Don Anderson | On polls, pollsters and politicians

This article is intended to encourage fulsome and informed discussion about polls and their impact, and to replace emotional, subjective and unfounded comments when polls results are released which do not favour one party.

I want to reference the period 2015 to present, to reflect on polls conducted by me over this time, how as a pollster I have assessed the voter response and in the final analysis how politicians have reacted to the results of polls over these seven to eight years.

From my perspective, a single poll is of little value, it is but a snapshot at a point in time. Far more important is the conduct of a series of polls which helps in the formation of a trend of data which over time presents real value and which facilitates a strong basis for predicting the likely outcome of an election.

Take, for example, polls conducted by me over this period. A significant number of polls was conducted between 2015 and the election of 2016. Those 50-plus constituency and national polls showed that the election would be close, with the incumbent People’s National Party (PNP), then enjoying a 42-21 majority in Parliament, tipped to win. To the surprise of many, including the Jamaica Labour Party (JLP), the PNP contrived to lose that election by the slimmest of margins, 32-31.

Following that 2016 election, some 6 national polls were conducted by me up to the beginning of 2023 and none of these gave the PNP even a glimpse of the tape or a realistic chance of winning a national election. They trailed on average by some 10 percentage points up to the election of 2020. At the same time, the number of persons who voted for the PNP in 2020 was over 170,000 fewer than those who voted for them in 2011. Over the same period the JLP lost a mere 4000 supporters. The 49-14 victory for the JLP in that election was clearly on the cards.

The PNP was torn and fractured, a fact that dogged them over the next two years to end 2022.

VOTING INTENTION IN DON ANDERSON NATIONAL POLLS CONDUCTED BETWEEN 2016 TO 2023

2016 2018 2019 2020 (1) 2020 (2) 2021 2022 2023 (1) 2023 (2) 2023 (3) 2023 (4)

% % % % % % % % % % %

JLP 31 26 29 30 36 26 31 28 25 26 22

PNP 34 19 18 22 20 17 18 28 30 29 25

DIFFERENCE P3 J7 J11 J8 J16 J9 J13 0 P5 P3 P3

OVERALL JLP HAS A POSITIVE GAP OF 1O% POINTS PNP POSITIVE GAP OF 3% POINTS PNP 3% ahead

The data shows that the PNP went into the 2016 election with a three per cent point lead (dead heat) in the polls, but lost the election 32-31. That is as statistically correct as a pollster can get.

So, in polls conducted by me over this time, the JLP enjoyed an average of 10 percentage points lead over PNP, with the gap in support between the JLP reaching 16 per cent points in favour of the JLP in 2020. That was reflected in a 49-14 victory. Very accurate polling.

In none conducted over this time were there murmurs from the JLP regarding the pollster. Indeed, there was obvious acceptance by the JLP of the results that consistently favoured them over this time. Quietly the JLP embraced those findings.

Fast forward to 2023. The polls conducted so far for the year have confirmed wide anecdotal sayings that point to a significant degree of disaffection with the JLP, for reasons best known to the voting public and perhaps felt by the party.

Pollsters have nothing to do with voter affection or disaffection, we only gather information from the voters and report them as found. This is true irrespective of who commissions the polls.

This pollster has gathered and reported on poll findings in 2023, in the same way he gathered and reported on poll findings for the seven years between 2016 and 2022.

Absolutely nothing has changed in the methodology, between then and now, virtually the same questions have been asked.

Yet, suddenly, this same pollster is “contriving results to suit the PNP”, after 6 consecutive polls which had the PNP a distant second in a two-horse race.

I present this only for the purpose of allowing the general public to make a very objective assessment of the attempts to discredit the findings, as if such pronouncements might change the fact of the data.

From a dead heat in the February 2023 poll commissioned by the PNP, that party first jumped to a 5 percentage-point lead in June. That number dipped to three per cent points lead for the PNP in the August 2023 poll commissioned by the RJRGLEANER group.

POLLS DO NOT DETERMINE WINNER

Polls do not determine which party wins an election. Polls are designed to provide information on how the voting public feels about the respective parties at points in time, but are also intended to provide information that can help to develop and guide strategy during a campaign and what specific messages should be created that can generate an advantage for one party versus another. Especially in the context of the growing degree of transactional behaviour on the part of various parties, polls should really be used as a guide to effective campaigning, helping to determine where the emphasis should be put in intensive canvassing as part of the attempt to give voters a good enough reason-why to support them.

Instead, politicians have placed a considerable degree of emphasis on what the polls say to the extent that they become overly caught up in the party standings to the detriment of really learning from the results how to create this important advantage to win votes. So the tendency when poll results are released is to go straight to the specific results which show party standing and all the other valuable data gets lost in the effort to vilify the pollster as if by discrediting the messenger the numbers will change. I don’t recall any election anywhere I have worked across the Caribbean where public criticism of the poll results has gained any party sufficient votes to make a difference to the outcome of the election.

POLLS CAN IMPACT VOTER SUPPORT/THE BANDWAGON EFFECT

Having said this, polls can impact the level of support, under specific conditions.

Generally, if polls or historical voting patterns suggest an election might be close, the release of poll findings has been shown to have little impact. Rather, undecided voters listen and watch keenly for what benefits will accrue to them or are promised to them to determine how they will vote, whilst confirmed voters who have already made up their minds are hardly swayed by poll findings.

If, however, the polls consistently point to a comfortable win for one party, the release of poll findings can have the effect of strengthening support for the party that is well ahead and dissipating support for the party that is well behind. It is the bandwagon effect. In other words, the wagonists come out in full force. People want to associate with winners and conversely do not want to associate with losers. We know that from the level of support or otherwise we give to Jamaican sporting teams over the years.

POLITICS IS ABOUT MARKETING, SIMPLE

Politics is about marketing and voters are looking for a party to give them a good enough reason to vote for them. This is very similar to a business doing surveys to find out what are the key things that will make a customer buy their product. Efficient marketing companies, especially in a highly competitive environment, will look at what share of market they have for sure, but more importantly will look at the reasons why customers are buying the competitors’ more than theirs and then use other data to seek to gain a competitive advantage.

Market share or voter intent do not determine who will win an election.

Don Anderson has been conducting market research throughout the Caribbean for 50 years and political polls for 34 years. He was former senior lecturer (part-time) in research methods at UWI. Send feedback to columns@gleanerjm.com