Politics

On the probability of not dying

May 29, 2008
Placeholder image!

If in doubt, the Old Left would nationalise or regulate. If in doubt the New Right would privatise or deregulate. And if in doubt, the soupy left-right blend that now unites us has its own default option: quantify and publish.

Today The Guardianreports that the government is preparing to publish the death rates of patients undergoing major surgery at NHS hospitals in England. Boris Johnson won the London mayoralty with a promise to publish New York-style 'crime maps', detailing the areas of London that suffer from the worst crime.

The practice of tabulating and comparing 'outputs' has been growing in policy for a number of years now. The production of rankings is always the highlight of the World Economic Forum's World Competitiveness Report, in which nations around the world are placed on a chart from the most competitive to the least. And New Labour has been infamous for its league tables, especially in education. The hope in such cases has been to create incentives to alter managerial strategy or try harder. Revealing a country to be the 35th most competitive in the world is meant to be a wake-up call for them to do better (unless it's France), as well as to give businesses an indication of where not to invest.

This policy trick is now being performed for the benefit of individual citizens, thanks to two developments.

First, there is the 'choice agenda', which takes the observation that consumer choice is what drives efficiency through markets, and applies it to public services. And second, there is the internet, which for the first time enables distribution of such information at virtually zero cost.

The ironic thing about these marketising strategies is that they aspire to something far more radical than anything present in actual markets. Markets have their third party adjudicators, such as Which? and independent regulators; but on the whole, people are perfectly content relying on brand loyalty, habit, taste and convenience to determine their choices, without needing tables to inform them which outlet is the best (or second or third best).

The notable difference is that there is qualitative differentiation in a proper market, which is generally lacking in the field of public services: we may not be able to place supermarkets in rank order, but we know vaguely what we think of each of them. The inability to produce this qualitative differentiation in public services is presumably why policymakers think we need quantitative differentiation - although this is scarcely the case with Boris's crime map, which will take an existing qualitative judgement ('this place is well dodgy') and transform it into a more robust quantitative one ('there were 187 muggings in this neighbourhood last year').

I wonder if policymakers have thought about what they expect people to do with this information. The analogy to market choice is not good enough, for precisely the quant/qual distinction just outlined. In a world of quantitative differentiation, it is illogical to choose anything but the 'best'. Who would choose an 'average' hospital to perform surgery on them? Who would even choose the 10th best hospital to do so? The same is true with crime: the only level of risk that people would ideally select is zero, but failing that they want as close to zero as possible. The crime-map will, no doubt, create a 'more perfect' housing market in the technical economic sense, as surprisingly crimeless areas experience an upsurge in valuation. Equally, unthought-of areas may gentrify, once artists discover that they're dangerous, but not that dangerous. Otherwise, the main consequence is surely one of rising unease.

The problem, it seems to me - with today's In Our Time fresh in my ears - is this. Statistics are about probabilities, and are therefore only useful to people doing things repeatedly, such as governments. Governments repeatedly spend money on hospitals and policing, and statistics help them spend that money more intelligently. If there is a 20% chance that a hospital will mess something up, this tells us nothing about whether it will or won't do so on any given occasion; but it does mean that over time it becomes more intelligent to divert the money to one with a higher success rate.

The psychology of the individual user is very different. Dying in the operating theatre either does or doesn't happen; getting mugged either does or doesn't happen. True, the latter can happen a few times, but by the time one has been mugged four or five times, the statistical probability of this happening is unlikely to alter one's impression of an area. The fact that some of these statistics (on surgery death and murder) relate to the ultimate one-off that is Life makes it all the harder to assimilate the worldview of the statistician with that of the lone, anxious person.