Paul Buckley and Nigel Hawkes disagree
The world is full of well-meaning evangelists out to change the way other people behave. Their targets are many: those who smoke, drink, get fat, engage deliriously in unsafe sex, even - heaven forbid - talk on their mobile phones while driving. Don’t they know these things are wrong?
It’s hard sometimes not to have a sneaking regard for the miscreants in their rejection of good advice. But that isn’t a view public health specialists can permit themselves. They want to know how best to get across the messages that are so often ignored. My claim is that the right way to do it is to ensure from the start that truth is being told. ‘Now, what I want is, Facts ...Facts alone are wanted in life’, as Mr Gradgrind put it in Dickens’ Hard Times.
Statistics are generally as close as we can get to facts in this imperfect world. They model reality better than any other technique so far invented. They provide forceful, memorable, and at times almost tangible representations of human behaviour and its consequences. Without them, we might as well uninvent the Enlightenment. We need more and better statistics to convince the unwilling.
‘Lies, damned lies and statistics...’. Disraeli was right. Statistics aren't facts. Subjectivity always occurs. It’s just a matter of where and when: deciding what questions to ask, what variables to analyse, how to analyse, how to present the results. Any supposition can be supported with the 'right' statistics. Take a quote from a current anti-aging cream advertisement: ‘80 per cent (of women) saw visible results in the appearance of their wrinkles.’ This is cited as 'proof ' that the product works. It's a statistic, but what is the fact? It doesn't prove (or mean?) anything.
Whilst better statistics would be welcome, do we really need more? I argue that less is more. What we really need is a better understanding of what we already have. The world is overrun with a surfeit of statistical information, much of which serves no useful purpose.
A statistic alone is little more than a number. Indeed, rather than, ‘provide... almost tangible representations of human behaviour’, statistics can disguise the richness of the multifaceted human condition. Statistics serve as much to obscure as to enlighten. In order to be of value, a statistic needs interpretation, interpretation requires understanding and understanding requires ability and motivation. Statistics are nothing without psychology and context.
Of course statistics are a matter of choice. Not all are equal in value. But to damn statistics by picking an example from cosmetic advertising is scarcely a conclusive argument. And it isn’t true that any supposition can be supported with the ‘right’ statistics: find me a legitimate statistic that shows smoking is good for you, or that the population is declining, or that marriage rates are increasing.
I’d never pretend that statistics can capture the whole of the human condition, any more than science can capture the whole of human knowledge. But they form bedrock without which we struggle to understand anything. Pity the policy-makers of the 1930s, who had to grapple with the Great Depression without even a way of measuring the GDP. Psychology didn’t get them far.
Indeed, psychology itself wouldn’t have got far without statistical approaches for summarising observations, understanding variation, and testing interventions. It would simply be a mass of undigested data, prey to charismatic individuals able to stamp their own impression upon it, rather like psychoanalysis. By all means let’s have understanding and interpretation, but without data systematically gathered and properly organised – statistics, in other words – what is there to understand and interpret?
A statistic is a statistic. Cosmetic advertising statistics may not be very valid, but they are nonetheless statistics. What constitutes a legitimate statistic is itself a subjective viewpoint. It could be shown that smoking reduces stress – a positive outcome, or that the population is declining for example inSouthern Italy, or that marriage rates are increasing for example in homosexual groups. It just depends on the parameters chosen. The fact remains that the choice of statistic and data to analyse can produce the answer required. AverageUKpay figures are vastly different depending on whether the mode, median or mean is used in the calculation.
Whilst I would accept that statistics help to summarise some data, without the human ability to assess information and make inferences, data per se is worthless. Statistics are only a tool. The policy makers of the 1930s needed to understand the meaning of any statistical data and to go beyond it to make inferences of what it all meant. Indeed the statistical data would have been very incomplete and required assumptions and inferences to be made.
Data can be systematically gathered and properly organised without being statistical; interview data would meet both these criteria. In no way would it be statistical, but would inform and aid understanding.
Statistics alone can never persuade. 98 per cent of smokers are aware of health warnings on cigarette packets – so why does over 20 per cent of theUKpopulation still smoke?
I’d be the last to deny that statistics can be misused, as I’ve spent the past three years documenting such misuse. People frequently use the mean when the median is more appropriate, as in the current discussion of top executives’ pay. But that’s not a good reason to throw the baby out with the bath water. Indeed, the fact that statistics are so often misused is a tribute to their persuasive power, not a reason to abandon them.
Nor am I claiming that statistics alone are sufficient. You are right to say that we need to go beyond statistics and make inferences. We also need to understand the limitations of any set of statistics, which is why the flood of data emerging from the UKgovernment through its www.data.gov  portal without statistical commentary is potentially dangerous.
But recognising the limitations of statistics is not the same as trying to do without them. Qualitative data is fine and can be illuminating, but trying to make policies on the basis of qualitative data alone is like navigating in a fog without radar. One either goes slowly, or one goes aground. Statistics are not a substitute for human judgement: but without them such judgement is blind.
I agree that appropriate statistics are a valuable tool, without which our understanding would be severely impaired. Neither would I argue that qualitative data alone is sufficient on which to base policy, though it does humanise data.
Rather than arguing whether statistics or psychology has the answers, the two together promote understanding. All information, whatever its theoretical stance, is of value if it aids understanding and clarity.
Whilst judgement may be somewhat short sighted, rather than blind, without statistics, ultimately people think of things illogically and then rationalise. To paraphrase Kant, they ‘see the world as they are, not as it is.’ It is therefore extremely difficult to change public behaviour. Someone somewhere will twist even the most elegantly argued proposition to fit what they want to believe, for example: ‘Stopping smoking is bad because you gain weight, so it’s better to carry on smoking.’ Even if the information is accepted, it does not necessarily lead to changes in behaviour. Most people accept that it would be good to stop smoking, and keep weight under control. Changing their behaviour is the difficult part!
Not only do we need good data, but also - given widespread false and misleading information – general education on what data means.