Reading 6 — the authority of numbers

Sample question

Reading Passage — The authority of numbers A A number has a peculiar kind of authority. Tell a committee that a hospital is performing badly and you have offered an opinion, which can be met with another opinion; tell it that the hospital's waiting times have risen by eleven per cent and you have offered a fact, which can only be met with another number. A figure quoted in a headline is rarely questioned in the way that a columnist's opinion would be, even by readers who have no idea how it was produced. It is tempting to explain this authority by saying that numbers are simply more accurate than words. The history of how public life came to be run by figures suggests a stranger explanation: that numbers are trusted less for what they reveal than for what they allow us to leave out. B Consider what a measurement requires. To count the unemployed, someone must decide who counts as looking for work, how long a person must have been without it, and whether a student with a weekend job belongs in the total. Each decision is a judgement, and reasonable people make it differently. Once it has been made and written into a procedure, however, it disappears from view. The figure that comes out at the end carries no trace of the arguments that went into it, and it can be passed from one office to another, and from one country to another, without anyone needing to know who made them. C That portability is the heart of the matter. A judgement belongs to a person, and it is only as trustworthy as the person is thought to be. A number produced by a fixed rule belongs to no one in particular, and it can be checked by anyone who follows the same rule. Historians of science have argued that this is exactly why quantification spread fastest not where experts were most powerful but where they were weakest — among officials who had to justify their decisions to a suspicious public and could not simply ask to be believed. A rule that anyone could apply was a defence against the accusation of favouritism. The number did not have to be right; it had to be impersonal. D None of this makes measurement a trick. A procedure that is applied the same way every time has real virtues: it treats like cases alike, it can be inspected, and it allows a result obtained in one place to be compared with one obtained somewhere else. A world in which every official decided each case by personal impression would not be fairer; it would be more open to prejudice, and far harder to challenge. Courts, tax offices and examination boards all depend on exactly this kind of consistency, and nobody would wish them to abandon it. The point is not that numbers conceal judgement dishonestly but that they conceal it by design, and that the concealment is part of what makes them useful. E The difficulty begins when the number stops describing an activity and starts to direct it. The economist Charles Goodhart observed in the 1970s that a statistic used as a target for policy tends to lose the stable relationship with the economy that made it worth watching. The principle has since been applied far beyond economics. A school judged by its examination results has every reason to concentrate on the pupils nearest the pass mark; a hospital judged by how quickly patients are seen has a reason to see them quickly, whether or not it treats them any better. In neither case need anyone be dishonest. People are simply doing, very sensibly, what the measure rewards. F What makes this hard to correct is that the remedy usually proposed is more measurement. If waiting times can be improved at the expense of care, then care itself must be measured too, and then the time taken to record care, and so on. Each new indicator is reasonable on its own terms. Together they produce an institution that spends a growing share of its effort on describing what it does, and whose staff come to experience the numbers not as information but as a form of supervision. G It does not follow that the answer is to abandon figures and return to trust in individual judgement, which had well-documented failures of its own. A more modest lesson is available. A number is most reliable when those who produce it have no stake in what it says, and least reliable when their pay, their reputation or their funding depends on it. The same waiting-time figure can be an excellent guide when it is collected quietly for the sake of understanding and a poor one when it is published in a league table. What changes is not the arithmetic but the incentive. H Perhaps the deepest point is the one with which the story began. We trust numbers because they seem to speak for themselves, free of anyone's opinion. That impression is not false, exactly, but it is incomplete, and the more confidently a figure is presented, the easier it becomes to forget. Every figure is the end of a chain of decisions made by people, and the right response to a striking statistic is neither to accept it nor to dismiss it, but to ask who decided what should be counted, and why. — Which heading best fits Paragraph B?

• Judgements hidden inside a procedure
• Why impersonal rules appealed to weak officials
• When a measure becomes a target
• Incentives, not arithmetic