Forget Iraq, We Care About Our Pockets

On the morning of polling day in the 1983 general election, I had a conversation in the local Labour committee rooms in Hackney. What do you think the result is going to be tonight, the ward secretary asked. I think the Tories will win by more than 100 seats, I replied. The ward secretary seemed taken aback. That wasn't the mood on the doorstep, he countered. In his view Labour would win by about 20.

Well, when the polls closed, it soon became clear that I was right and that the Labour ward secretary was wrong. Mrs Thatcher came back with a majority of 144, Labour slumped to 28% nationwide and just escaped falling into third place.

Fast forward 14 years to the 1997 election. Another election-day conversation with another Labour official, only this time with the leader of the party. The subject was the same as before, only this time it was my turn to ask the question. What do you think the result will be? We might win by 50, he replied. More like 150, I retorted.

The purpose of retelling these anecdotes is not to boast. The truth is that my powers of prediction are no better than anyone else's. My secret on both occasions was a simple one. Unlike the people I was talking to, I consistently believed the opinion polls. The polls said Thatcher would win by a landslide in 1983, and they said the same about Blair in 1997. I merely repeated what the polls were saying.

I believed the polls then, and I believe the polls now. That is why, for example, when people ask me what I think will happen in the US presidential election, my answer is that John Kerry is winning. I say this, my abusive rightwing American email correspondents may like to note, not because I am a jerk or a pinko or because I come from a country that has lost its manhood, nor even because I want it to happen, but simply because that is currently the conclusion of most of the polls, especially in the battleground states.

Which is not, I stress, the same thing as saying that Kerry will win. Maybe he will and maybe he won't. No opinion poll is foolproof. All of them have margins of error. And none of them is ever a prediction. They are merely a necessarily approximate snapshot of the state of opinion at a point in time. And since the US election is not until November 2, and the nuances and momentum of the polls change all the time, no prediction made in August can be anything but a more or less informed guess. All you can say, right now, is that it is a close election and that Kerry is narrowly in the lead. Calling me a jerk doesn't change that, you bigots.

I accept, though, that those of us who believe the polls currently have some explaining to do on the domestic political front. This is particularly true in the combined wake of last week's Guardian-ICM poll, which found Iraq at the bottom of a list of prospective election issues, and Friday's Financial Times-Mori poll, which concluded that defence and foreign affairs were the most important issues facing the country.

Shurely shome mishtake, even good observers have said on seeing these results, among them Peter Preston, who confessed to "plain confusion" on these pages 24 hours ago. But I think the confusion is more apparent than real, and I also think it is important to be able to show that this is so, not solely for reasons of amour-propre , but also even for reasons of the civic good. It would be good for our collective politics if we could stamp out the lazy conceits that the polls can't be trusted and that they can be made to support anything you want them to say.

There are many variables between different polls - date, size of sample, method of contact, weighting and the rest. More than anything else, though, the results of different opinion polls depend on the questions that the pollsters ask. Confusion in; confusion out. Clarity in; clarity out. That is why the key to unravelling the ICM/Mori "confusion" is to be clear that they asked very different - not contradictory, just different - questions.

In its poll for the Guardian, for example, ICM asked its sample: "Now, thinking about the next election and the issues that will be important in your own decision on how to vote, which two or three of the following will be most important in your own decision on which party you will vote for?" Respondents were offered 10 subjects, of which "Iraq" - that one word alone, note, not some other formulation like "the Iraq war" or "Tony Blair's imperialist lies about Iraq" or any of the almost infinite number of possible permutations - was one. Just 12% of all voters selected "Iraq", which came bottom of the pile topped by the NHS on 59%.

Meanwhile, going into the field that same weekend for the FT, Mori was asking its sample: "What would you say is the most important issue facing Britain today?", before asking them to add "other important issues facing Britain today". Mori's sample were not prompted with a list of issues - unlike the ICM sample - and their unprompted replies were then allocated by the pollsters into Mori's regular categories. In this latest FT poll, "defence/foreign affairs/international terrorism" came top for "most important" single issue (24%) and retained that place when "other important issues" were added in (38%).

Does this mean that the polls are saying contradictory things? It is possible - rogue polls can happen - but very unlikely. It certainly is not happening in this instance. The key thing to notice is that the two polling organisations were simply asking different questions: ICM about the issues that would influence the way you voted at the election, but Mori about the most important issue facing the country.

These two polls, far from being contradictory, are in fact compatible. It is perfectly possible to see a subject as the most important facing the country while simultaneously not giving it great importance in your decision on which way to vote. In the 80s, polls regularly showed that unemployment was deemed the most important issue facing the country, and even that Labour had the best policies for dealing with it. But that didn't mean that people voted Labour. They voted Conservative because, when it came to voting, issues like taxes and economic competence - on which the Tories had the advantage - mattered more.

A similar dynamic is at work today. Foreign affairs, defence and international terrorism may indeed be the most important issue facing the nation - a pretty capacious bag that category is, by the way, containing many other issues besides Iraq. A party which could find something compelling to say about that bunch of issues would certainly have an advantage - as the Lib Dems have discovered. Even so, when it comes to the election, as ICM found, it's the same old story. Next year, most people will cast their votes because of the things that make a difference to themselves and their families - health, education and the rest of it. Which is why Labour will win. Just as the polls say they will.