Analysis of poll findings ‘seriously flawed’

Dear Editor,

Mr Chris Ram’s analysis of Vishnu Bisram’s poll findings is seriously flawed, biased and misguided.  It is not “Bisram [who] got it wrong” (letter caption SN, Dec 9) but it is Mr Ram who is wrong in his disaggregation and misinterpretation of the poll’s findings. He fails to demonstrate an adequate level of understanding of opinion polls to do a proper analysis.

As Mr Balram Rambrich (GC, Dec 16) pointed out, there is unadulterated bias in Mr Ram’s hatchet job on the NACTA poll. There were three other polls that he did not criticize for their failure to even come close to the election outcome. For example, the APNU poll gave APNU 40% of the votes (obtained from only five regions) and the closest rival (presumably the PPP) 30%.  The APNU poll was based on a turnout of 90% when the actual turnout was 73%. The Bhaskar Sharma poll gave the AFC 35% with the other parties trailing.  And another poll whose findings were revealed by AFC promoters had the PPP at 34%, APNU 29% and AFC 27%. Mr Ram failed to explain the anomalies of the other polls and why he opted not to analyze them and why he applied his caustic pen to NACTA’s poll.

On the NACTA poll, substantively, as Mr Rambrich pointed out, the poll is accurate in that it identified the PPP with 52% when the party got 49% which is well within the margin of error, and the findings were from the Friday before the election.  The same is true for the AFC support which was identified at 14% when the party got 11%.  The poll was off for APNU’s national support of 40%, but another updated poll sent to the media in Trinidad on the evening of Saturday before the elections showed changes in the mood of the electorate with the AFC (hovering at 11%) and the PPP (hovering at 50%) slipping among Africans, APNU (36%) gaining among Africans and the Indians splitting their votes.  It was not practical for new numbers to be published in the media in light of Election Day restrictions. All the new numbers were within the margin of error and the numbers for the regions had also changed substantially reducing Mr Ram’s 16 errors out of 48 findings.

And using Mr Ram’s analysis, he identified sixteen instances where he felt the poll was off target – only sixteen out of 48 instances – and he did not factor in the undecided voters or the unknowns I identified; this is therefore a very good and an accurate poll.

Mr Ram’s analysis of the numbers is deeply flawed and superficial. He did not take into consideration the almost 3% undecided voters who needed also to be included in any analysis using simply numbers.  He also failed to mention that the poll identified several dynamics (turnout rate, pull of race, division among Indian voters between PPP and AFC, Nagamootoo factor, etc) which were at play that would alter the support of the parties.  The late entry of Nagamootoo into the race introduced one of the most volatile dynamics. Mr Ram also failed to note that the poll numbers he analyzed were three days before an election and therefore would not reflect changes that occurred up until the day of voting.  Any serious, objective, impartial, unbiased professional analysis would take note of the many factors outlined herein.

It is easy to take pot shots at those polls and pollsters that show up one’s sacred cows. I encourage Mr Ram and others of his ilk to establish a polling institute in Guyana and let us see how much better they fare.

Mr Ram said the poll caused the PPP to lose the elections but he did not say how. I did not know he was sympathetic to the PPP and wanted it to win a majority.  I think the tone of the campaign, the perceived neglect of their supporters, and their refusal to heed the warning of the NACTA polls cost the PPP their majority. The NACTA polls were not conducted to help or hurt a party but to provide information on how parties are doing among the electorate. Polls alert politicians to act. It is up to them to heed or reject the findings. Has it not occurred to Mr Ram that the parties (AFC and APNU in particular) could very well have used the information from the NACTA polls to address their weaknesses in canvassing voters in areas where they had weak support?  I am told that APNU and AFC took the poll’s findings seriously using them to work harder on the ground making tremendous gains.  APNU and AFC went house to house to convince voters to support them and that strategy worked.

I am told the PPP was convinced it would win 60% of the votes and as such ignored the polls’ findings, opting instead to believe its own comrades on the ground. It would be productive if party officials would listen to their constituents, especially what they tell independent pollsters like NACTA. Polls should give politicians an indication of popular issues and the mood of the people. Unbiased, random, nonpartisan polls that show public preferences for various policies or parties could – should – be an important guide to politicians. Polling has become very accurate over the past few years and it does accurately reflect current attitudes within the electorate. Political leaders should be aware of these polling numbers. One positive aspect of the NACTA poll which is indisputable is that political leaders should pay close attention to it.

Mr Ram said the margin of error of the poll was too high. Let me inform him that the errors of a poll are not manufactured by the pollster but come from the sample size.  The larger the sample, the smaller the margin of error and vice versa. A pollster cannot be criticized or critiqued for the margin of error, exposing Mr Ram’s limited understanding of polls.

In conclusion, Mr Ram’s analysis of the NACTA poll is flawed and biased and cannot be taken seriously, for the poll got several things right which he fails to give credit for.

Yours faithfully,
Vishnu Bisram

Around the Web