Political scientist questions accuracy of interpreting election polls

Over the last few decades, the accuracy of election polls have decreased due to commercial, ‘non-political science’ pollsters, whose findings are often misinterpreted due to a lack of informed literature on voting behaviour.

That’s the argument being put forward by Professor Christopher Charles, political scientist at the University of the West Indies (UWI).

During his presentation on political forecasting at the American Political Science Association (APSA) Conference in Los Angeles on Thursday, Charles highlighted wanton misinterpretation of data by pollsters and political commentators, which often go uncorrected. This, he said, has led to a breakdown in public confidence in poll findings.

Founded in 1903, APSA is the leading professional organisation on the study of political science and serves more than 100 countries.

“In June 2023, the Don Anderson Poll found that 54.5 per cent of Jamaicans have a positive view of the prime minister, while 53.2 per cent have a positive view of the Opposition leader,” Charles told the gathering.

“The margin of error is +3% or -3%, so the leadership evaluation is a statistical dead heat. Yet there was a headline of a PNP lead guided by the fact that 50 per cent said that the country was heading in the wrong direction – an assumption that automatically means the PNP is leading. And this was not corrected by the pollster or political commentators,” he stated, adding that the same poll noted that 30.2 per cent of Jamaicans would vote for the People’s National Party (PNP) if the election is called today, while 25 per cent would vote for the Jamaica Labour Party (JLP).

“Since the margin of error is +3% or -3%, the PNP is only marginally ahead with 2.2%,” he explained. “The Westminster system is leader centric, so it is the leader approval that matters and less so the party standings in the poll, which in this case was a very marginal lead.”

“Leaders popularity drives voting behaviour and the leaders were in a statistical dead heat. However, the interpretation was that the PNP was leading,” he continued. “The pollsters have the numbers which is good. However, interpretation of these numbers also require in-depth understanding of political science and the empirical research on voting behaviour.”

Charles also told the conference of a history of daunting repercussions for pollsters and political observers in Jamaica’s small, tribalised and violent society – from physical threats to professional alienation, some of which, he said, he has had to deal with in his career.

“I have been attacked professionally, which has made younger scholars with the quantitative skills and method to do forecasting hesitant. There were also attacks from other sectors of the society that younger scholars do not want to take on at the start of their academic career,” he further explained to The Sunday Gleaner on Friday.

ANDERSON: THE RESULTS ARE VERY CLEAR

Pollster Don Anderson – who some regard as taking over the mantle from legendary political sociologist the late Carl Stone, who many say pioneered the systematic study of voting behaviour in Jamaica – said he was not particularly concerned with perceived misinterpretation of findings from his polls.

Of greater concern, he said, is people attacking the findings when it does not suit them. Nonetheless, one must contend with that, he added, noting that generally there is a greater appreciation of poll findings now compared to former years.

“I believe that the data I release on behalf of whoever it is that I’m working for, whether a party or media, I believe the results are very clear. I don’t think there is any ambiguity in the data, but obviously, everybody looks at data from the standpoint of their own positions,” the pollster of more than 40 years told The Sunday Gleaner on Friday.

Anderson has conducted political polls across the Caribbean, including in Antigua, St Kitts, the Cayman Islands and Grenada. Ninety-five per cent of his business comes from private contracts, he explained, as rebuttal for anyone who thinks he may be swayed by any political bias.

Addressing Charles’ views regarding the PNP’s percentage points lead over the JLP, as outlined in his latest poll, Anderson said that “as long as they are within the margin of error there is argument for both sides to claim that they are not far ahead or not far behind”.

“Most of our polls since 2017 have indicated that the gap that the Jamaica Labour Party has led the People’s National Party is well in excess of that margin of error of three per cent,” charged the head of Market Research Services. “In February of 2022, when we released data from a poll we conducted, the JLP were 13 percentage points ahead of the PNP, which is tantamount to a 6-nil drubbing in a football game.”

“When the gap in the July poll of 2023 was reported, the PNP was five percentage points ahead, which means, theoretically, either two percentage points, or eight percentage points, which is a more competitive environment than that 13 percentage points. So there is really no room for misinterpretation; people just read it how they want to.”

JLP: A TOUCHY ISSUE

Months after the release of the political polls, some members of the JLP hierarchy still view the subject as “very touchy”, declining to share their thoughts on how the findings were interpreted.

Asking not to be named, one party member told The Sunday Gleaner that, “There is no such thing as an incorrect poll but there can be an incorrect sample. So if I go into ‘Jungle’ and poll JLP versus PNP, there is no way JLP can win five per cent of the votes. So when a man says that he doesn’t believe a poll, it means that he doesn’t believe that the sampling was done in a manner that reflects what we are seeing.”

He continued, “When we did our JLP polls in 2016 we tried a different approach to sampling and that is why we were the only ones who predicted us winning. When everyone else was surprised that we won, we were not,” cautioning that polls only reflect individuals’ feelings and not necessarily how they will vote come election.

WIGNALL: WE ALL HAVE OUR BIASES

Political commentator, Mark Wignall, who worked with Carl Stone during his heyday, said that just like in yesteryears, pollsters will never fully rid themselves of the criticisms that follow their work.

“A lot of the commentators would like to pretend that they are free from their biases, but we all have our biases but the skill in dealing with that is to scale it down but I don’t think a lot of us can do it,” Wignall told The Sunday Gleaner. “So you are always going to find that there are variations in terms of interpretation.”

“I don’t know that the present set of analysts are as versed as Carl was in his craft, but I’ not going to criticise them as I definitely believe they are doing their best to try and get it right,” he said.

“Any thought of a criticism of sampling is just politics as usual. I wouldn’t pay it that much attention at all.”

corey.robinson@gleanerjm.com