Nov. 4, 2002 – According to the findings of two public opinion polls published over the weekend by The Avis, Gov. Charles W. Turnbull will come out ahead in Tuesday's general election and avoid a runoff by capturing a majority of the votes cast.
But the methodologies of the surveys are open to question, and in one case the polltaker, a government employee, isn't answering.
An article in the Saturday issue of The Avis said Rattan Polling Service, "a firm with over a decade of political prognosticating experience on the big island," released figures projecting "blowout victories" for incumbents Turnbull, Delegate Donna M. Christensen and at-large Sen. Almando "Rocky" Liburd.
The article did not mention that Rattan Polling Service is owned by Raymond James, special assistant to the commissioner of Housing Parks and Recreation. James is being paid $60,000 a year — up from $50,000 since Turnbull last summer granted more than $8.5 million in retroactive pay hikes to unclassified government employees.
The Avis did not indicate what type of poll was conducted — such as telephone, written questionnaire, door-to-door, show-of-hands, online, etc. — or how many persons were queried. And James told the Source he would not answer any questions about it.
The survey apparently was only of people on St. Croix, however. The Avis quoted an unnamed spokesperson for Rattan as saying the poll was of persons "from all walks of life and from all parts of the island," singular.
According to James, the Rattan poll was conducted over the last two weeks of October. He described it as an independent survey that has been taken at election time over the last 10 years.
However, officials at the Lieutenant Governor's Office said Rattan Polling Service is not a registered or licensed V.I. business. "It's just a public service kind of thing," James said.
Based on the results of its survey of about 200 people, Rattan projected that Turnbull and running mate Vargrave Richards will sweep the polls. The Avis article said the poll found 41 percent of the respondents support the Turnbull team, compared with14 percent for John de Jongh, 11 percent for Gerard Luz James II, 7 percent for Alicia "Chucky" Hansen, 6 percent for Michael Bornn and 4 percent for Cora Christian.
James claimed that the poll has a margin of error of plus or minus 5 to 7 percent, meaning it could be expected to be off by that much in either direction. That margin of error would depend on the methodology used in selecting the sample and carrying out the survey plus any demographic adjusting of the data, as well as the number of persons surveyed.
The other poll results, reported by The Avis in its Sunday-Monday issue, are from a series of three telephone surveys conducted by John Boyd, a former University of the Virgin Islands accounting and finance professor. Boyd also pegged Turnbull as the winner; however, his surveys found support for Turnbull gaining on St. Thomas but slipping on St. Croix over a three-month period.
Boyd said he conducted telephone polls in August, September and October on St. Thomas, St. Croix and St. John, each time querying 300 residents selected at random from the phone book. Businesses and any respondents who said they were undecided or would not vote were excluded from the poll results, he said.
Over the three-month period, he said, support for Turnbull jumped on St. Thomas from 54 percent in August to 68 percent in October, while it slipped on St. Croix from 44 percent in August to 36 percent last month.
The margin of error for any of the three single polls, Boyd said, is about plus or minus 7 percent. But he claimed that the margin goes down considerably when the polls are combined. "That margin drops to about 3 percent considering the island-wide results," he said.
However, according to public opinion research experts, because of the three separate time frames, it is statistically possible that a significant number of the initial respondents could have changed their views subsequently because of events occurring in the intervening periods. Also, in order to attain a 3 percent margin of error with random sampling, about 1,600 persons must be surveyed. In addition, excluding any respondents who said they were undecided would give a skewed measure of the public pulse, as the ultimate decisions of a significant number of "undecideds" can determine the outcome of an election.
Boyd said the discrepancy between his survey findings and those of Rattan could be wavering support for Hansen. "It wasn't so much margin of error as it was timing, with Chucky rising and falling so much," he said.
In the last gubernatorial election, Turnbull funded Boyd's election survey. But this time, Boyd, who described himself as a Republican, said he chose to conduct the poll on his own.
Jean Etsinger contributed to this report.
Publisher's note : Like the St. John Source now? Find out how you can love us twice as much — and show your support for the islands' free and independent news voice … click here.