Election fraud or just inaccurate: the problem with political polling

By Isabel Cocker / Santiago Times Staff

Presidential candidate Sebástian Piñera caused quite a stir with his accusations of election fraud, ahead of the second round of the presidential elections in Chile. But the candidate himself benefited from the numbers the polls stated, giving him a lead that later turn out to be at least 7 percent less than expected. The debate on the effect of political polls remains, stronger than ever. 

In 2013, 3 out of 4 people suspected political polls of bias. Seeing as this was itself a poll, perhaps we may never know the true figure. However, after the inaccuracies seen in the recent Chilean election, in the 2017 UK general election, Brexit and the U.S. presidential election, among others, public perception of polls and surveys is bound to have worsened.

Yet, we still cite them, get buoyed up or worried by the results, following them avidly in the run up to political events. Journalists, myself included, love to give our articles a boost by mentioning Cadem or CEP and analysing the latest results. Even after proof that they fail, we still return to them for the next round of battle.

Part of the reason for this is the wish for stability and a view into the future. A presidential election is a big change for a country, and citizens wish to know as soon as possible what the results could be. By following polls, people can begin to plan for the next few years and the media can map out their own coverage of the election cycle. Elections are all about uncertainty, and anything that allies our own fears is bound to attract our attention.

Even after proof that they fail, we still return to them for the next round of battle

As citizens, we are nosy. We want to know what other people are thinking, how they are reacting, and whether our opinions fall in line with the rest of the population. Polls give us a window into the political point of view of people around us – a divisive issue not suitable for conversation with strangers. By following analysis of polls, we get the psychological boost of knowing that we are part of a crowd which, in turn, concretes our own opinions.

Polling companies and some scientists argue that the public mistrust polls because citizens tend to look at individual results rather than looking at overall trends measured through long-term polling. For example, in the U.S. election, polling suggested Hillary Clinton would win but it also showed her support declining massively over the election period. The subtext by the polling agencies is that although they mistook the exact percentages of voters who would be voting, they had been showing a surge in interest for Trump and a narrowing of the gap between him and Clinton. Therefore, on election day, Trump’s win may have been predicted by looking at the long-term shift in opinion of the American population, instead of analysing poll-by-poll.

However, it can also be contended that polls themselves have an indirect effect on the electorate. Blind faith in the accuracy of the figures can lull voters into a false sense of security or motivate the less-popular candidate to rally their votes. Undecided voters who see a large majority for the preferred candidate are more likely to use their vote either as a “protest” or to not vote at all – because they do not think that their vote will make any difference.

After Brexit, 11% of the “leave” voters later regretted their decision, according to yet another poll released in April this year. This suggests that many citizens voted against the established “remain” side not because they explicitly wanted to leave the E.U., but because they wanted to register a protest against a system they felt wasn’t working for them. Another part of the reason that the Brexit vote passed in favour of the “leave” side was an unexpected dip in participation by younger voters, those who would traditionally be “remainers”.

Both of these occurrences were caused partly by the overconfidence of the “Remain” camp, who saw constantly-positive opinion polls as proof of their imminent victory. Remainers seemed to accept victory almost before campaigning began and many rebelled against this, seeing it as arrogance and a continuation of the institutionalisation of the British government. Other people, seeing the high margins predicted for the Remain side, simply stayed home because they thought their side was bound to win. This resulted in a victory for the “Leave” side which had not been predicted by any of the so-called reputable polling agencies.

In these instances, polling companies change their protocol and tinker with the formula they use to predict results, so that by the next election they can come out declaring that they will not make the same mistakes as last time. However, they often do. Once more referring to Britain, in June 2017 there was a snap General Election called by the Prime Minister and, once more, the polls got the results completely wrong.

The youth who had stayed home in Brexit and been burnt by the result came out in force to protest the Conservative party, and many of the older generation unhappy with the referendum result and other political pressures also changed party allegiances. It was the case for many that they were not voting necessarily for the Labour party. Instead, they voted Labour or Liberal Democrat to actively prevent the Conservative Party from keeping their majority, a protest vote which changed the face of the Government. The massive swing from right to left in the political landscape had not been foreseen by polling bureaus who use voter statistics from previous elections to project turnout.

Perhaps it is for their own good that Cadem, one of the most influential Chilean survey companies, decided not to alter their own formulas in between the first round and final rounds of the presidential nomination, citing “no time and no knowledge of first-round the voter profile”. Whilst it is true that they predicted a much stronger victory for Piñera, and almost overlooked the influence that Sánchez would have on the population, so did all of the other Chilean pundits. At least if they get it wrong once again, they can just blame the same old formula.

The polls do not work because, in the end, they do not consider emotions. They look at numbers, analyse figures, consider statistical probabilities based on gender, age, and social status. But they don’t consider the emotional effects they themselves create, and they don’t consider that people can’t be analysed solely by numbers. People, just like elections, are illogical and uncertain.