"The pollsters got it all wrong," said MP Colm Imbert in a letter to the editor. So, even before getting a ministerial portfolio, Imbert was practising misleading statements.
In fact, H.H.B. & Associates, the firm contracted by Guardian Media Limited to conduct opinion polls for the election season, forecast a People's National Movement victory of 22 seats to 19. The actual result was 23-18.
The pollsters who got it wrong were Hamid Ghany of UWI's Constitutional Affairs and Parliamentary Studies Unit (Capsu), Vishnu Bisram of the North American Caribbean Teachers Association (Nacta) and Nigel Henry of Solutions by Simulation (SBS). Technically speaking, Henry didn't get it either wrong or right since, on the eve of the election, SBS said the race was too close to call. But H.H.B. managing director Louis Bertrand didn't mince his words, saying: If a poll cannot predict marginal seats, it is not a poll. This is inarguable. If as a pollster you can't make a forecast, then you are just as ignorant as the ordinary person in the street, except for having been paid a few hundred thousand dollars to admit your ignorance.
Until this election, SBS had an impressive record of accurate predictions. Their 2015 election poll, however, found that 51 per cent of their respondents supported the People's Partnership and 48 per cent the PNM. However, noting American research which showed that people's expectations of who would win an election was a more accurate predictor than statements of who they would vote for, Henry found that 37 per cent expected the PNM to win and 36 per cent the PP. Even so, he could not come up with a technique to penetrate that statistically tied curtain. Instead, SBS's telephone poll of 1,997 persons forecast that the winning party would have 21 or 22 seats: and that was definitely wrong.
Capsu's forecast was far more egregious. Released one week before the election, Capsu interviewed 2,328 respondents, who comprised 39.2 per cent Afro-Trinidadians, 38.3 per cent Indo-Trinidadians and 21.7 per cent mixed descent. This is a relatively large sample, yet Capsu found that the PP had a nearly seven per cent lead over the PNM in the marginal constituencies of La Horquetta/Talparo, Point Fortin, St Joseph, Toco/Sangre Grande and Tunapuna. All these seats were won by the PNM. Capsu also found that Kamla Persad-Bissessar was favoured as prime minister by over 42 per cent of respondents, as compared to 35 per cent who preferred Keith Rowley.
Nacta, which did not reveal its sample size but claimed to have conducted face to face interviews, said that the election could go either way and that 47 per cent of respondents preferred Kamla as compared to 42 per cent who preferred Rowley and that the PP had more national support.
H.H.B., however, was able to make its accurate forecast by polling just 120 persons in each of the five marginal seats.
So why did everyone else get it so wrong? There may be many factors, but bias could have been key. By this I do not mean that the bias was deliberate or conscious, and it may not even have been political. But anything from the phrasing of the questions to the choice of interviewers may have skewed the results, either because the respondents chosen were in some way a biased sample or because their responses did not reflect their actual opinions.
Bertrand also got the forecast right because he concentrated on the marginals. The other misleading statement Imbert made in his letter was that "seats deemed by the pollsters to be too close to call were, in fact, won by large majorities, making complete nonsense of the pollsters' predictions." While this was true of the three pollsters who got it wrong, H.H.B. was right on target. For Tunapuna and Toco/Sangre Grande, Bertrand forecast a PNM victory of 55 per cent and, for San Fernando West, his forecast was 60 per cent. The actual percentages in all three constituencies was 59 per cent for the victorious candidates.
None of this means that polls have been discredited in T&T politics. In fact, the lessons learned from this election will help pollsters to be more accurate for future elections. The key inference to be drawn, in my view, is that the swing voters in the marginal constituencies are just as monolithic in their thinking as the diehard voters. It must be so, or else the five or six marginal seats would not have all swung the same way.
With a voter turnout of nearly 67 per cent, according to the Elections and Boundaries Commission, the PNM got 378,447 votes while the combined total for the People's Partnership was 341,597. So the PNM won this election by a slim majority of 36,850 votes, which is a mere five per cent of all those who voted with a 51.5 per cent overall majority.
This means that corruption is not a key issue for most voters and that the Government does not have a strong foundation to win national support for its policies. At a time when hard decisions will have to be taken, this does not bode well for ordinary citizens.
Kevin Baldeosingh is a professional writer, author of three novels, and co-author of a history textbook.