I really don’t like “does X make you more or less likely to vote” questions. On policy questions they are particularly abhorrent as they are used to try and measure salience – or more often, to try and deliberately overstate the salience of an issue. It’s inevitably some pressure group, the campaign for ponies or whatever, asking a question saying “If party X offered you a pony would you be more likely to vote for them?”. Everyone says yes, as they like free ponies, and you end up with a press release saying that 70% of people will change their vote based on the issue of pony-ownership, and ponies are going to be the key election deciding issue. Sigh.

When it comes to questions about party leadership the question isn’t that bad (it’s one of the few contexts where I use that question structure myself) because it isn’t normally being used to gauge the salience of party leadership, it’s normally being used to see if a leader or potential leader has a positive or negative impact.

There is nothing wrong with a question asking if Jeremy Corbyn has made people more or less likely to vote Labour… but you need to be careful with interpretation. The overall figures will include a big chunk of people saying Corbyn makes them less likely to vote Labour who would never have voted Labour anyway, and a big chunk of existing Labour voters saying Corbyn makes them more likely to vote Labour (it’s why when YouGov ask the question we try to encourage those people to say no difference by offering opinions of “No difference, I was/wasn’t going to vote X anyway”).

The interesting bits are people who DID vote Labour last time, and say they are less likely, and people who voted for parties other than Labour last time and say they are more likely. But even then “more or less likely” is not a particularly high criteria to meet – “less likely” is a long way short of “not going to”. So a headline like the Independent’s today saying “Corbyn loses fifth of Labour voters”, based on 20% of Labour voters agreeing with a statement that with Corbyn as leader they are more likely to vote Tory is over-egging it. Those voters aren’t necessarily lost, they may still vote Labour tomorrow, their likelihood of Labour has just dropped to some degree. We’ll have to wait for some voting intention polls to see if there has been any substantial net damage to Labour’s support.

81 Responses to “Interpreting more or less likely to vote questions”

1 2
  1. YouGov in Sunday Times – Con 39 – Lab 31 – UKIP -16 – LD 6

  2. UKIP’s VI seems to be hlding up well in all these polls. Is this despite or because of the leadership’s relative silence recently, I wonder?

  3. @Pete B

    Probably despite, and probably helped by the migrant/refugee situation. Almost every European country now has a UKIP-style party polling 10% plus so it is highly unlikely that they will just fade back into their pre-2012 obscurity.

  4. @Pete B

    Most of the public don’t notice pronouncements or lack of, from politicians.

    UKIPs’ share has been, under normal circumstances, between 10%&15% for the last ~3 years.

  5. Meanwhile, if polling for our election was tight Canada’s is something else – https://en.wikipedia.org/wiki/Opinion_polling_in_the_Canadian_federal_election,_2015. Tories, Liberals and NDP all polling about 30%.

  6. As is the one for tomorrow’s Greek election:


    Actually there are two different sets of tightness involved. Firstly which of SYRIZA and ND will come first and so get the winner’s bonus of 50 seats in the 300 seat parliament. Secondly will a number of smaller Parties (the junior coalition Party, ANEL; the SYRIZA rebels, LAE; the centrist EK) will clear the 3% barrier and so get seats. If they all do that will mean nine Parties represented (from seven). In any case it looks as if forming a government will be difficult.

  7. Jack Sheldon and Wood
    Thanks for the replies. I take your point that UKIP now seem to have reached a level which will not suddenly collapse. As Farage is such a ‘marmite’ character, I just wondered if it’s actually better if he keeps his mouth shut, as he has already peresuaded everyone he’s going to.

  8. @Roger Mexico

    Perhaps ANEL, who have an unfortunate acronym, will squeak in through the back door.

    Looks to me as if whichever party wins – and therefore gets the seat bonus – will have to form a minority or, as before, a coalition with a much smaller partner. What odds on yet another election in 2016?

  9. @pete b

    I reckon UKIP would have better prospects if Farage hadn’t “unresigned”. Suzanne Evans for example might be better as leader. Farage keeps UKIP’s ceiling lower than it otherwise would be.

    You would expect UKIP’s VI to be boosted at the moment due to the migrant crisis, as the issue of immigration is always in the news. I think it will rise even further as we get closer to the referendum, because whatever the result of Cameron’s renegotiations, many won’t be satisfied.

