Pick a number between six and 36.
Chances are one of the four main UK parties will have hit that number in an opinion poll during November. Only 12, 20-22, 25-26 and 28 have yet to be taken. But there’s still plenty of time before the month is out.
Such unpredictability in the polls used to be rare in British politics. This year it’s become the norm. Labour, the Conservatives, Ukip and the Lib Dems have all bounced around the percentages throughout 2014. We’ve seen big variations often in the space of 24 hours – and sometimes less.
On Monday three polls were published which placed Labour variously on 30 per cent, 32 per cent and 36 per cent, while the Tories were put on 29 per cent, 33 per cent and 35 per cent.
You can find every opinion poll published since August 1970 here.
It’s understandably hard to draw meaningful conclusions from such clusters of data. But place them in a broader context and some intriguing patterns emerge. I’ve gone back through all the polls of 2014, and recorded each party’s monthly “ceiling” and “floor”: the highest and lowest rating they have scored.
It’s not an overall measure of performance, but it does give a sense of a party’s consistency: be it holding steady, gaining ground, or sinking downwards.
I’ll start with Labour.
Labour’s poll ratings ranged across a gap of nine points in January. At the time of writing the gap for November is seven points. As such the party’s performance in the polls is more consistent now than at the start of 2014 – albeit spanning a lower range of numbers.
By this measure, Labour’s most consistent spells in the polls occurred in February and April. In both these months the party shuttled around a gap of only five points.
Labour was sometimes touching the low 30s right back at the start of the year.
February is also the month when Labour enjoyed its highest poll “floor”: 36 per cent. A mark of how the party’s fortunes have changed is that 36 per cent is currently Labour’s “ceiling” for November.
Nevertheless, it’s useful to be reminded that Labour was sometimes touching the low 30s right back at the start of the year. Slumps in the polls have punctuated the party’s ratings throughout 2014. But the graph also shows how Labour’s floor held steady in the low 30s from May right through to October. It didn’t sink below the symbolic figure of 30 per cent until November.
Since the start of this month, 29 per cent has been touched in only three opinion polls, but that’s been enough to completely overshadow the two times Labour has hit 36 per cent and the four times it has hit 35 per cent: both of which feel like thoroughly respectable scores in a four-party system.
Now look at the Conservative highs and lows for the same period:
The party has had a strikingly consistent ceiling, but one that has never pushed higher than 36 per cent. Its floor has tended to sink slowly downwards, however. The low point of 25 per cent in June was followed by a slight recovery, but not – as yet – to the low of 30 per cent seen in January.
June was the month of the Tories’ greatest poll variation: a span of 10 points between its highest and lowest scores. The graph shows how this gap has in general got bigger as 2014 has unfolded. The smallest range was in January when the Tories kept their poll ratings within a span of five points. Since then only in April (six points) has the party come close to the same kind of consistency.
The Conservatives have been unable to shift their poll ceiling upwards.
The fact the Conservatives have been unable to shift their poll ceiling upwards at the same time as Labour’s has come downwards is one of the reasons the outcome of the general election feels so unclear. For the Tories to win a majority or even the most seats, it’s not enough for them to move ahead of Labour sporadically. They need to move ahead, stay ahead, and expand their lead. So far the polls do not show this happening.
Ukip’s ratings have been the most inconsistent of the lot.
Right from the start of 2014 the party has careered around the polls. It has never once maintained a span of numbers that’s come close to matching the minimum range of five points achieved by both Labour and the Conservatives.
The largest gap occurred in October, when Ukip scored as low as 13 per cent and as high as 25 per cent. The Tories will be thankful their all-time low of 25 per cent did not occur in the same month.
But Ukip has managed what neither Labour nor the Tories have been able to do this year: raise both its poll ceiling and floor.
So long as Ukip performs so inconsistently in the polls it’s hard to predict with confidence what impact the party will have in the general election – particularly as much of its support is so localised. An increasing number of poll ratings in the mid-20s won’t harm its chances, however.
For the Liberal Democrats, the polls have been so gloomy for so long that very rarely do their scores make headlines.
One exception came in June, when the party sank to 6 per cent for the first time this parliament. They will be relieved that 6 per cent has remained their floor ever since:
The Lib Dems have also been the most consistent when it comes to poll extremes, never venturing beyond a range of six points and in one month keeping it as tight as three. The party can take comfort at having such loyal supporters. It can’t take comfort from much else.
In this final graph I’ve brought together the poll ceilings of the four main parties:
Labour’s gentle decline contrasts with Ukip’s erratic peaks and troughs, while the Lib Dems continue to defy their critics and make it into double figures at least once every month of the year.
The Conservatives’ persistent inability to push into the high-30s is clear. If the party does ever take a permanent poll lead over Labour, it will probably be thanks to a Labour slump rather than a Tory surge. But only one point now separates the peaks for both parties. If the trend continues, the Conservatives could be in front just in time for Christmas.
Ian Jones runs ukgeneralelection.com. Follow his work for daily doses of number-crunching, news-spotting and result-pondering.