The validity of opinion polls

2024 has been correctly described as the Year of Elections, as more than two billion people from 50 countries across the globe are expected to exercise their franchise. In modern history, the importance attached to opinion polls during times of national elections has increased considerably over time. The business community, in particular, use the results of such surveys to ascertain the impact of policies that might be implemented by the predicted winners on their businesses.

Conducting an opinion poll to predict the outcome of a national election is a challenging task. In the Sri Lankan context, the history is replete with examples of polls being conducted to achieve the agendas of groups with vested interests. Unfortunately, no independent authority exists in the country to determine the reliability and the statistical foundation of the opinion polls that are frequently cited by the media.

Nevertheless, the credibility of opinion polling has been a topic of contention not only locally but internationally as well. In 1936, the reputed magazine Literary Digest conducted a poll about the US Presidential Election with a sample size of 2.4 million, a massive number by any yardstick. The publication predicted that the Republican candidate Alfred Landon would get a 57% vote share while incumbent President Franklin Roosevelt would receive 43%. The actual result diametrically differed from projections – Roosevelt managed 62% support, compared to 38% of votes for Landon.

Exactly two decades ago in India, at the conclusion of the Lokh Sabha Election, every Exit Poll predicted the return of Atal Bihari Vajpayee’s NDA government. But to the dismay of pollsters, Sonia Gandhi’s Congress-led UPA emerged victorious. In the contemporary history, the performance of surveys conducted in the UK to predict the outcome of the Brexit referendum has been recognised as a monumental blunder. The integrity of polling suffered a further blow when Donald Trump emerged victorious at the 2016 Presidential Election, proving wrong pollsters who projected a landslide victory for Hillary Clinton.

In this backdrop, the opinion poll conducted by the Institute of Health Policy (IHP) on the Presidential Election Voting Intent – titled Sri Lanka Opinion Tracker Survey (SLOTS) – has raised many eyebrows. Because of the enormous attention this particular poll has obtained from the local and international media, a close examination of it would be meaningful. The findings of this poll are based on a monthly sample size of randomly selected 553 individuals on average. As per the Election Commission, there were 16,692,398 registered voters in the country at the end of 2022. 

Therefore, the average sample size of 553, which is used by the IHP for its poll, represents just 0.0033% of the registered voters in the 2022 Electoral Register. The gross inadequacy of the size of the sample is a major drawback associated with this controversial survey. Further, there is no information about the gender and age of the individuals who constitute the selected sample to determine its representativeness.

Amusingly, all the respondents in the IHP poll have expressed a preference to a particular candidate, which could be termed as a unique phenomenon. In any selected sample, there could be individuals who have already decided to refrain from voting at the forthcoming election while there are some who are still undecided about their choice. The exclusion of individuals like Dilith Jayaweera and Dhammika Perera from the options available to respondents is another glaring shortcoming as those two have already hinted about their ambitions for the top post.

The results of published polls of this nature could be used to persuade the undecided voters to vote for the individual/party who is leading in the survey in addition to aiding the outfit which is ahead of others to generate funds for their election campaign. In democracies, opinion polls are familiar during elections. However, limitations are imposed in many countries, from two to 21 days prior to the poll — Canada, France, and Italy, are few such examples.