  10. Faisal Islam points out –

    “Osborne & Corbyn have almost identical favourable/ unfavourable ratings”. :-)

  11. Polls now are pointless. The country will look radically different after 5 years of Tory rule. Norman Lamb is predicting the NHS will collapse and even Tory councils are saying they can’t cut anymore. My council has implemented a cut of a third in their budget already and soon won’t have enough to fund statutory requirements. Plus tax credit cuts haven’t been felt yet. Also junior doctors are not happy and could go on strike.

    Either the Tories will have to ease off planned austerity a la 2012-13 to avoid a collapse in public services, or public services will be so bad. Either way the Tories will be in trouble by 2020.

  12. @bigd

    To be fair, that’s what they said in 2010. I remember when commentators were saying that 2010 was a good election to lose because the measures to get the country back on track (whether from Tories, Labour, whoever) would make the 2010 winner so unpopular that they’d suffer a huge loss in 2015.

  13. @Omnishambles

    It is also fair to say that the longer a Government has been in office the less forgiving the electorate is likely to be!

  14. BigD

    “Polls now are pointless.”

    Only if you think they are a guide to how people will vote in the UK GE in 2020. Since that isn’t what polls do – they just provide a snapshot of current opinion – your statement seems a tad hyperbolic.

    Your examples of local government (in England) and NHS (England) facing imminent disaster apply, of course, to only one of the three GB political systems.

    Oddly, the good folk (and doubtless the bad ones as well) of England seem to be quite happy with these states of imminent collapse of their services – if we judge by the polls.

    Quite useful things polls. ;-)

  15. @graham

    Well yes, but it also depends how credible the opposition are judged to be.

    The electorate weren’t too happy with Labour by 2005, but the Tories weren’t credible enough to replace them

  16. omnishambles @graham

    “Well yes, but it also depends how credible the opposition are judged to be.”


    Large numbers of voters aren’t that fussed about which particular set of career politicians make the decisions. They just want their own interests to be protected.

    The trick that successful politicians succeed with is to persuade enough of the electorate that they are the ones to do that – even though some of their policy objectives may not be exactly to the voter’s taste..

  17. Useful stuff from the British Election Study – but, sadly, initial assumptions that gained traction in the immediate aftermath, will doubtless continue to dominate.


    Or, as the BBC put it. “During the campaign every leader but the SNP’s Nicola Sturgeon proved unpopular, but David Cameron was less disliked than Miliband

    Perversely, when Labour gained votes from former Lib Dems they helped the Tories win seats.

    In total Labour votes won from Lib Dems alone secured seven seats for the Tories.”

    In light of Corbyn’s election, it might be worth focussing on two “red herrings” that the BES identified. –

    A pair of red herrings: the SNP, and left–right ideology

    There are two oft-cited reasons for Labour’s failure that we do not address here, because we currently think those explanations might be red herrings, at least in the main.

    The first is the ‘SNP threat’. As discussed above, we currently find little robust evidence that attitudes towards the SNP and expectations about a hung parliament resulted in gains for the Conservatives from Ukip or in vote losses for Labour from former Lib Dems. We cannot say for sure that this didn’t matter, but our explorations suggest it is a difficult effect to pin down. We find much clearer and more robust evidence that perceptions that there was going to be a hung parliament enhanced votes for ‘challenger parties’ overall: Ukip, Plaid Cymru, the Greens, the Lib Dems and the SNP. This might have cost Labour votes, but not in the ways people may assume.

    The second red herring is Labour’s left–right position – that is, the question of whether Labour was either overly or insufficiently left-wing.30 Generally, our data shows that people were more likely to vote Labour in 2015 when they thought the party was more left-wing, and less likely to vote Labour when they thought it was centrist. Figure 8 shows the ‘predicted probabilities’ of voting Labour based on a statistical model of BES respondents’ positioning of Labour on a left–right scale. The respondents who were most likely to vote Labour were those who saw Labour as sitting on the left of the political spectrum, with the absolute peak coming in at around 2 on this 0–10 scale – slightly to the left of where respondents placed the party on average (the vertical line). This suggests there is very little to the argument that Labour was too left-wing to attract voters. At the same time there is not much to support the argument that Labour was not left-wing enough. There was very little difference in the likelihood of voting Labour between someone who thought Labour sat at the left-most end of the scale (0) and someone who saw it as just left of centre (4) – it is only when people saw Labour as sitting to the right of this point that support really drops off.

  18. “The respondents who were most likely to vote Labour were those who saw Labour as sitting on the left of the political spectrum, with the absolute peak coming in at around 2 on this 0–10 scale – slightly to the left of where respondents placed the party on average (the vertical line).”

