How big is Labour’s lead?

Keir Starmer polls better than Rishi Sunak. But not 20 points better

July 11, 2023
Strength to strength? Keir Starmer campaigning with Keir Mather, Labour candidate ahead of the Selby byelection this month. Image: PA Images / Alamy Stock Photo
Strength to strength? Keir Starmer campaigning with Keir Mather, Labour candidate ahead of the Selby byelection this month. Image: PA Images / Alamy Stock Photo

To adapt the famous aphorism about Donald Trump, opinion poll stories should be taken seriously but not always literally. Polls are, by a long way, the best way to measure the public mood, but every number they generate is subject to a margin of error. Care should be taken not to dramatise small shifts, cherry-pick figures that suit the politician or journalist commenting on them, or overinterpret results that depend on the precise words used in the questions asked.

Bearing all that in mind, how do things stand as the next general election approaches? Plainly, Labour is well ahead. But how far ahead? Is its lead growing? If it is, why? And can it sustain its lead until the next election? In what follows, I have combined the data from different polls, to iron out fluctuations due to sampling error. If we just looked at individual polls, we could not be certain whether Labour’s in the past few days has been going up (Redfield & Wilton, Omnisis, YouGov, Techne) or down (Deltapoll, Opinium).

Here goes. It does look as if Labour, whose lead had been gradually contracting in the winter and spring, has crept back up. Here are the figures from the five companies that conduct weekly polls, during the two months when Labour’s lead seems to have been rising. (Using the same five companies for the before-and-after calculation means that we are comparing like with like.) 

A five-point increase in Labour’s lead is politically significant. However, look at its two components: Labour’s share up three, Conservatives’ down two. If those were the figures from a single company, we could not safely conclude that anything was happening other than sampling fluctuation. By averaging the results of five companies which together interview almost 10,000 people each week, there is a very high chance that Labour’s lead really has risen—but not absolute certainty. 

Assuming that Labour’s lead has risen, why has it done so?

Deltapoll and Redfield & Wilton track Rishi Sunak’s and Keir Starmer’s ratings each week. They tell similar stories. The opposition leader outperforms the prime minister, and his rating is holding up. In late April/early May, Deltapoll gave him an averaging a net rating of plus four (well 45 per cent, badly 41 per cent). Since late June it has averaged plus three (45-42 per cent). 

Wording its question slightly differently, Redfield & Wilton also indicates stability: plus five in three polls in late April/early May (satisfied 40 per cent, dissatisfied 35 per cent), plus seven in three polls since late June (40-33 per cent).

The action has been in Sunak’s ratings. Both companies show them down. Deltapoll: late April/early May: minus 10 (well 41 per cent, badly 51 per cent); since late June: minus 20 (35 to 56 per cent); Redfield & Wilton: late April/early May: minus six (satisfied 34 per cent, badly 40 per cent); since late June: minus 14 (29 to 43 per cent).

All of which looks good for Labour. However, three notes of caution. First, although the Conservatives are as unpopular as they were in the run-up to Tony Blair’s landslide in 1997, Labour and its leader nowadays are not doing anything like as well as they did then. The Tories are hated far more than Labour is loved.

Second, both Deltapoll and Redfield & Wilton ask an alternative election-choice question to voting intention, and this suggests a closer race than the headline numbers.

Each week, Deltapoll asks: Putting aside any support for a political party you may have, which of the following do you think would be best for the British economy? Its latest three polls give average figures of:

A Conservative government led by Rishi Sunak with Jeremy Hunt as Chancellor of the Exchequer: 32 per cent

A Labour government led by Keir Starmer with Rachel Reeves as Chancellor of the Exchequer: 44 per cent

Don’t know: 24 per cent

Redfield & Wilton asks: At this moment, which of the following individuals do you think would be the better Prime Minister for the United Kingdom?  

Keir Starmer: 42 per cent

Rishi Sunak: 32 per cent

Don’t know: 26 per cent

So, leads here of 10-12 per cent, not the 22 per cent indicated by voting intention polls. This suggests that when voters shift from the current mood of late-mid-term protest (which may well cost the Tories dear in next week’s byelections), to one of choice of government, Labour’s lead may well shrink. In particular I would expect most of the current “don’t knows” who voted Conservative last time to revert to the Tories when the election comes.

Third note of caution: as I have argued before, Britain’s electoral geography is tilted against Labour these days. Since I wrote that analysis, Labour’s hopes of gaining seats in Scotland have risen; and May’s local elections suggest that anti-Tory tactical voting is back. Taking everything into account, Labour may need a lead of 8-10 per cent in the Britain-wide popular vote to secure an overall majority in the next House of Commons.

So: prediction time. I currently expect Labour to end up with 40-44 per cent of the vote, while the Tories win 31-35 per cent. If the result falls within those ranges, Labour will be the largest party, and has a 50-50 chance of securing an overall majority. However, these should be regarded as medium-term economic forecasts, subject to revision—and just as likely to be as wrong as the Bank of England’s recent inflation forecasts.

Like economic forecasts, I shall be updating my predictions as the election draws near. I also hope to narrow the range of outcomes for each party towards the end of the campaign. If by some chance one of my forecasts between now and polling day proves right, that is of course the one I shall remind readers of afterwards.