When looking at opinion polls, it is tempting to look at the short-term and the headline data alone. The eight point Labour lead in today’s YouGov poll is attracting a reasonable amount of attention, just as the two point Labour lead in the same poll yesterday did.
To counter this short-termism. I thought it would be interesting to delve into Labour’s recent polling in more detail.
If I wanted readers, I’d do this in the form of a list, or perhaps trail it as ‘you won’t believe what’s happened to Labour’s poll lead..‘ But I did this for my amusement, and I’m not going to tart up a blog post about polling numbers to score a few hundred extra office eyeballs. Frankly, if you care about poll numbers you’ll want to read this. If you don’t, you won’t enjoy it. So you may as well stop now.
Mind you, if you do read this, It’ll mean you’re better than other people, because you’re not reading a tiresome controversialist pumping out their opinion macro. So there is an incentive.
I’ve looked at the internals of all YouGov polling so far in December, and compared Labour’s polling with the same period a year ago. I’ve done this to try to smooth out the inevitable bumps and dips in samples, reducing the chance of a single day’s polling throwing everything out of whack. The following is based on the first 12 YouGov polls of December 2013, compared with the first 12 polls of December 2013.
What can we see?
First, Labour’s overall poll share is down. This is perhaps salutary for those who have, like me, written posts and columns asserting confidently that in 2013 Labour has been setting the agenda, winning the argument and so forth.
A year ago Labour polled an average of 43.3%, within a range of 42 to 45. Over the first half of December 2013, Labour polled an average of 39.2%, within a range of 38-41.
This represents a decline of 4.1 points, or roughly a tenth of Labour’s support. To put it another way, Labour’s best poll share so far in December 2013 is equal to Labour’s worst poll share in December 2012. (For comparison, in the first half of September 2010, just before Ed Miliband became leader, Labour averaged 37.9%, within a range of 37-39. Since Ed Miliband became leader, Labour’s support is now roughly a point higher, but has gone significantly higher in the recent past).
Given the fall in share, obviously Labour’s lead over the Tories has also declined. A year ago, Labour had an average lead of 11.5%. So far this month, that figure is 6.1%.
Given that Labour’s share has fallen four points, that the lead has fallen by only five and a half points suggests that this year has seen more Labour softening than Tory strengthening.
This is demonstrated in this chart from Anthony Wells’ excellent UK polling report, Labour vote share seemed to dip in spring this year, gradually soften over the summer, and maintain, or even increase over autumn, before showing signs of softening a little again in December.
The Conservatives, on the other hand reached a nadir in late spring, recovered over summer and fell back a little in autumn. they may be recovering a little in December.
This leaves us where we are now, with Labour down on a year ago, the Tories a touch higher, the Lib Dems down a point or two, and UKIP up two or three points.
So those are the trends, but what comprises that fall in the Labour vote?
Here, we have to dive into the internals. Ready? Then let’s get this party started.
LABOUR BASE FROM 2010 SOFTENS
First, there’s been an increase in 2010 Labour voters saying they don’t know or won’t vote.
This may not seem like a significant change, but it roughly represents a point fall in Labour’s overall vote share.
If you combine the 2010 Labour voters who now say they won’t vote (or don’t know) with those who say they’ll vote for another party, then this time last year, Labour retained 83.2% of their 2010 voters. Now, that figure is down to 76.2%. This accounts for roughly half of Labour’s poll fall.
Where did these voters go?
They’ve scattered across the board. There are very small increases in the number of 2010 Labour voters who now say they’ll vote Lib-Dem and Tory, and a slightly larger increase in the number who say they’ll vote UKIP.
However, the numbers and sample sizes are so small, it’s hard to read much into that. That said, the number of 2010 Labour voters who now say they’ll vote UKIP is now the same as the number who say they’ll vote Tory, which is amusing, if not really statistically significant.
MANY LIBDEM SWITCHERS, BUT FEWER THAN BEFORE?
Obviously though, Labour has a far broader support than just 2010 Labour voters. There are those who didn’t vote Labour in 2010 but now say they will. To state the bloody obvious, these people are why Labour are ahead in the polls.
The most significant of these groups are 2010 LibDem voters.
Last year, 38.3% of 2010 LibDems said they’d be voting Labour. This year, 34.6% said the same (this excludes the c25% of 2010 Lib Dems who say they’re not sure how they’ll vote, a number that still seems to be reasonably consistent).
That shift is worth about two-thirds of a point in the polls, though Labour has still gained roughly 6 points from 2010 Lib Dems.
The number of 2010 Tories who say they’ll vote Labour has also fallen, but the proportion is small. Last year, excluding Don’t Knows, Labour had 5.8% of 2o1o Tories. Now, that’s down to 4.9%. However, this represents a mere third of a point.
In total, fewer 2010 Tories and Lib Dems are saying they’ll vote Labour, which is worth around a point of Labour’s decline. ((again, I’ve updated the data, but the datawrapper chart tool takes a while to update, so this may show slightly different figures))
WHAT ABOUT NON-VOTERS?
This suggests two things. First, although we don’t have separate data from YouGov it seems likely a significant proportion of Labour’s current support comes from 2010 non voters, and that Labour support among this group has also declined over the year.
For the first point, research by the Fabians last year demonstrated that more 2010 non-voters now say they’ll vote Labour than for any other party, representing almost as significant a proportion of the Labour vote bloc as former Lib Dems.
