BARISAN Nasional is expected to retain power in this general election.
So says a survey by pollster Merdeka Center.
This is in a stark contrast to March’s survey by another pollster, Invoke, which predicted the collapse of Barisan, and PAS winning zero seats. A more recent poll by Invoke showed a more tempered result: 40.5% of 1,961 registered voters said they are confident of a change in the federal government.
Opinion polls and surveys are not new in our electoral landscape, but GE14 seems to be cluttered with them. Especially with new players – like Invoke headed by PKR leader Rafizi Ramli – jumping into the fray.
With less than a week to polling day, these polls and surveys can be overwhelming for the average voter. And with their recent bad track record worldwide – Brexit and Donald Trump’s victory in the 2016 presidential elections were two major boo-boos – why should we believe them?
As one writer in the British online newspaper The Independent described them, “Opinion polls are the worst way to find out what people think apart from any other. The alternatives to respectable polls are astrology, gut feeling or just asking people who are available.”
Opinion polls about election are important to a democracy, says political analyst Professor Datuk Dr Shamsul Amri Baharuddin, if they are carried out properly and satisfactorily.
“Information about political access, political literacy and institutional awareness among the population in a democratic system could be found through election polls,” he says.
He explains public polls are essentially a market tool to determine market share, sales, and consumers’ choice using a survey method usually based on limited random sampling.
“So an election poll is a predictive tool to assist ‘the market share’ of support for a politician and political parties, the ‘acceptance and consumption’ of potential voters of the political parties’ manifesto, and which individual prominent politician is the public favourite (in terms of trustworthiness, effectiveness and other elements of leadership).
“Compared to roundtable discussions or focus groups, survey-based polls may not be providing deeper information about people and issues but they have a broader reach and bring quick results.
“It must not be ignored that polls also help journalists, political analysts, and citizens get information about a campaign. Those using social media find data from polls useful in providing ‘cheeseburger-sized’ information that can be consumed and shared quickly across various social media platforms.”
Polls can also determine how much a political party, individuals or NGOs would want to spend for their campaign or how much they can ask for from donors.
“That is why they say election is expensive because it is not only about posters, buntings and banners or the makan and teh tarik that has to be provided, but also about conducting the polls. In Malaysia it can run to millions of ringgit,” says Dr Shamsul Amri.
He adds, generally, polls are useful not only during the years leading to an election, or during an election year, but also the years in-between elections.
“It is an exercise of finding out the state of current political mood and in some sense predicting the ‘political weather, where the wind of general support ‘blows’ so to speak.”
Without doubt, polls are important tools for individual politicians, political parties and other politically oriented organisations, he notes. “Through polls they would know which candidates are favourites of the public and who could eventually win the voters’ support.”
Here are some answers to your burning questions about polls compiled by Sunday Star:
How can a sample of 800 to 1,500 truly reflect the opinion of over 30 million Malaysians?
Election polls use the survey method to find out whatever information and opinions they need from the targeted public. A survey can only be truly valuable when it is reliable and representative. In Malaysia, the voters are over 21-year-old adults. So, the target group is clear.
According to the Election Commission, in 2017, not all 32 million of the Malaysian population were voters. The number of registered voters is 14.4 million – some 3.6 million of eligible voters are not registered. For surveys or election polls, the relevant target known as ‘survey population’ (relevant only to the election polls) is only half the size of the total population.
How do we determine the ideal survey sample size and the population (or target group)?
In an election poll, the population is the group of voters.
“We can’t survey or interview every one of the 14.4 million voters. The challenge is the selection of sample from this known number of voters,” says Dr Shamsul Amri.
A sample is a selection of respondents, who are chosen in such a way that they represent the total population as good as possible. Too big is a waste time and money. Too little will not provide reliable insights.
The calculation of the margin of error (positive and negative deviation you allow in your survey result) and confidence level (how many of the respondents actually lie within the ‘margin of error’) is necessary to decide the final number of survey sample needed.
Assume you want only 100 respondents. You would achieve 95% of confidence level if 80 respondents replied, that represents the margin of error of 5%. If 94 replied margin of error is 2.5% and if 99 replied the margin is 1%.
If you want 300 respondents to complete answering the survey you would usually distribute 600-700 questionnaires. Not all will return them but you need only 300 returns.
Then one has to evaluate the content of the acceptable response. A response rate of 20% is considered a good response, and a 30% rate is considered excellent.
How come my friends or I have never been polled?
As there are approximately 14.4 million voters in Malaysia, it means in a poll with a sample of 1,000, only 1 in 14,400 people will be interviewed. Most polling now is done by phone, both landline and mobile, but most people don’t pick up or hang up on cold calls.
Dr Samsul Amri also points out that there are plenty of choices of survey teams doing election polls in Malaysia, but all are based in Klang Valley.
How accurate are polls?
The selection of sample and administration of the returned questionnaire and content are critical. Says Dr Shamsul Amri, “In a war-torn country like Sri Lanka, for more that two decades, it was impossible to conduct such polls satisfactorily even though there were regular elections because they were conducted only among the Sinhalese, not among the Tamils.”
The American Association for Public Opinion Research’s best practice guidelines say the quality of a survey is best judged by “how much attention is given to preventing, measuring and dealing with the many important problems that can arise”, rather than just the size or scope of the polls.
Can the polls influence the outcome of an election?
An opinion poll is a survey, or snapshot, of political opinion and cannot predict the future.
As an Opposition party worker, who declines to be named, puts it, “People change their minds and focus more on election campaigns. If opinion polls can determine the result of the election, parties won’t spend money campaigning.”
There is, however, a “bandwagon effect” which has been widely quoted by political analysts around the world.
Australian National University politics professor Ian McAllister, who has studied election trends since 1987 as part of the Australian Election Study, has described the phenomenon as “backing a winner”.
“Voters like to back a winner, they don’t like to back a loser so if they see a party leading in the polls they’re much more likely to gravitate towards that,” he was quoted as saying.
How should the public take opinion polls?
The public should use an opinion poll just as a guide, says Dr Shamsul Amri. If you observe there is consistency in the results produced by a particular survey team, then you should stick to it and follow its findings. Of course you can check with other teams to give a sense of comparison.