setUoYouRPROFILE's blog: "tufui"

If election polls were perfectly accurate — with a margin of error of 0 percentage points, 20 times out of 20 — then your services as a voter would not be required on Thursday. But guess what. Your services as a voter will definitely be required on Thursday, because public opinion polls can be all over the map. And in Ontario, that’s exactly where they’ve been lately. The still evolving science of predicting mass preferences based on limited samples has left many of this province’s voters scratching their collective foreheads. Depending on which of several eleventh-hour polls you choose to consult,gucci outlet the incumbent Liberals are either trailing the Tories by a narrow margin — suggesting a hung legislature for the next few years — or they are comfortably out in front, headed for what could be a third majority in a row. The outcome of the vote will have an impact on just about everyone in the province, but professional pollsters have more at stake than most. Their reputations, and their capacity to attract or retain clients, depends on their ability to make accurate predictions and to avoid statistical train wrecks. “If you consistently get it wrong,” says Frank Graves, president of EKOS Research Associates, “you’re not going to be in the business anymore.” Graves is among those who are calling Thursday’s vote a definite win for the Liberals — possibly a majority — and a certain defeat for the Conservatives. “A Liberal majority is a good call,” he says. “The Tories are simply not ahead.” In fact, if the EKOS numbers are right, the Tories under Tim Hudak are well behind, with the support of just 29.7 per cent of Ontario voters, compared to 39 per cent for Dalton McGuinty’s Liberals. The New Democrats under Andrea Horwath trail with 23.1 per cent. For those keeping track, the EKOS poll was conducted between Oct. 2 and 4, involved a sample of 2,081 respondents, and has a margin of error of 2.2 percentage points, 19 times out of 20. It was an IVR poll (interactive voice response), in which a computer calls people on the phone and they register their opinions by using the phone keypad. Meanwhile, a recent poll conducted on behalf of the Star by Angus Reid Public Opinion and released Tuesday yielded a strikingly different result, putting the Progressive Conservatives on top with 36 per cent, followed by the Liberals with 33 per cent and the NDP with 26. Conducted from Oct. 3 to 4, the Angus Reid survey employed a very different methodology — an online sounding of 2,233 computer-equipped respondents who had previously volunteered to be consulted on such matters. The poll claimed a margin of error of 2.1 percentage points, 19 times out of 20. “I’m skeptical of that methodology,” says Graves. “In this case, I’m suspicious of the results.” Among a raft of surveys conducted by major firms in the past few days, only the Angus Reid poll showed the Tories out in front, although another sampling — this one by Forum Research Inc. and also carried out on behalf of the Star — put the two parties in a dead heat. Other large polling firms, including Nanos Research and Ipsos Reid, have also conducted surveys of the Ontario electorate in the past few days and their results mirror the EKOS findings, giving the Liberals a fairly significant lead over the Tories. “Regardless of what transpires over the next 48 hours, the McGuinty Liberals will govern the province after a third straight election,” Ipsos Reid predicts on its website, referring to a poll released Tuesday. That survey reflected a sampling of 1,020 adults, who were questioned over the phone. Margin of error was 3.1 percentage points, 19 times out of 20. Jaideep Mukerji, managing director at Angus Reid, defends his company’s poll showing the Tories ahead of the Liberals, but he seems to be talking about a different survey. In fact, he is. Angus Reid quickly conducted a smaller follow-up poll, whose results were published Wednesday. It puts the Liberals ahead of the Tories, by 37 per cent to 33. The NDP had 26 per cent. “Every pollster found the same thing — the Tories blew their lead,” Mukerji said. “We’ve certainly seen a boost in support for McGuinty. More and more, people believe the Liberals will win the election.” Pollsters say many factors can cause disparities between different polls. They include the timing of the field work, the wording and order of the questions asked, the means used to select respondents, the technology employed to interact with them, and the kind of data massaging — or data “weighting,” the industry’s preferred term — that is later invoked to refine raw results. “Polls generally perform quite well,” says Graves at EKOS. “But all these methodologies can be used well or poorly. It’s kind of a tricky game.”