Gambling on polls that ‘may’ not count

Without filters, social media will lose its credibility of being a reliable vehicle of information

From a united Europe to Brexit; from Theresa May to Theresa May be; from Hilary Clinton to a triumph for Trump; from a close contest between Marie Le Pen and Emmanuelle Macron to a sweep for Macron, predictions, forecasts and polls have become a coin-toss exercise. Analysts have been made to look fumbling, pundits have been made to sound mumbling and experts have been made to eat their own humble pie with the actual numbers beating the predictions regularly. This in times of great advancement in research techniques, data analytics, voter profiling, behaviour predictors, etc, is almost incomprehensible. As the science of researching human psychology assumes numerical perfection, the science of decision-making based on it becomes highly imperfect. In the last one year the two leading powers in opinion poll metrics — the US and Europe — have seen shocking deviation from graphs and trends that dominate the debate, discussion and influence dramatic decisions.

Questions are now being asked as to the accuracy and authenticity of these polls. Academics are shell-shocked, voters are confused and politicians are enraged. Opinion polls are and will remain a major indicator for political preferences. However, they need to make certain adjustments to this instant click world. Their predictions are leading to decisions that are wide scale and far reaching. The UK polls were so convincing that it was a safe bet for Theresa May to call for a snap election. Mid-April lead for Conservatives was 20% with a clear majority. Most “respectable” polls were predicting this eventuality. Why did she end up with a minority in June? Some major factors need to be taken into account while reading and interpreting these polls.

Voter intentions and voter turnouts are not correlated: In Trump’s case and in May’s case the polls could not predict the fact that how likely are people who were interviewed to actually vote or not. With turnout around 60%, the 40% who donot turn up may be the majority of the those who were forming a trend or vice versa. Only in compulsory voting as in Australia where you are fined for not voting do the polls remain accurate. For Trump voter turnout was much greater than those who preferred Hilary. Similarly for Labour and Jeremy Corbyn the voter turnout for youth was unprecedented.

The second factor is the campaign: From the time the election is announced to the time the vote is casted the campaign can swing the voters either way, especially the floating voters. Theresa May’s campaign was a disaster. The “strong and stable” theme crashed on her dithering stance on Brexit plan and was not helped by the two terrorist attacks in London. Her campaign was all about her and not Britain and she was perceived as wooden and cold. Jeremy Corbyn on the other hand focused on “hope and the future of Britain” and came out more inspiring and warm.



The third and most important factor is the impact of the viral social media: From Trump to May the ability for each word and statement to be framed or twisted to create a narrative is terrifying in its speed and ferocity. Hilary’s inability to deal with the last flood of email discovery hours before the campaign finished made the swing voters stay at home. In Theresa May’s case a song released days before the vote titled “ She is a liar, liar, liar..” hit the top of the charts showcasing her instability giving her party little chance and space to retaliate.

These are the realities of the new world of hyper awareness and over responsiveness. As they say, what brought you here will not take you there. To cope with these challenges a change in thinking, strategy and legislation is required at the policy and research levels. It is very evident that research is much more accurate in countries where voting is compulsory. This has been authenticated by the results in Australia. Newspoll predicted a Turnbull Coalition victory last year, and that is exactly what happened.

Firstly, lawmakers in countries must debate on having mandatory voting. In countries like Pakistan it is even more important to ensure that more and more people vote as voter turnouts are low giving rise to more rigging in polls. We have seen in our country that voter cynicism is the highest in the educated classes and to cater to their disillusionment a column of none of the above can be added to prevent their vote going waste or being misused by others.

Secondly, research companies need to improve their targeting by closing the gap between voter intent and voting itself. This can be done by simple leading or indirect questions that determine the likelihood of their turning up on election day and then deducing their opinions on that basis. They also need to make their tracker tools more current to absorb the impact of social media virality.

Thirdly, social media platforms need to develop more sensitive software to comment on the veracity of news. In the academic world, software picks up the plagiarised material. More research needs to be done to identify fake accounts, tweets, photo-shopped news, to counter this label of social media being a post-truth alternative facts forum. Without these filters, social media will lose its credibility of being a reliable vehicle of information.

Whereas informed decision-making is the leadership buzzword, the source, speed and credibility of the information is vital to prevent global decisions being made almost on a mere roll of the dice.

Published in The Express Tribune, June 18^th, 2017.

Like Opinion & Editorial on Facebook, follow @ETOpEd on Twitter to receive all updates on all our daily pieces.