On the second, If we assume that this support has ebbed in line with that of other groups (falling by roughly a tenth), this would account for the remaining fall in Labour support since last year
There are two ways of viewing the support for 2010 non-voters for Labour.
You can either see it as a strong endorsement of the new approach Labour has taken since 2010, which has inspired people to vote, or you can argue that the most important fact about non-voters is that they do not vote, and so support they claim to give should be taken lightly. Whichever view you take, it’s likely this group is less supportive of Labour than this time a year ago.
VARIANCE BY AGE, SEX, CLASS AND REGION?
The suggestion that Labour’s support among 2010 non voters has declined somewhat is supported by the fact that Labour’s decline in vote share among social groups with a lower propensity to vote is in line with that of other groups.
There’s little really to note here, other than the consistency. You could argue that Labour’s fall among 18-24 years olds is more noticeable than in other groups, but it feels something of a stretch. The same applies to men and women. This month, Labour is averaging 37.6% among men and 40.2% among women. A year ago that was 42.3% and 44.3% respectively. That’s basically the same decline.
There is a regional difference though. In London, Labour share seems to be holding relatively steady, while it has fallen more sharply in Scotland.
So what does all this mean?
It means everything I usually say is totally right, of course. Everyone should heed me.
There is a worrying lack of heeding at the moment, and this heed-deficit can only lead to disaster.
To be a little more constructive, it’s probably better to reverse the question. What doesn’t it mean?
What it doesn’t mean:
This doesn’t mean that Labour is destined to win or lose the next election.
Just because Labour lost about four points this year doesn’t mean Labour will lose another four points in 2014. Labour’s current poll lead is healthy enough to secure a general election victory. In general, oppositions lose support in the 18 months before an election, but there are no guarantees. I can guess that on average an opposition would decline by about three points over the next year and a half, but it’s not for certain at all.
What it doesn’t mean:
This doesn’t mean people can write end of year political columns claiming that Labour has ‘won the argument’ over the course of 2013.
Or at least, they can’t write them and claim that they mean in data terms, rather in a sort of ‘communion with the zeitgeist’ sense. Whatever the level of Labour’s connection with the zeitgeist, the polling numbers don’t suggest Labour’s connection with voters got stronger this year.
What it doesn’t mean:
Labour can point to a single group of voters, region, or electoral bloc, and say ‘aha, if only we could persuade a few more red-headed UKIP tempted residents of Coastal towns, we’ll be fine‘. The most interesting thing for me about Labour’s performance this year is how diffuse the decline has been. Labour’s lost a little bit to everybody, but not a lot to anyone.
Now, I’m a fan of detailed data crunching, and I long for the day when I can atomise the electorate (literally, sometimes). I don’t discount the value of picking out small groups of voters and focussing on them at all. It’s essential to a hard-fought victory. However, this data makes me consider the absence of a big political swell that shifts the dial, rather than long for ever greater application of targeting. It reminds me that targeting actually works best as an amplification of a wider message, not a strategy in itself.
To put it another way, good retail requires good wholesale.
What it doesn’t mean:
That Labour can rely on our ‘base’ or on a defined group of switchers to “guarantee” victory.
Electoral strategists like to play with electoral blocs as if they were lego, endlessly combining and combining them together in search of a majority. This is, for my sort of person, an awful lot of fun. (Yes. I’m like that. Out and Proud) The better your data, the better the accuracy. But individual voters aren’t part of a bloc. There are ‘fringe’ 2010 Labour voters, and ‘fringe’ 2010 Lib Dems who say they’ll vote Labour.
Overall, many 2010 LibDem voters are politically aligned to Labour, think of themselves as on the ‘left’. I have little doubt that a lot of them will vote Labour in 2015. However, that is no guarantee that another point, or two, of 2010 Lib Dems might melt away, just as some 2010 Labour voters might decide that this time, they don’t think we should win. The ‘fringe’ of these blocs can decline, almost imperceptibly.
This means the danger of assembling these electoral blocs is that you discover that they’re blocks of ice and have melted as you handle them, and so are rather smaller than you hoped. When you’re right on the margins of victory, perhaps the question to ask is not, “how do I hold onto the marginal one per cent“, but “how do I find the next ten per cent?”
What it doesn’t mean:
It doesn’t mean the next election is already won.
Right now, Labour has a six point lead over the Tories. That’s almost exactly how much Labour’s lead has declined over in the last 12 months. This doesn’t mean the same will happen this year. It doesn’t mean doom and gloom and tearing your hair out. The Tories haven’t picked up a lot, and that is keeping Labour ahead (if the Tories were scoring as they did in Autumn 2010, or Winter 2011, the polls would be suggesting a popular vote tie)
However, it should be a reminder to Labour strategists that this could happen again this year. If the Tories pick up a point or two from disillusioned 2010 Tory voters. If Labour’s proportion of 2010 voters who won’t vote increases a touch, the number of 2010 LibDems falls from c34 to c30%, or if Labour’s non voters return to apathy, then suddenly, instead of a comfortable lead, Labour will be an effective tie at the end of 2014. None of these things are earth shattering changes in public opinion. None of them are predicated on a huge shift in electoral blocs, or a major surge by any party, just a little loss at the fringes.
There are two ways to respond to this, I think. One is to spend your time worrying about how to hold on to these fringes. To think every day about how to keep hold of everyone who now says they’ll vote Labour. the other path is to consider what it would take to expand the range of people who would vote Labour, and let the fringes take care of themselves.
Instinctively, I prefer the latter approach. That’s just me though. But, as I said before, I am usually right, and everyone should do as I say.