Polling in the US election was a disaster – and the UK industry must learn lessons from it
Select committees of parliament have a deservedly high reputation. And naturally we in the Lords are particularly proud of ours. Week after week, we produce reports, firmly based on evidence sessions with expert witnesses, which have a real impact on public policy.
November’s report from the Communications and Digital Media committee on “The Future of Journalism” is one example, setting out detailed proposals as to how journalism can prosper and be made more diverse in the digital era.
Last week’s “Employment and COVID-19: time for a new deal” from the Economic Affairs committee sets out the measures necessary to achieve a sustainable recovery and to offer a positive future post-virus.
Until recently, however, the select committee procedure had a serious flaw. Out popped the report. Weeks or months later the government produced what was usually a dry-as-dust response. There might then follow a debate on the floor of the House in which sensible peers made sensible speeches. But that was it. There was no follow up, and the reports disappeared into that other-worldly Arcadia of the gone and forgotten good.
The Lords has now made a significant step to improving that process. Now, under the auspices of the Liaison Committee, chaired by Lord McFall, our Special Inquiry Committees – which cease to exist once they have published their reports – are holding follow-up sessions, to see what action has been taken and what more remains to be done.
The Committee on Opinion Polls, which I chaired, and which reported in 2018, has just held such sessions. It heard inter alia from Sir John Curtice, Britain’s iconic pollster who chairs the British Polling Council (BPC) Today 21st December the Liaison Committee published its findings.
Important recommendations of the committee are being taken forward by the BPC. Better training for journalists in opinion polls is one example. The BPC, as the committee recommended, is now a more proactive organisation. It hasn’t adopted some of our recommendations – regarding for example guidance on polling techniques. But it has progressed.
Our original report was followed by the 2019 British General election. That was a great day for opinion polls, with the result pretty well bang on. So naturally the pollsters lay relatively smug in their beds.
The alarm has now gone off. Polling on the American Presidential election got one thing right. Joe Biden won. But otherwise the results were a disaster for polling. Day after day, poll after poll, the data piled up: Biden was winning by a massive average eight per cent margin nationally.
No doubt the American polling industry will be conducting its own post-mortem. My own instinct is that the problem lies in non-respondents. These days a large majority of people who pollsters contact refuse to participate. It looks from America as if the non-respondents are not the same as the respondents and have different – in this case less anti-Trump views.
But having had a bad run before 2019 – wrong in 2015, wrong on the 2016 referendum, wrong on the 2017 general election – our pollsters can’t rely on their 2019 success as a get-out-of-jail card. After all a man with a pin would sometimes get the right results.
The new hearing gave us a chance to bring this crucial fact to the fore. The British polling industry needs to establish some formal and impartial investigations into what the American polling disaster implies for Britain.