Wired868 Wired868 for smart sport news and opinion
“[…] Knowing public opinion helps parties in the closing weeks of an election… Well-done opinion polls that are made public give everyone access to the same information. This approach helps our democracy.
“[…] The Ghany approach raises two main methodological concerns that merit closer examination: sample size (approximately 125 per constituency in Trinidad) and the professed margin of error in each constituency…”
The following Letter to the Editor was submitted to Wired868 by Market Facts and Opinions (MFO), a Trinidad and Tobago-based Caribbean market research company:
As a matter of principle, MFO typically refrains from commenting on polls done by other parties. However, given the broader implications of the recently published Ghany poll, we feel it is necessary to provide a professional perspective—particularly as it relates to the standards and integrity of public opinion polling in Trinidad and Tobago.
For the past 40 years, we have not engaged in political polling nor intend to do so.
Knowing public opinion helps parties in the closing weeks of an election. Reporting on opinion gives the news media good stories. Regardless of who commissions the poll, all polls must be done well—otherwise, they will provide a distorted opinion view.
Well-done opinion polls that are made public give everyone access to the same information. This approach helps our democracy.
Political polls are not new in Trinidad and Tobago. In 2020, Louis Bertrand of HHB, sponsored by the Trinidad Guardian, ran a series of polls on the marginal constituencies.
On 27 July 2020, the Trinidad Guardian reported that both the People’s National Movement (PNM) and the Opposition United National Congress (UNC) expressed satisfaction with the poll.
At the time, Bertrand stated: “With two weeks to go before voters cast their ballots, 16 per cent of the respondents said they are still undecided, and 11 per cent indicated they would not vote. The margin of error is plus four, minus four.”
He had done samples of 300 and 200 in the constituencies.
The Ghany approach raises two main methodological concerns that merit closer examination: sample size (approximately 125 per constituency in Trinidad) and the professed margin of error in each constituency.
With such a small sample, the reliability of results becomes questionable, especially when used to draw conclusions about demographic subgroups or voter behaviour at a granular level.
The construction of the sample, as described theoretically, would not help the researcher to understand potential voting shifts. His ‘invented’ approach would not improve sensitivity to understanding those constituencies’ non-partisan voters.
It would have been more effective to follow Bertrand’s approach of understanding how his respondents reported voting on the last occasion.
Interestingly, for the Tobago samples (each with 250 respondents), the poll claims a margin of error of 6%. Yet in Trinidad’s more heterogenous constituencies, a significantly smaller respondent base of 125 claims to carry a margin of error of 2%. The statistics escape us.
(via UNC.)
NACTA noted that the data was collected through in-person interviewing. This method is touted as the Caribbean standard. While face-to-face methods have their place, it is important to note that the 2010, 2015, and 2020 political polls have increasingly adopted telephone-based interviewing.
Moreover, due to security concerns, MFO, as one of the larger Caribbean research houses, has phased out door-to-door polling—except in areas with limited access to telecommunications services. We instead utilise our in-house call centre.
Knowing the total number of persons contacted versus those who participated would be interesting. Why? Ghany’s poll recorded significantly improved responses about voter intention compared to Bertrand’s 2015 and 2020 polls.
(via PNM.)
As a matter of comparison, in San Fernando West for the 2020 elections, Bertrand recorded 15% of those who refused to indicate their preference. His margin of error for that constituency was 6%. We await such disclosures.
We urge caution in interpreting and relying on the results of this poll without more information on its methodology. In our industry, an old saying is: “Even broken clocks are right twice a day.”
Mr Ghany’s predictions could be correct even though his methodology is questionable. However, we believe the conduct of this poll has done more harm than good to the reputation of public opinion research in our region.
We share these thoughts not to undermine, but to advocate for better standards, greater transparency, and more robust approaches in our industry—for the benefit of all stakeholders in our democracy.
“[…] The 10 years under the Dr Keith Rowley-led PNM saw a vicious attack on the trade union movement and Read more
“[…] Most carnival costumes (beads, bikinis, feathers, clothes, uniforms, caps, wire bending, etc) are now manufactured in China at reasonable costs—transport Read more
“There are two things that are important in politics,” said Mark Hanna, a 19th-century businessman and political kingmaker in Cleveland, Read more
“[…] When police officers disobey direct orders from superiors to wear body cameras, it is a serious breach of protocol Read more
In the run-up to our General Elections, I indicated how the world’s events impact our country and how our method Read more
“[…] Trinidad and Tobago now joins the United Arab Emirates and Canada as the only countries in the world with Read more
Want to share your thoughts with Wired868? Email us at editor@wired868.com.
Please keep your letter between 300 to 600 words and be sure to read it over first for typos and punctuation.
We don’t publish anonymously unless there is a good reason, such as an obvious threat of harassment or job loss.
Do well-done opinion polls really help democracy? Point out the facts to support that opinion please.
Agree!!! With the stakes being as high as they are a lot more rigor should’ve been built into the methodology. Glad that a reputable research company has raised these concerns. All very valid and if taken on face value can mislead (hopefully not deliberately) the casual observer.