This article is from the source 'guardian' and was first published or seen on . It last changed over 40 days ago and won't be checked again for changes.

You can find the current article at its original source at http://www.theguardian.com/politics/2015/jun/10/polling-industry-general-election-labour-

The article has changed 2 times. There is an RSS feed of changes available.

Version 0 Version 1
Polling industry must come under state control, say Labour peers Polling industry must be more tightly regulated, say Labour peers
(35 minutes later)
Britain’s polling industry should be brought under the control of a state-backed external regulator in the wake of their failure to predict the outcome of the 2015 general election, Labour peers are proposing. Britain’s polling industry should be brought under the control of a state-backed external regulator in the wake of their failure to predict the outcome of the 2015 general election, Labour peers are proposing. Lord Foulkes has tabled a private member’s bill bringing the multi-million-pound industry under the control of a state-appointed regulator rather than the current self-regulatory body, the British Polling Council. The state regulator would be empowered to consider whether polls should be banned in an election period.
Lord Foulkes has tabled a private members bill bringing the multi-million pound industry under the control of a state-appointed regulator rather than the current self-regulatory body, the British Polling Council. The state regulator would be empowered to consider whether polls should be banned in an election period. The council has already set up an independent inquiry into the pollsters’ collective failure to predict the outcome of election. The media have been excluded from sitting on the inquiry, which is largely populated by statisticians. It is being chaired by Patrick Sturgis, professor of research methodology in social statistics and demography at the University of Southampton, and is due to report next year.
The council has already set up an independent inquiry into how the pollsters collectively failed to predict the election. The media has been excluded from sitting on the inquiry that is largely populated by statisticians. It is being chaired by Patrick Sturgis, professor of research methodology in social statistics and demography at the University of Southampton, and is due to report next year. But in evidence to the inquiry, the Labour peer Lord Lipsey has accused the industry of encouraging exaggerated reporting of their findings, tailoring questions to fit the views of those who commission polls, and abusing anyone who dares to criticise their methods. He suggest that pollsters “insist that they achieve a degree of statistical reliability which they neither can nor do achieve, and only sign up for the loosest of regulation”.
But in evidence to the inquiry, the Labour peer Lord Lipsey accuses the industry of encouraging exaggerated reporting of their findings, tailoring questions to fit the views of those who commission polls and abusing anyone who dares criticise their methods. He suggest pollsters “insist that they achieve a degree of statistical reliability which they neither can nor do achieve, and only sign up for the loosest of regulation”. Lipsey, the joint chair of the all-party parliamentary statisticians’ group, writes: “The opinion polls in the 2015 election campaign were not only wrong. They greatly influenced the nature of the campaign and very possibly the result. In a democracy, this is no small matter. You would think that professional pollsters would regard it as part of their duty to assist those who report their work in explaining its limitations. They do not. Rather, they encourage exaggeration (as it is likely to achieve more follow-up publicity for the poll).”
Lipsey, the joint chair of the all-party parliamentary statisticians group, writes: “The opinion polls in the 2015 election campaign were not only wrong. They greatly influenced the nature of the campaign and very possibly the result. In a democracy, this is no small matter. The Foulkes bill has little chance of becoming law, but it reflects a backlash in political circles at the inaccuracy and prevalence of the polls in the last election and the potential impact on the result. The Tory election strategist Lynton Crosby has called for polls to be banned in the two or three weeks leading up to an election.
“You would think that professional pollsters would regard it as part of their duty to assist those who report their work in explaining its limitations. They do not. Rather, they encourage exaggeration (as it is likely to achieve more followup publicity for the poll).” Lipsey called for the inquiry to investigate whether pollsters practise “huddling” – adjusting their figures into line with those of other pollsters on the grounds that, if they are to be wrong, it is better to be wrong collectively. He suggests that newspapers often ask questions that fit their political agenda.
The Foulkes bill has little chance of becoming law, but it reflects a backlash in political circles at the inaccuracy and prevalence of the polls in the last election and its potential impact on the election outcome. The Tory election strategist Lynton Crosby has called for polls to be banned in the two or three weeks leading up to an election. He argues that the problem of polling has been magnified by the breakdown of social class as a guide to voting intention, the decline of face-to-face interviewing, and the switch to telephone panel-based polling.
