19 Sep 2017

It has not been a good year for pollsters, thanks to the shocking results of Brexit and the US and UK elections. Should political leaders still rely on them and can their accuracy be improved?

The 2016 Brexit referendum cast serious doubts on the polling industry due to its unexpected results. Much of the surprise was blamed on pollsters, who had consistently predicted that a remain' result was guaranteed. Many post-referendum analyses concluded that the polls simply did not ensure representative samples and thus failed to accurately predict the results.

Skepticism surrounding polling was further compounded when the recent UK and US elections also culminated in unexpected outcomes. To make matters worse, politicians such as President Trump have taken to dismissing polls altogether, especially in instances where their approval ratings have declined drastically.

However, should political leaders be so quick to write off polls? In the political sphere, understanding public sentiment is crucial to good governance. If carried out with the right methodology, polls still remain one of the best ways to gauge public opinion and measure reactions to policymaking and constitutional design.

How polls are carried out

In politics, it is common to find opinion polls that survey citizens' general opinions of their governments. These include tracking polls, which monitor a candidate's success or approval ratings, and exit polls, which are a sample survey of people who have just voted at the polling stations. In the days and hours leading up to the declarations of results, such polls provide information for politicians and pundits alike to discuss possible outcomes.

These polls can be carried out in different ways, depending on the information that is being sought. Independent polling companies such as Gallup and YouGov attempt to get representative samples of the population and predominantly gather information from them through telephone polls or online surveys. Ultimately, however, anyone can conduct polls as long as they follow the respective methodologies.

The impact of polls on politics

A nudge for policymakers and leaders

The first and most important advantage of polling is that it gives a voice to the people. French President Emmanuel Macron's approval rating plummeted soon after he revealed his plans for market reform, which caused unhappiness and confusion among voters. Such discontent from both left-leaning and right-leaning citizens is a warning signal for the government to listen to the people, who currently feel that their concerns are not being addressed.

The second advantage is that polling raises awareness and enables a larger discussion about affairs ranging from policy reforms to broader issues, such as climate change. A campaign in Singapore to raise awareness on mindsets towards persons with disabilities, for example, started off by conducting surveys to glean insight on public perception of the matter. Thereafter, governments can tailor campaigns to educate people accordingly.

The third advantage is that polls have the potential to change the decisions of leaders. In Japan, where opinion polls are carried out to measure approval ratings of leaders, Prime Minister Shinzo Abe's approval rating recently dropped to below 30 per cent due to a recent scandal where he allegedly provided favours for a friend's business.

Since low voter support puts pressure on leaders to better cater to the public's wants, non-partisan surveys are helpful in conveying to officials that they need to take action. In this instance, the dismal approval rating spurred Abe to reshuffle his cabinet to restore the faith of citizens.

Why some polls miss their mark

The 2016 US election is often cited as an example of polls gone wrong. In reality, they were not entirely inaccurate, as Hillary Clinton did win the popular vote. Numerous polls had actually predicted that Clinton would win only by a thin margin, and post-election studies showed that an eventual difference of a few percentage points was not unusual. What are some of the factors that diminish the accuracy of polls?

Since it is not an exact science, one of the factors could be the potential margin of error. There are various reasons why this margin of error occurs. Overly small sample populations may not reflect the opinions of the entire nation. Even if larger sample sizes are obtained, they may not correctly reflect the economic and social variations within the population.

The margin of error also occurs because in certain instances, biased sampling is carried out, whereby polls are made accessible to a particular group of people in order to achieve a biased response.

Another factor that could affect accuracy is that polls themselves can be biased, depending on who is backing them. Push polls, for instance, can sway voters by using loaded questions and unethically disseminating information under the guise of surveys. It is also plausible that those being surveyed could provide false answers, as people may feel pressured to provide socially desirable responses to avoid being judged by the interviewer.

The third factor is that opinion polls capture public sentiment only at the time of survey, which could change depending on events that occur afterwards. During elections, two commonly observed phenomena are the bandwagon and underdog effects. With the former, voters may hop on the bandwagon and switch their vote to the winning party. In the case of the underdog effect, voters may choose to vote for the losing side out of sympathy.

Therefore, pre-election polls have the potential to swing the result of elections by influencing both voter and campaigner attitudes. In this year's UK election, for example, early polls in the UK indicated a guaranteed win for the Conservatives, but the actual election saw an unexpected surge of support for the Labour Party, especially for Jeremy Corbyn, who has often been portrayed by the media as the underdog. After the election, Prime Minister Theresa May was criticised for taking the Conservative lead for granted as Corbyn was holding spirited rallies that especially appealed to the youth.

Ensuring poll accuracy for strategic political engagement

Despite the risks that polls pose, they can be a valuable tool for political leaders looking to strategically engage citizens, if the following steps are taken to ensure their accuracy.

  1. Choose the right random sampling method

    To ensure that data collated accurately reflects the opinions of the general population, pollsters need to choose a random sampling method that represents the diversity of the country's citizens. For example, quota sampling would better account for social, economic and cultural variations in a country in order to create a proportionately diverse yet random sample.

  2. Select the best medium

    When it comes to medium, online surveys tend to produce more accurate results, whereas phone polls may reach a sample that does not represent the general population, say YouGov.

    The internet is also useful for pollsters to reach a bigger sample size and tap into big data to achieve more accurate analytics. However, one needs to keep in mind that although big data reduces statistical error, results can still be subject to sampling bias.

  3. Ensure questions are clearly worded

    Polling questions should be worded well and in a manner that is easy for respondents to understand, so that ambiguity and confusion will not ensue. Jargon and double-barrelled questions such as Should the government focus more on healthcare and less on education? should be avoided, as different issues must be addressed separately. Questions should also provide multiple-choice answers that are direct.

    Despite the hit that political polls have taken to their reputation in the past year, their value cannot be ignored. When administered correctly, they can provide policymakers with numerous useful insights without compromising on accuracy.

This piece was written by Prethika Nair.