    I don’t see that that supports the case that Labour wasn’t too left-wing. If Labour essentially boiled off all supporters who don’t identify as left-wing, and there aren’t enough left-wing voters to win an election on their own, then Labour’s problem was absolutely its low appeal to people who were not left-wing.

  19. @ Onmishambles – but Osborne changed course in 2012. If he hadn’t it would have been a different story. Remember they promised to eliminate the deficit by 2015. If they had stuck to that the economy would have tanked! The truth is public services could just about cope with the 10-15 cuts. We’re now in a different scenario. You can’t keep cutting forever. When even Tory councils are predicting doom you know something is about to go horribly wrong!

  20. Also does anyone know what turnout models are used? If Corbyn can attract non-voters, surely that won’t show up in the polls because of the turnout filter? I suspect the Ashcroft underestimation of the SNP vote in May was because it failed to predict the spike in turnout in Scotland.

  21. Omnishambles

    It takes a while for cuts to affect quality as services can run on fumes for a while.

    I am also sure you can see that cutting a further x% and saying that is the same as the first x% is illogical. Think of that logic applied to your wage packet.

  22. @ Hawthorn – I totally agree!

  23. Regarding ‘The Mail on Sunday’.

    I always infinity was a theoretical mathematical construct.

    In fact, it can actually be calibrated by the depth of the bottom of the barrel the MoS find their stories from.

  24. Correction

    Regarding ‘The Mail on Sunday’.

    I always thought infinity was a theoretical mathematical construct.

    In fact, it can actually be calibrated by the depth of the bottom of the barrel the MoS find their stories from.

  25. I’m not sure which Mail article you’re referring to but your post made me curious so I went and looked online. I must say what a complete load of crap! An article about Sadiq Khan that’s headlined “Labour’s ‘Mayor’ suggests Corbyn is unfit to be PM”. Did he say that? Having looked at the article I can’t find a quote that supports the headline. All he said was he should have sung the national anthem. I totally despise the Mail and anyone who reads it!

  26. Your examples of local government (in England) and NHS (England) facing imminent disaster apply, of course, to only one of the three GB political systems.

    It applies in Scotland too. Edinburgh council are facing cuts of at least £105M this year, with more extreme cuts to come after the 2016 election. Thousands of jobs are under threat & the unions have not ruled out strike action.

  27. I was referring to the stuff about JC’s marriage 35 years ago.

    However, the story you refer to ticks the sames boxes.

    They do that sort of thing from time to time….

  28. Well, any potential cuts in Scottish NHS Could be reduced by not spending 2 million pounds on homeopathy.

  29. BigD – good question. The short answer regarding todays polls is that YouGov don’t currently use a turnout filter at all away from elections, ComRes use their new demographic model of turnout, based on people’s demographics.

    It raises a good question of how we deal with the problems of What Went Wrong. The polls at the election overstated turnout and didn’t correctly reflect the difference in turnout between socio-economic groups. One way of addressing is what ComRes have done – include socio-economic factors in your turnout model

    The problem of this is what you raise – if the pattern of turnout changes, you’d get it wrong. In theory an alternative approach is better, sorting out the sample so it is more representative in terms of turnout without forcing a socio-economic model upon it. Except, what if you can’t? What if there isn’t some simple alternative approach that does that? You may end up taking an approach like ComRes have done not because it is without flaws… but because there is no better alternative, and because in practice the socio-economic pattern of turnout is pretty constant (sure, the precise proportions change… but turnout is always lower among the young and poor than the old and rich).

    Behind closed doors there are lots of hard methodological decisions being made.

  30. ICM poll of marginal seats in today’s Sun on Sunday shows a Con to Lab swing of 2.1% – which would imply Labour gaining 17 Tory seats with Labour on 249 and Tories 313. The Tories would be struggling to survive as a minority Governmwnt.

  31. @ Anthony – thanks for the reply. I only raise it because the media have highlighted anecdotal evidence of people saying they’ve never voted but would vote for Corbyn. Is there anyway of using election results to work out what proportion of each socio-economic group vote? I know the marked register is released after an election. Could a sample be looked at to try and work out what proportion of various groups voted? The marked register from the upcoming London Mayoral election will be quite useful. If suddenly you get higher turnouts in poorer areas that might suggest Corbyn has managed to get these voters to actually vote!

1 2