thetorydiary

Stephan Shakespeare's Polling Report.

Just a short note tonight about 'margin of error': an obvious but widely misunderstood measure used with every poll.

When it says after a poll, "margin of error +/- 3%" it means that there's a 95% likelihood that the actual proportion of the total population that would answer the survey in the way indicated is within three percent of the number given. In 5% of cases the real number may be even more different.

Now what most people don't realise is that this margin applies to each of the percents. In other words, if the poll has the Conservatives on 40% and Labour on 30%, then it would still be within the given ‘margin of error’ for the Conservatives to be on 43% or 37%, and for Labour to be on 27% or 33%.

That means when the lead is given as 10%, the real lead can be between 4% and 16% and still be within the ‘margin of error’. Now of course it’s more likely to be in the number stated and not the extreme of the margins, but no pollster quoting a 3% margin of error (which is what you get with a poll of 1000) can consider, say, a 5% or an 8% as significantly different. When you see the polls bouncing around, it doesn’t matter there’s been a big shift in actual opinion.

The true value of daily polling is that when you look at a whole sequence of polls you can be confident that the trend-line is real. With a weekly or monthly poll, it takes much longer to know whether a particular 3-point jump means anything at all. Whichever time series you're looking at, make sure you watch for the pattern, don’t over-interpret small movements.

Stephan Shakespeare

Stephan Shakespeare, CEO of YouGov, writes his daily guide to election polling.

Yesterday I wrote about the three types of swing.

Let’s remind ourselves how swing is defined: at the last election, Labour had 36% of the vote and the Conservatives had 33%. If this time it turns out to be, say, Labour 31% and Conservatives 40%, that would be called a 6% swing from Labour to Conservatives (half of the change in the respective shares of the total vote).

But we need to remember that this does NOT mean 6% of people changed their minds and moved from voting Labour to voting Conservative. That nominal 6% swing actually consists of all kinds of churn, a significantly larger amount of movement between lots of parties, and between voters and non-voters. You could theoretically even achieve that swing simply by a big chunk of Labour voters staying at home, with no-one changing sides at all.

But let’s now consider only Type 1, those people who are actually considering changing their vote from Labour at the last election to Conservative this time around. I want to invoke Prospect Theory, which I first discussed here three years ago. I won’t rehearse this well-evidenced Nobel Prize-winning theory: in summary, people tend to over-value what they already have compared to a rational valuation of a new prospect. It’s a vital theory for elections as it explains the dynamics of hope v fear.

In a straight fight between giving up what you have in exchange for something new, the odds are heavily stacked against the new prospect. In order for people to vote for change, they must either (a) really dislike what they already have, or (b) really love the new offer. In 1992, they didn’t dislike the Major government enough, nor was the Kinnock alternative seen as very attractive. By 1997, they despised their government, and all that Blair had to be was likeable and safe.

So the choice for the Cameron Campaign depended on whether they thought our current mood towards the government was more like 1992 or 1997. They decided on 1997, and went for likeability and safety rather than the strong, highly differentiated offer they would have to attempt (and which Kinnock failed to provide) if it was 1992. About six months ago, it all seemed very 1996: things looked bad for Brown. But of course, history doesn’t repeat itself exactly, and we are neither in 1992 nor 1997. The twist was a fragile recovery from a frightening recession. People started to credit Brown - just a little - with steering the nation out of a disaster. His negatives reduced. The Conservative modern-young-man safety-first strategy started to seem a little shaky.

I think we’re delicately poised. The public, as measured by opinion polls, has certainly shifted towards change – but the 5% advantage to Labour that is currently built into the electoral map might mean the shift isn't yet decisive enough. Labour’s strategy is clear and simple: play hard on fear of loss, fear of change. But how should the Conservative campaigner respond?

Yesterday’s taunt of “we’ve experienced Brown’s experience” is too clever, too indirect, and should be dropped. A conventional professional strategist’s view, and possibly the right view, is that they should now go strongly negative, that they should hammer unrelentingly and viciously in a sustained attack on Brown’s fitness to govern.

Many people don’t like going negative, but they should remember that the finest and most successful example of negative campaigning in British political history was practised – albeit with attractive smiles - by Tony Blair, ably assisted by Brown and Mandelson. Day after day they ripped into Major as “weak, weak, weak,” mercilessly, hour by hour they pushed their unremitting assault on any remaining shred of respect for any living Conservative, the party of sleaze. And recall how they frightened pensioners by whispering in their ears that the Tories would clip their pension?

That was the very art of poison. It would be going too far, would feel too uncomfortable today, but if the Conservatives want to be surer of victory, they should realise that Type 1 Swingers need to be jarred out of their comfort zone, they need to be reminded of how they felt six months ago, and made to worry that it’s actually risky for them to stay with the old, the incumbent. And they should also be offered a carrot. Otherwise too many of them might just stick with what they’ve got.

We’ve seen the beginnings of a tougher Conservative campaign, but nothing approaching the toughness that Labour showed in 1997. And while they have already committed in their spending plans to a carrot or two, they aren’t yet putting them into their shop window bathed in bright lights. I suspect that will happen before long.

Stephan Shakespeare, CEO of YouGov, writes the first of a daily guide to election polling.

Over the coming weeks, we'll be interested nor just in 'swing', but in three different varieties. In this space I'll be examining each one, but let me start today by defining them. They are not new, of course, but the lesser varieties may this time come into greater prominence.

The first type of swing is the most obvious: Type 1 is between the two realistic contenders to form a government, and we clearly expect this swing to be from Labour to Conservative, though we're uncertain about the size.

Second is the swing from major parties to minor parties, and vice versa. This can be divided into two sub-categories: 2a is the one we are well familiar with, 'tactical voting', where people vote for a party not because they support it, but because the party they prefer cannot win in their constituency, so they vote for the preferred of the two real contenders. And Type 2b, the non-tactical vote for minor parties (Green, Ukip, BNP etc).

We may see more of 2b in this election, because people are more cynical about politicians and parties than ever before, and express far lower levels of loyalty. "The main parties all say the same thing," some will decide, "So I might as well express my support for the party (or politician) that comes closest to my real views, even if they can't win".

Types 1 and 2 are ok for the pollster: we are reasonably well practised at measuring support for minor and major parties. The swing that really worries us is Type 3: between voters and non-voters. Many people who don't end up voting tend nevertheless to think of themselves and express themselves as voters, because they feel it's the right thing to do. They want to feel like active citizens, they may really believe they will vote, but on the actual day practical obstacles or plain laziness means they don't make the  effort. The drivers of voting behaviour may (or may not) be changing, for social as well as political reasons. And if the swing to or from non-voting is disproportionate - that is, if one party's stated supporters are more or less likely actually to make the journey to the polling booth - then that could make a significant difference between the opinion poll number and the election result. Pollsters often apply 'likelihood to vote' filters to their polls, especially on the eve-of-election when it's slightly more reliable, but there is no known reliable methodology for it.

It may not matter. We've been accurate so far, and we will probably end up with 'politics as usual'. But with the stated differences between the main parties perceived as narrow, and following the expenses scandals as well as the wider disillusionment with mainstream parties, this could be the election where Swing Types 2 and 3 become a bigger factor.