Monday, February 04, 2008

Opinion polls: Cameron vs Kinnock?

The Tories’ poll leads have been dropping recently (one poll last week even put Labour ahead), and of course there’s much media comment on whether and why David Cameron’s efforts are stalling. I’ve also seen several pundits remark that Labour’s poll leads while in opposition were much more impressive.

Those of us of the anorak-wearing persuasion know that this is technically true, but deeply misleading, because polling methodologies have changed over the years to deal with the pro-Labour bias.

So, what follows is an effort – and it’s a bit rough and ready – to compare opinion polls now and earlier in the 1990s. Skip the first bit if you trust me completely and/or don’t care for nitty gritty.

I’m using ICM voting intention figures because they were first to start adjusting their polls after the 1992 result so surprised people. While other pollsters were routinely giving Tony Blair ludicrous leads in the 30s and even 40s, ICM were consistently more plausible. Since 1997, they’ve made further methodology changes (as have the others).

(Hat tip to Anthony Wells’s excellent UK Polling Report for the figures.)

I’ve taken the average of ICM ratings in the three months prior to the 1992, 1997 and 2005 elections being called – because all we have to go on at the moment are the hypothetical ‘how would you vote in an election tomorrow’ polls, and I want to compare like with like as far as possible. I’ve then looked at how these averages compare with the actual results to calculate how accuracy has changed over time.

Pre-campaign 1992 polls gave Labour a lead of 3 points, whereas the Tories finished 7.6 points up – a Labour bias of 10.6.

Pre-campaign 1997 polls gave Labour a lead of 17.8 points, whereas Labour finished 12.8 points up – a Labour bias of 5.

Pre-campaign 2005 polls gave Labour a lead of 5.5 points, whereas Labour finished 2.9 points up – a Labour bias of 2.6.

To convert older polls into new money, so we can compare them with contemporary polls, then, we need to subtract 8 points from Labour’s lead in pre-1992 polls, and to subtract 2.4 points in 1992-97 polls.

These are the poll leads/deficits for different oppositions, averaged over given periods and converted into contemporary equivalents (original figures in brackets):

Kinnock’s Labour
(last year of Thatcher, Dec 1989-Nov 1990)
+7.6 (+15.6 in old money)

Kinnock’s Labour
(Jan-Dec 1991)
–4.6 (+3.4 in old money)

Smith’s Labour
(last year of his life, Jun 1993-May 1994)
+7.5 (+9.9 in old money)

Blair’s Labour
(Jan 1995-Dec 1996)
+16.6 (+19 in old money)

Cameron’s Conservatives
(last year of Blair, Jul 2006-Jun 2007)

Cameron’s Conservatives
(post-‘non-election’, Oct 2007-Jan 2008)

So: since Gordon Brown contrived to end his own honeymoon, Cameron has been doing about as well as he had been against Blair (although the three ICM polls this January average a lead narrowing to 4.7 points). Cameron is doing a good deal better than Kinnock did against the new John Major, but not as well as Kinnock had done against the declining Thatcher nor John Smith against Major. He has come nowhere near Blair’s – very sustained – performance against Major.

I’ve not looked at the relevant numbers, but he’s certainly doing better than his three predecessors managed against Blair. The other thing to bear in mind is that he may also be doing a bit better, in reality, than the polls say, assuming that the small pro-Labour bias from early 2005 still exists.

No comments: