Playing the percentages: Steve Coronella surveys political opinion polls

It seems you can’t listen to a news report, access an online journal, or dive into a social media post these days without being told the latest mesmerising findings of an opinion poll commissioned by a reputable news agency from a research market specialist.

No topic is left unexamined – in numerical terms anyway. Our sexual preferences, dietary and travel habits, and political inclinations are routinely given statistical values. Then the pundits are off and racing toward some dubious conclusion about society as a whole. Newspaper editors and TV producers love this type of easy reporting, packaged with glitzy graphics and ready for mass consumption, and such polls regularly make the headlines of the Sunday papers and weekend news broadcasts.

Some polling companies have been notoriously off the mark in recent times – the passing of Brexit and Donald Trump’s presidential triumph being the most notable blown calls – but public opinion surveys remain an essential part of any election or referendum campaign. And they can inadvertently sway democratic decision-making. For example, as University of Warwick economics researcher Eleonora Alabrese discovered in a recent study: “In broad terms, if national polls predict a landslide election, turnout tends to be lower. Turnout is also generally lower in safe seats.”

How do opinion polls work and can they be trusted?

Netflix, Prime Video, Disney+, Apple TV+: 10 of the best new shows to watch in June

---

We no longer try to understand why Real Madrid win the Champions League final - we just accept

---

The Irish philosopher who was father of ‘cathedral thinking’

---

100 great restaurants and places to eat around Ireland this summer

---

The first known political opinion poll in the US took place during the 1824 presidential election, when voters in Delaware and North Carolina recorded a preference for Andrew Jackson over John Quincy Adams, as did voters generally. Owing to the presence of a four-man field, however, Jackson failed to secure an electoral college majority. So in accordance with the terms of the 12th amendment, this notoriously contentious election was finally decided by the US House of Representatives – a scenario some present-day Republicans would love to duplicate in 2024 given their recent high jinks surrounding the election of Speaker Kevin McCarthy.

Of course, no matter which way the political winds are blowing, all polls are dependent on a sound methodology. In 1936, a survey of 2.3 million voters conducted by the Literary Digest – an influential general interest magazine published by Funk & Wagnalls – indicated that Alf Landon would defeat Franklin Roosevelt in the presidential election. Roosevelt was in fact re-elected by a landslide and had two more successful runs at the White House, both correctly forecast by market research expert Elmo Roper.

An investigation conducted after Landon’s failed candidacy by George Gallup, another opinion poll pioneer, attributed the polling discrepancy to “participation bias”. The Literary Digest’s survey involved the mass mailing of postcards to a target audience more inclined to vote for the Republican Landon and more likely to return their unofficial ballot. (Sounds like an electoral approach Donald Trump would support.)

Gallup’s own polling during the 1936 election involved a far smaller, more scientifically-based pool of respondents and his findings predicted the actual outcome.

You can believe me when I say that the accuracy of opinion polls also depends on people answering honestly. When respondents are less than truthful – or plain inaccurate – in describing their views on a certain topic, this produces “response bias”, which can be overtly whimsical to frustrate pollsters, or slyly malicious in order to advance a position more extreme than a respondent might actually hold.

Response bias can also work in another direction – to help disguise a person’s true preferences altogether – and can produce the “Bradley Effect”. This is named after Los Angeles mayor Tom Bradley, an African-American who lost the 1982 race for California governor to a white candidate despite being ahead in the polls leading up to the election.

It’s surmised that rather than reveal their preference for a white candidate, many voters registered as undecided or indeed in Bradley’s favour when polled – resulting in another sociological phenomenon known as “the spiral of silence”.

Though it sounds like a contemplative chamber where Superman might go to de-stress, this high-falutin academic theory, proposed by German political scientist Elisabeth Noelle-Neumann, states in a nutshell that people are inclined to keep their beliefs to themselves until they know what other folks are thinking.

Of course, no discussion of opinion polling – or indeed everyday journalism – would be complete without bringing in the inevitable margin-of-error issue. This is no doubt an important element of opinion polling, but frankly, after I’d looked up a few articles on the subject, the arcane mathematical symbols and formulas had my head swimming.

Suffice to say, as with most things in life, it’s wise to take any opinion poll survey with a pinch of salt.