifying the failing may fall foul of what the private sector economists call "survivor bias." Failed private firms simply disappear, leaving most survivors "above average."
But the usually labour-intensive public sector cannot drop out of "unprofitable" sectors or go bust any more than it can easily get rid of disruptive, foreign-born or special needs pupils or get them to sit their GCSEs privately so as to maintain grade average. The same is true of hard cases in the NHS, the chronically ill or mentally unstable.
So if some statistics, such as GCSE results, flatter Labour\'s education record, others – like yesterday\'s batch – do the reverse. The Teach First recruitment programme for top flight graduates edged ahead of Goldman Sachs this year. The headlines were modest.
For more information, please visit
http://www.guardian.co.uk/commentisfree/[...]white-statistics-education-ons
Instagram is adding automatic in-app text translation - The Verge views: 1530
Here’s a ‘Brand-New’ Massive Multilingual Dataset for Machine Translation views: 1527