Wednesday, June 18, 2008

Goodhart’s law and Brown’s poll collapse

In the middle of a piece about Gordon Brown, Jonathan Freedland says this:

All this came to a head of course with last autumn's phantom election. Besides the machinations clearly designed to give him a poll lead, the uncertainty created a new part of the Brown persona: that he was indecisive.

That short aside, “machinations clearly designed to give him a poll lead”, leapt out at me. I’ve argued before that Labour’s poll leads last summer and early autumn were artificial and illusory, but Freedland’s phrasing here makes me see more clearly some of the reason for this.

It’s Goodhart’s law (usually found in economics): when some visible measure has been used as a surrogate to indicate the level of some underlying phenomenon, and then is targeted directly by policy, it ceases to be a reliable indicator of what it used to be.

In this case, the early Brown strategy of presenting him as the calm, competent leader of the nation - with a flurry of smallish yet nice policy changes - served to shift the voting intention polls a fair amount. But during this period, general satisfaction with the Government stayed very low.

So, for a while, Brown’s newness broke the link between declared voting intention and deeper political attitudes – it produced a shift in the former at the cost of making it a poor indicator of the latter. Which is why, when the Tories got their act modestly together during their conference week, and Brown made a couple of presentational mis-steps, the whole thing came crashing down so suddenly.

As a ten-year Chancellor, perhaps Brown should have known about Goodhart’s law.

No comments: