A survey claiming support for Yes has dropped to just 25 per cent has been criticised after it emerged no weightings had been carried out.
The latest survey by pollster TNS claimed support for Yes had dropped by five points since its last survey, whilst support for No had dropped by only four. The team behind the survey also claimed to have uncovered a huge surge in those describing themselves as ‘Don’t Knows’.
However it has now emerged that the methodology employed by TNS did not weigh the results depending on how participants voted in the last election. Weightings ensure that the breakdown of respondents taking part in the survey take into account the parties they voted for in previous elections.
According to polling expert Professor John Curtice, the TNS survey included more people who had voted Labour in the last Scottish election than voted SNP. This, said the academic, raises questions over the accuracy of the survey which has received headline coverage in the Scottish media.
Writing in his blog, Professor Curtice highlighted other ‘issues’ which he claimed may have led to a rise in the number of so-called ‘Don’t Knows’. Other polling organisations have failed to record any significant increase in the totals of those yet to decide.
It also emerged that TNS had introduced a subtle change to the wording of their question, which the academic claimed might explain the apparent increase in ‘Don’t Knows’.
Previous TNS polls had contained an intro to the referendum question which read: “If this referendum were to be held tomorrow, how would you vote…”
The latest poll however asked: “How do you intend to vote…”
Writing on his blog, Professor Curtice said: “The new formulation invites people to forecast what they will do in a year’s time rather than what they would do now, and perhaps we should not be surprised that rather more people are less certain about the former than the latter.”
In contrast to other polling organisations, TNS employed a methodology that involved knocking on people’s doors and asking them to participate. This, suggested Professor Curtice, may well have led to an increase in those who do not follow politics closely and who are less likely to actually vote.
The academic said: “All of the other recent polls were conducted over the internet with people who have volunteered to fill in that company’s surveys when asked, whereas TNS’s was conducted by knocking on people’s doors and persuading them to take part.
“Unsurprisingly the latter approach tends to be more successful at securing the participation of those with little or no interest in politics. And in line with that pattern, whereas Panelbase found that 76% of their sample said they were certain to vote in the referendum, TNS estimate the proportion at just 62%.”
Recent polls have veered wildy with a YouGov poll suggesting Yes trailing its No rivals badly – a thirty point gap opening up (No 59 – Yes 29). However a Panelbase survey published one day later turned this on its head and indicated Yes ahead marginally, 44% to 43%.
The Panelbase poll was commissioned by the SNP and also suggested those intending to vote Yes would do so regardless of whether Labour or the Conservatives led the next UK government.
There will now be growing suspicions that pollsters themselves are unwittingly producing results favoured by the respective sides in the constitutional debate rather than revealing actual trends amongst the electorate.