Skip to content

Entirely neutral government statistics? Really? Where?

In today’s Scotsman academic Arthur Midwinter claims that the Scottish Government is not being honest about statistics. This would not make them an isolate case…

For nearly two decades I’ve worked around about governments. In all of that time I don’t think I can identify a single occasion on which a single civil servant, when faced with two possible statistics, chose the one that made the relevant government look bad. In fact, I can go further. I have worked closely with political parties, lobby organisations, business groups, journalists and charities. In all the time spent with these organisations I simply can’t think of a single occasion in which anyone ever used the version of the data that made them or their paymasters look bad. Certainly some were worse than others, but the concept of politically-applied-statistical-purity will be along to surprise us about ten minutes after the Loch Ness Monster.

Arthur Midwinter has spent a career producing some of the most valuable and important statistical work about Scotland, particularly in the sphere of local government. Much of what he has done has been as an independent adviser to an organisation which commissioned him to do the work. In the commission, in the appointment, in the oversight, in the people chosen to contribute, in the source data provided, in the negotiations about the final report, in its presentation, in its future use – no compromise? Nowhere in all of this process has the commissioning body exerted influence on the direction of the work? If not, I would argue that the commissioning body isn’t doing its job properly.

Seldom do we ask questions to which we have no idea what the answer is or where we have no vested interest whatsoever in the answer coming out one way or another. The use of data and evidence is Hegelian, not Platonic. What I mean by that is that it is only useful if we accept that one side assembles the evidence to support its argument and the other assembles it differently for its purposes. Then, debate and discussion ensues, and hopefully the best argument wins. Not the best data – we are not in a Platonic world of perfect, neutral answers. The best argument which hopefully involves data and evidence which makes it stronger.

Now I get as frustrated as anyone with governmental distortion of data. For example, the Office of Budget Responsibility is an organisation set up by a politician with a pretty extreme agenda (George Osborne) with a specific agenda embedded in its name (it isn’t called the Office for Statistical Responsibility). Crudely, it’s role is to report on how well austerity is going. Does the OBR produce reports on projections about how a Keynesian approach to the economy would have been working? There has been media gossip suggesting there have already been some differences of opinion between staff and paymasters. And even in its own areas, how ofter does it revise its last estimate? Perfect purity? Give me a break.

This is not a defence of the Scottish Government which in the past has pushed evidence which to my mind gave a misleading picture (and one which at that time worked against my own massaged evidence base). Nor is it a defence of the Scottish Executive before it (which did the same things as far as I was concerned but which doesn’t seem to bother the Prof). It is a defence of Scotland and a defence of human intellect. I’m sick of the idea that Westminster is intrinsically better and more honest that Holyrood and that we should ‘defer to our betters’. After all, Westminster did a great job of predicting a financial crisis that the left saw coming for miles.

But above all, I’m sick of the idea that the numbers can do our thinking for us. They can’t. They’re just numbers. Professor Midwinter is good with the numbers, but then again, apparently the answer to life, the universe and everything is 41. (I await the accusation of inaccuracy contained in my previous sentence and realise that it will invalidate my entire argument. If only I’d checked it with a statistician first…)

Robin McAlpine