A READER - the source of most of this column's best subjects - sent me an interesting chart last week (right). It shows the incidence of reported crimes near a pub in Bromley, Kent, from August 2000 to November 2004. Is there a link between extended drinking hours and more violence? In the case of the Tiger's Head, the answer is yes - violence doubled when hours were extended. When the late-drinking licence was revoked after 18 months, crimes fell again, to less than half the previous level.
The chart looks like a simple graph. In fact, it's much more. 'It's immensely powerful - there are so many applications in local government and the public sector,' says Bromley councillor Tony Owen, who drew this one. It's a 'control' or, better, a 'capability' chart, and it measures a process's capability and variation - failure to understand which, according to quality prophet W Edwards Deming, is 'the central problem in management and leadership'.
As targets are to conventional, dysfunctional, management, measures of capability and variation are to the positive systems alternative. The exact opposite of the poisonous 'WMD' (weapons of management destruction) described on this page two weeks ago, such measures are, alas, criminally neglected - utterly ignored by boards, ministers and Whitehall policy wonks alike. Yet it is no exaggeration to say that their better understanding and use would do more to stabilise the NHS, improve public finances and raise UK productivity generally than all the government targets and interventions put together. (Actually, the latter necessarily make things worse, and not the least of the beauties of capability/control charts is that they show you why. Read on.)
The control chart was invented by Dr Walter Shewhart, godfather of the modern quality movement and mentor of Deming, in the Twenties. 'Control' here is slightly misleading: Shewhart saw it above all as an aid to predicting performance and thus management judgment and decision-making. It can take many forms, but at its simplest - as in the Tiger's Head example - it depicts the working of a process over time.
A critical component of the chart is its control limits, based on the calculation of standard deviation, which show the extent of natural variation in the system. Around the Tiger's Head, under standard opening hours, reported crimes per month could total anywhere between 12 and zero and under extended hours 25 and zero. Anything inside the control limits is 'normal', part of the system, and the process is stable or in control. A point outside the control limits, on the other hand, is evidence of an abnormal external cause that needs to be investigated, and if necessary removed, to make the process stable again. Only when the process is stable is predictability renewed.
The control chart has been aptly described as 'the voice of the process'. Unless it is heard, managers have no way of knowing if any particular number is a natural variation or something that warrants intervention. This is a critical distinction: if managers mistakenly tamper with a stable process, believing an occurrence is exceptional, they introduce an external cause, which destabilises it. Targets do the same thing.
If a system is stable, as a matter of logic you can only force it to deliver a target beyond its limits by improving it, distorting it or fiddling the numbers. It's impossible to know where and what to improve without a process voice to tell you - so, in the absence of capability measures, distortion and fiddling are the inevitable result. If the system is unstable, meeting the target is useless, because you have no way of knowing if you can do it again. Thus does the current public-sector regime of targets and interventions make systems work worse, raising costs and destroying morale thus Deming's insistence on the importance of the leadership task of interpreting and acting on variation.
Traditionally, control charts have been thought of, if at all, in terms of manufacturing industry. But this is a misconception, says Cranfield Management School's Steve Mason: 'There are many applications they could be used for, including at board level. We need to get boards away from looking at columns of spreadsheet numbers to something more meaningful.'
Another strong advocate, no stranger to these columns, is John Seddon, whose consultancy, Vanguard, puts capability measures at the heart of a philosophy that has helped a number of public- and private-sector organisations to performance improvements that make a mockery of official targets: benefits payments in four or five days against an official target of 50, or cutting resolution of IT helpdesk enquiries from days to hours. 'The test of a good measure is whether it helps in understand ing and improving performance,' he says. 'That's exactly what systems measures - of which control charts are one - do, and targets don't.'
The puzzle is why these measures are so scandalously neglected. 'There are few of us teaching them, and they don't figure in MBA curricula,' says Mason. It probably doesn't help that they are part of the unsexily titled subject of 'statistical process control', identified exclusively with nerds in white coats. Another reason is that so few people in the public sector have worked in environments where their virtues are recognised. 'Where I worked, at Philips, if we supplied Sony with more than a couple of faulty parts per million, we'd be out as a supplier,' says Owen at Bromley council. 'Here, 20 per cent of the holes in the road are unfilled.'
But although no one who 'gets' systems measures ever returns willingly to targets and specifications, even in the private sector the gains are all too often reversed by senior managers who take fright at the implications. This, says Seddon, is because those measures are part of a package - a systems view of the organisation - and can't be applied piecemeal as a handy 'tool'.
It's hard for managers to accept that, from a systems perspective, their function of devising top-down financial plans and budgets is the source of all the subsequent variation with which the inhabitants of the system have to cope (or not) as best they can. What is that if not the biggest leadership issue there is?
The Observer, 21 August 2005