Not long left?

So, where are we? With the latest UK polling report average suggesting Labour support is at 38 per cent, are things going well or badly for Labour?

Some, from both left and right, argue that it is impossible for the Tories to win the next election.

On the other hand, there is academic research which suggests that after the 2012 local elections the Conservatives were likely to be marginally ahead in the popular vote in 2015, but that analysis was before the rise of UKIP in the 2013 local elections.

Leo Barasi has also produced an analysis that suggests that the polls in May 2013 project to an effective tie between Labour and the Tories at the next election, though, as Leo acknowledges, it’s weakly predictive.

Can we look more deeply at previous elections and draw out some trends? Let’s try. (I should state my gratitude to Mark Pack for compiling a dataset of post war opinion polls, which has made this analysis possible)

So what can we say?

First, every election winning opposition party at this point in the cycle had a higher share of the vote than Labour currently enjoys.

Second, the Conservative share of vote, while not impressive, is above that some re-elected governments had at this point in the cycle.

(However, there have been changes in polling methodology which might tend to increase the mid-term share of vote of unpopular governments)

Third, in almost every case, the opposition share of vote has declined between this point and the general election.

The only exceptions have been when the opposition was performing poorly in mid-term and went down to heavy defeat (1955, 2001).

In recent years, the decline has been more significant among successful oppositions.

How big is this decline, and when might we expect to see it?

Well, in recent elections, the decline for successful oppositions has been between an eighth and a fifth of their share at the current point, or around seven share points.

Unsuccessful oppositions vary more, but aside from the perennially unpopular 97-05 Tories, also show a significant decline.

So, to be rather conservative about it, you might anticipate a decline of three to five share points for Labour.

Naturally this isn’t the whole story. There is no inevitability in politics.

In almost every case, the opposition enjoyed an increase in their share of the vote at some point during the 22 months before the election, despite eventually losing share.

So, while the Conservative party lost just over six points over the 22 months before the 1979 General Election, going from 51% to 44.9%, they also went as high as 55%.

Even the 1997-2001 Tories, in polling terms perhaps the least impressive opposition of the post war period, briefly reached the heights of 40% nine months before the 2001 Election (this was at the height of the fuel protests).

You can see this expressed here: this chart shows the opposition polling position 22 months out, at their polling peak before the election, and the final result.

The opposition peaks are usually achieved between 22 and six months before the election.

On the other hand opposition polling ‘troughs’ are usually found in the election campaign itself, with ten of the fourteen oppositions seeing their lowest poll share in the three months before the General Election and eight of these getting their lowest poll share in the last month.1.

So what can we conclude from this?

Well, obviously it’s not great news that no successful opposition has won from having a share of vote in the high thirties/low forties at this stage in the cycle.

Further, if the next election follows the pattern of previous elections, you’d likely expect to see Labour’s polling support to fall.

However, this fall would only be clear in the run up to the election itself, and there will likely be a peak in Labour support before this, perhaps driven by external events (Fuel strikes, economics changes, government scandals etc).

If that peak comes close enough to the next election, and is significant enough, Labour might outperform these expectations2.

Further, there may be reasons to believe that the next election will not behave as those in the recent past have done.

Andrew Harrop has observed that the views of post-2010 Labour converts are to the left of even 2010 Labour voters. This suggest these converts might be ‘stickier’ to Labour than usual mid-term supporters of opposition parties3.

However, even given the likelihood of a future peak and hopeful Lib Dem ‘stickiness’, it’s best to be somewhat cautious.

If the past is even a close guide to the future, then you’d be fairly conservative in expecting to see Labour lose some three-five points between now and the next election.

That would put Labour roughly in the 33-35 bracket, with that level of support becoming only becoming clear in the election campaign itself.

I’ve deliberately talked little about the Conservatives current polling in this post, but absent a very low poll share by the Conservatives, such a projection does not suggest a straightforward victory.

This might lead you to conclude that Labour needs to either do an outstanding job in retaining current supporters, or find an extra element to our electoral coalition.

Or of course, we could hope for our polling peak to hit at the right moment!

One final comment: it bears repeating that just because things have tended to happen one way in the past does not mean they will happen this way again in the future, but it’s probably more likely to happen that way than the reverse.

For example, Mike Smithson is right to identify that Margaret Thatcher was rated as a less capable PM than Jim Callaghan in Ipsos-Mori’s polls as the 1979 election approached.

Yet it is equally true that this is the only election from 1979 to date in which the leader on ‘most Capable PM’ did not end up as the leader of the largest party.

Why was 1979 so unusual in that respect? Looking at the data, it seems likely that the events of the winter of discontent damaged the government so severely that Labour could not recover, even though Callaghan personal standing did recover somewhat.

Events matter, but it’s risky to rely on a winter of discontent for bucking electoral trends.

  1. The three recent opposition troughs before the election campaign are the Conservatives in mid-1977, Labour in mid-1982, and Labour shortly after the replacement of Margaret Thatcher []
  2. This for me is the best explanation of the 1979 result, where Labour was not polling particularly badly overall []
  3. I suspect this is true up to a point, but does not mean that post 2010 converts are ‘locked’ into Labour. We’ve already seen a decline in the proportion of 2010 Lib Dems saying they’ll vote Labour, for example []

8 Responses to “Not long left?”

  1. Dan Filson

    Just as a matter of history I don’t buy that in 1955 “the opposition was performing poorly in mid-term and went down to heavy defeat ” What happened was that the Tories replaced their aging leadership (Churchil) and Labour was slow to do so (Attlee) with the result that a still glamorous Eden was up against the aging Attlee and the prince–in-waiting less glamorous and more cerebral Gaitskell. My father fought a seat won by the Tories in 1950 and lost by 2,105 votes. Nationally the net loss of seats by Labour was 18 – hardly a heavy defeat.

    • hopisen

      Bevanite split, disinterested leader, no new policy offer, losing by-elections in Sunderland? 51-55 was a pretty weak opposition.

      • Dan Filson

        I concede that the leadership, not just Attlee, were either tired, old or split and this may or not have been reflected in by-election results. My main point was that 1955 was by no means a crushing defeat, just the steady build of a Tory majority upon the 1951 flimsy foundation. The housing building may have helped (take note: Ed M – you can show dynamism by getting PUBLIC housing built).

        In my father’s case being up against a decorated WW2 Battle of Britain ace didn’t help, but I think a London seat with a sitting MP only getting a Tory majority of 2,105 indicates it was hardly a runaway win. Note the 82% turnout, not bad eh!

  2. Stuart Ingham

    Are we dismissing Labour’s 70ish% retain rate of the 80’s as a product of pre-advanced ‘weighted’ polling?

    If so, do we not have a rather small sample size of elections where we trust the polling methods. One of which shows an opposition increasing its support and anther where it only lost a point of two.

    Or am I letting hope produce an unwarranted scepticism?

  3. adrian alexander

    surely we should be polling the marginal seats to see which way the public are moving on a regular basis.If you looked at Oldham east,it has moved from marginal to fairly strong labour by consistent doorknocking by the local party,which mike has mentioned from time to time.


Leave a Reply