Lipsey calls for the inquiry to investigate whether pollsters practice “huddling”, adjusting their figures into line with those of other pollsters on the grounds that, if they are to be wrong, it is better to be wrong collectively. He suggests newspapers often ask questions that fit their political agenda.
He argues the problems of polling has been magnified by the breakdown of social class as a guide to voting intention, the decline of face to face interviewing and the switch to telephone panel-based polling.
He writes: “Perhaps in consequence, at least in general elections, the results have not been impressive. The polls were rightish (in the sense of getting the right winner) in 1997, 2001, 2005 and 2010. They were catastrophically wrong in 1992 and 2015. As they would pick the right winner by chance one time in two, that means that their actual performance was of 67% success against a likely performance using a pin of 50% success. Even allowing for the small sample, this is not convincing.”He writes: “Perhaps in consequence, at least in general elections, the results have not been impressive. The polls were rightish (in the sense of getting the right winner) in 1997, 2001, 2005 and 2010. They were catastrophically wrong in 1992 and 2015. As they would pick the right winner by chance one time in two, that means that their actual performance was of 67% success against a likely performance using a pin of 50% success. Even allowing for the small sample, this is not convincing.”
In his evidence, Lipsey claims: “Most polling companies care about their reputation and go about their work in a professional manner. But they have now got it badly wrong in a third of recent elections. And of course they may also have got it wrong in most of the polls conducted between elections, when there is no real result to check their published figures against.In his evidence, Lipsey claims: “Most polling companies care about their reputation and go about their work in a professional manner. But they have now got it badly wrong in a third of recent elections. And of course they may also have got it wrong in most of the polls conducted between elections, when there is no real result to check their published figures against.
“When the polls get it right, the pollsters exalt. When they get it wrong, they blame the statistical margin of error. Formally, if they interview a perfect sample of voters, there is a 19 in 20 chance that the figure they get for each party is +/- 2-3% of the true figure. “When the polls get it right, the pollsters exult. When they get it wrong, they blame the statistical margin of error. Formally, if they interview a perfect sample of voters, there is a 19 in 20 chance that the figure they get for each party is +/- 2-3% of the true figure.
“That is already a big margin, for it applies to each figure in an opinion poll. Take a poll that puts Labour and the Tories level on 30% of the vote. This means each party’s true standing is likely to be between 27% and 33%. So the poll could mean the Tories are six points ahead or Labour is six points ahead, though the truth is more likely to lie towards the centre of this range. One poll in 20 would be expected to have an even bigger error – as was for example probably true of the late poll in the Scottish devolution referendum which put yes in front.“That is already a big margin, for it applies to each figure in an opinion poll. Take a poll that puts Labour and the Tories level on 30% of the vote. This means each party’s true standing is likely to be between 27% and 33%. So the poll could mean the Tories are six points ahead or Labour is six points ahead, though the truth is more likely to lie towards the centre of this range. One poll in 20 would be expected to have an even bigger error – as was for example probably true of the late poll in the Scottish devolution referendum which put yes in front.
“However the unanimity of the polls means that in the 2015 election, the statistical margin of error is unlikely to be the sole explanation. The more polls you have and the bigger the total sample, the smaller the statistical error.” “However, the unanimity of the polls means that in the 2015 election, the statistical margin of error is unlikely to be the sole explanation. The more polls you have and the bigger the total sample, the smaller the statistical error.”
The Foulkes regulator would specify approved sampling methods; give guidance on the wording of questions to be put to the public in a political opinion poll; and put in place arrangements for the publication of polls including whether such publication should be allowed in a period that the authority considers appropriate before the election or referendum to which they refer. The Foulkes regulator would specify approved sampling methods; give guidance on the wording of questions to be put to the public in a political opinion poll; and put in place arrangements for the publication of polls, including a possible ban on polling in the run-up to referendums and general elections.
“Because of the polls, the campaign was all about the consequences of a Labour minority government, with the SNP holding the balance of power at Westminster,” Lord Foulkes said. “If the public had known the true position, it would instead have been about whether they really wanted a majority Tory government, committed to cuts, to austerity and to a referendum on the EU.”