Do you think online polls are rigged?

What’s your reaction to an online poll? Do you think they’re the honest and unfiltered voice of the people, whether on politics, pet competitions or favoured hairstyles, biscuits and bands? Or do you immediately wonder how rigged the vote might have been?
While upvoting a Jaffa cake over a Hob Nob might properly be viewed as just a bit of fun, online political polls have been raising other questions, about reliability, manipulability, and therefore, influence, ever since they dramatically entered the voting scene around 2012.
Yet they did have an interval when none of these things seemed immediately relevant. When they were first introduced as a mainstream polling tool during the 2012 presidential election in the US, online voter polls were seen as a significant breakthrough, full of promise for more accurate assessments of voter preferences.
While this sounds laughable now, the initial enthusiasm was based on solid results. Like Obama, the polls crossed the finish line. Online political polls turned out to be accurate, correctly pegging Obama as the victor over Mitt Romney. They often outperformed more traditional polls based on carefully selecting a representative group to poll, then applying a secret sauce of weightings and adjustments.
In an interesting newsletter recently, New York Times political analyst Nate Cohn highlighted this forgotten glory moment for online polls, when it appeared that even measuring the voting intentions of Xbox users could produce polling gold.
“The most prominent online pollsters — Google Consumer Surveys, Reuters/Ipsos and YouGov — all produced good or excellent results. With the right statistical adjustments, even a poll of Xbox users fared well,” he wrote. “These successes seemed to herald the dawn of a new era of public opinion research, one in which pollsters could produce accurate surveys cheaply, by marrying online polls with big data and advanced statistical techniques.”
That dawn did not come and you can probably guess why. While it at first seemed that an online poll would produce a broader and more democratic, population-representative selection of voters — which initially, it did — the internet soon began to internet. As Cohn notes, “just about everything” became more difficult.
Now, only British-based YouGov still produces solid online surveys, he says, which he credits to long experience and a solid model for selecting panels and weighting responses. Few of the original online pollsters are still around, though many new ones have sprang up, with generally mediocre polling accuracy.
Back in 2012, most people accessed online polls through a desktop PC and browser, by visiting a website. “A banner ad on the right websites could plausibly reach a huge — and broadly representative — swath of the population,” to construct a poll panel, he says.
Now people use apps and mobile devices and it’s harder to confirm identities. Cohn notes that online audiences are also now highly segmented and different social media audiences tend to produce different results.
I think it’s also hard to recall now how different the online world was in 2012, and that’s if someone was active there at all. In the US, a fourth of the population — 24 per cent — still wasn’t online (in Ireland, some 80 per cent were online in 2012, according to Statista figures). Now, it’s nearly 100 per cent.
And while the majority of people in both countries were using social media, it wasn’t to the extent we do now. In 2012, people globally spent 90 minutes a day using social networks. By last year, that had jumped 67 per cent to 151 minutes. That’s a lot of scrolling.
So we’re all more exposed to increasingly sophisticated manipulation through posts, “dark” political adverts that only we see and therefore any claims are not easily open to public challenge, and networks of fake accounts and bots that can make an issue, opinion or poll appear to be more popular (or less so) than it is.
And with ever more people opting to get their news only from TikTok or YouTube videos or social media posts, users themselves become the only filters and fact-checkers to a torrent of claims and counterclaims, fake posts, doctored videos, and other forms of deliberate misinformation.
Repeated studies indicate that younger people and more conservative audiences are less likely to view any form of mainstream reporting or debate, instead “doing their own research”, relying on peer groups and algorithm-driven online sources, which tend to reinforce rather than challenge or deconstruct.
In addition to the difficulties of getting accurate information on which to base an opinion — or a candidate choice — prospective voters also are now exposed to everything from widely circulating conspiracy theories to rogue foreign-state campaigns to sway voters.
Online polls are affected by all of this, too. Pollsters also have to deal with “bogus respondents”, bot accounts, widespread difficulty in verifying people’s identities, and many other reasons for low-quality, unreliable voter data. That’s when they bother — Cohn says many give little detail on data sources or their weightings. Many of us, rightly or wrongly, question the accuracy of modern polls. But with online polls, scepticism is justified. The online poll is the one almost certain to be wearing the data dunce cap.