As a teenager in the 1980s, I used to work for an opinion poll company, NOP Market Research, in their offices in Covent Garden. I think I was paid £3.00 an hour plus unlimited free coffee. It was challenging work. We were given sheets torn from telephone directories but while there was no shortage of people to ring, persuading them to take part was difficult. They were too busy. Or felt it was too intrusive. Perhaps around a tenth of the people I spoke to agreed to be surveyed. Even those who did take part often regretted it. I had to rattle through the questions as fast as possible in a pleading tone to prevent them from hanging up. (Often there would be batches of questions from different clients in the same survey – so I might suddenly switch from asking about politics to washing powder.)
To make it even more difficult, after a while quotas were reached for different categories. That meant we were not to interview any more of them to ensure a representative sample. This was before the woke era but you can still imagine the diplomatic minefield after a note was handed to me saying “no more women”, or “no more pensioners” or “no more C2s.” “Oh, so you are not interested in my opinions?” they would say after the demographic “screening” questions prompted the interview to be terminated. “No, I’m afraid we have spoken to enough women already. Would it be possible to speak to your husband please?” The receiver slammed down.
This traumatic early experience has left me sceptical of opinion polls ever since. I realise that in the internet era, the methods have changed but that just raises a different set of challenges.
Yet scepticism is different to denial. The polls usually tell us something if looked at in the right way. For instance, the polls on voting intention currently show a huge Labour lead. When you get stuck into the data they usually show that is not mostly due to Conservatives switching to Labour. More Conservatives have switched to Reform UK (who have hit 13 per cent in the latest YouGov poll.)
But an even bigger loss of support from the Conservatives is to those intending to abstain. It would be wrong to assume such people are apathetic. They are mad as hornets. Often they have strong Conservative beliefs and feel angry that the Conservative Party is failing to uphold them.
This cohort is hard for the pollsters to pick up. The reason is that often in their disgruntled state they will be disinclined to take part in the survey at all – whether it is online or over the phone. The YouGov poll I mentioned above does include those who said they were “certain not to vote”. Such people were three times more likely to have voted Conservative last time as voted Labour. But I suspect a great swathe of Conservative-minded abstainers are being missed.
By-elections may offer a better guide. On October 19th, Labour gained two hitherto safe Conservative seats – Tamworth and Mid Bedfordshire. But the reason was that Tories stayed at home rather than switch to Labour. While the Conservative vote collapsed, Labour’s stayed about the same. In Tamworth, it was up just 811 on the 2019 General Election. In Mid Beds, it was actually down 156 on when Jeremy Corbyn was the Labour Leader.
By contrast, in the run-up to the 1997 election when Labour won a landslide victory under Tony Blair, voters switched over to Labour. There was positive enthusiasm. A couple of roughly equivalent by-elections show the difference. In the 1997 Wirral South by-election, turnout was 71.5 per cent, Labour’s candidate got 5,360 more votes than the previous General Election. In the 1996 South East Staffordshire by-election the turnout was 62 per cent, Labour gained 4,167 votes on the time before.
Polling on specific issues, rather than voting intention, is even more tricky. My recollection is that those who really are “don’t knows” often like to give an answer anyway. They will try to pick whatever sounds safe and reasonable. This might mean backing the status quo. “Should we do x or leave things as they are?” will mean that the “don’t knows” tend to back leaving things as they are. But the “don’t knows” also like a middle option. “Should we abolish X, suspend X for a trial period, or leave things as they are.” With that one, you might get the undecideds to switch from the status quo to the reasonable-sounding trial period.
Another constraint is that an initial view might be picked up but then, were the matter to be pursued, the opinion might change. Often we get polls saying that the British would like to rejoin the European Union. But when Omnisis ask about this, they then ask if people would still want to join if there was a requirement to join the Euro as our currency instead of the Pound. That showed most people would not wish to rejoin on those terms. Yet that is realistically the choice that would be on offer.
Then there is ensuring the question is impartially worded. There was a poll from Global Counsel recently which was cited as showing support for higher public spending rather than tax cuts. But the options were: “The Government should prioritise spending on schools and hospitals.” Or: “The Government should prioritise tax cuts.” What if it had asked about spending on Quangos/civil servants/Overseas Aid/HS2/Mickey Mouse college courses/welfare payments to asylum seekers, etc, etc. What if it had asked about the increase in spending on the NHS and asked if that had produced higher standards? Those would have been skewed the other way.
A neutral way would have been to ask about public spending v tax cuts without highlighting popular or unpopular bits of public spending or assumptions about whether or not it be spent effectively.
Winston Churchill was dubious about opinion polls warning:
“The leader who constantly keeps his ear to the ground is hardly in a position to be looked up to by his constituents if he is detected in that somewhat ungainly posture.”
There is certainly much to be said for leading rather than following public opinion. Having a gap of five years between General Elections should allow time to carry out measures that might be unpopular in the short term, if a Government is confident it will be vindicated by the results.
Rishi Sunak does not have much time and as the General Election gets closer, the fixation with the polling is sure to become ever more intense. In a democracy, you can hardly blame politicians for such curiosity. The risk is for the Government to be misled as to the reasons for its unpopularity. Those who voted Conservative in the past are largely still Conservative. But to their dismay they have concluded that the Conservative Party no longer is – thus ceasing to merit their support.