You are here

Getting tie’ up with polls

Sunday, September 28, 2014

There is a truism in reporting on elections polls that newspapers and the media in general, love a horse race. It was seen, for example, most recently in the Scottish independence referendum where the media kept reporting a tight race with polls tied, although the ‘No’ vote had seized a commanding lead as the countdown to the actual polling day got closer. The eventual result, 55.7 per cent ‘No’ to  44.3 per cent ‘Yes’ was anything but close.

It is also seen in the United States presidential elections where the media report on the tightness of the race for the popular vote when the elections are decided, not by popular vote, but by the Electoral College count, in which one of the two main parties normally has a commanding lead over the other. It is not impossible for one party to win the popular vote, as Al Gore found out in 2000, while another wins the Electoral College which elects the president.

Just as it is in both the UK and the US, the popular vote can be a misleading statistic in T&T. To say that two parties are effectively tied 34-32, with the difference contained within the 3.8 margin of error, does not show much change in the election race. In the only tied election we have ever had in our history, the UNC managed to win 49.90 per cent of the vote to the PNM’s 46.51, a difference that represented 3.4 per cent points and 18,927 more votes and could not cement victory.

Why then, you may wonder, do pollsters do these polls that publish results which are misleading at best, meaningless at worse? One needs first to appreciate that most published polls are commissioned by newspapers that normally choose the headlines and have a vested interest (it’s a better news story or ‘bombshell’ to report on a competitive race).

Last Sunday’s Express Solution by Simulation poll was conducted, for example, to appear on the first day of the annual media survey which determines readership and crucial media revenue. The fact that the poll, which promised ‘explosive’ findings, resulted in the newspaper being sold out in several areas meant it was a good business decision.

The point that the poll was conducted from September 15 to 18, in the middle of the budget debate, when the People’s Partnership was at its most favourable after dispensing baby funds and pension increases, could account for the high marks given to the Prime Minister.

The PNM leader was also probably at his lowest with a savage attack from Housing Minister Roodal Moonilal on his tenure at the HDC dominating the news over the three days of the poll. A poll is always just a snapshot in time. The fact that either development would produce temporary spikes or falls in polling that normally don’t last should not be lost on anyone.

So that despite these factors, the polls still showed a tie in the popular vote (which in the first-past-the-post system is meaningless) spells bad news for the UNC and the People’s Partnership. A tie in the popular vote, as suggested by pollster Nigel Henry, means that the UNC or PP is probably on its way to losing the election.

The UNC, historically, has always had a higher voter turnout in its core constituencies than the PNM. The poll suggests that the PNM leads the UNC by 17 per cent in its core constituencies, while the UNC leads the PNM by 23 per cent in its core constituencies. Those figures skew the national popular vote totals to suggest a competitive race when, as noted by pollster Henry, the PNM will quite likely win the marginal seats which determine the election. 

Based on the historical trends, to be assured of winning, the UNC/PP will have to show at least a consistent four per cent gap from the PNM, which it has been unable to do. When conducting a poll with, for example, a four per cent margin of error, it means that to be assured of winning, the UNC or PP needs to be polling at least eight percentage points ahead of the PNM in the popular vote. Sunday’s poll, therefore, is bad news for them.

The other bad news for the ruling coalition is that given the high voter turnout among its supporters, especially in the marginal constituencies, the PP does not have much room to increase its voter turnout. While in the PNM constituencies and polling divisions, where there is a higher percentage of “not sures,” there is a much greater chance that the PNM can up its voter appeal through candidate selection and get out the vote campaigns by bringing out voters who will probably never vote for the UNC/PP.

The polls also suggest that, to be more meaningful, further polling would have to be done in the marginal constituencies. But such polling would be more expensive since it would require a higher number of respondents, and done over four to eighth constituencies.

While the PP and the PNM quite likely know the true state of the race, newspapers will avoid that kind of expenditure. Especially since, everyone probably knows the answer to the question of who will win the marginal seats and the next general election.


User comments posted on this website are the sole views and opinions of the comment writer and are not representative of Guardian Media Limited or its staff.

Guardian Media Limited accepts no liability and will not be held accountable for user comments.

Guardian Media Limited reserves the right to remove, to edit or to censor any comments.

Any content which is considered unsuitable, unlawful or offensive, includes personal details, advertises or promotes products, services or websites or repeats previous comments will be removed.

Before posting, please refer to the Community Standards, Terms and conditions and Privacy Policy

User profiles registered through fake social media accounts may be deleted without notice.