With the independence referendum now one year away, the first Scotland on Sunday/Scotsman/ICM poll shows that the No camp have decent sized lead (Yes 40%, No 60%) but which is by no means unassailable, something that supporters of the 2011 Yes to AV! Campaign will no doubt confirm.
The referendum argument’s will likely be won and lost on perceptions for Scotland’s future economic performance, and impact on household budgets. If people could be convinced that an independent Scotland could thrive – in doing so making people £500 a year better off as a result – support for independence leaps to a victorious 56% yes, 44% for no. Even if people thought their finances would be unaltered it would be all to play for (yes 47%, no 53%), but given a household deficit of £500 the yes camp could forget (yes 22%, no 78%).
Which begs the question: what do people think would happen to the Scottish economy if the country were independent? The headline findings of the poll imply what the answer is: a No lead indeed is underpinned by a view that independence would be bad for Scotland’s economy (48%) as opposed to good for it (31%). This is a critical measure of support and perhaps the main one for both sides to focus on over the next year. Until Yes can move the number on this, it’s unlikely they will move their overall support in the direction they want.
Weighting a Scottish poll.
The question of how pollster should make an online sample of Scots politically balanced, as well as demographically representative of the profile of all Scots is taxing to say the least, evidenced by the different approaches adopted by all the polling firms who have already joined the Scottish independence polling fray.
There are a number of decisions for a pollster to ponder: firstly a basic one: does the online sample contain the right number of people by demographic sub-group, something that few pollsters would fail to correct – although different information sources can yield slightly different demographic profile targets to set. More important though, is the extent to which pollsters want to tie the data back to a prior election result – to make the sample politically representative. On this there is much disagreement – some pollsters don’t believe it’s right to weight to prior election results at all, because some people can falsely align their past votes to their present voting intentions, thus ‘forgetting’ what they actually did a year or two ago – this would imply that pollsters are setting incorrect recall targets.
Others though, particularly ICM, believe it is imperative to weight politically. The spiral of silence theory suggests that people can avoid telling pollsters what they intend to do, saying they ‘don’t know’ how they might vote when really some of them do know but won’t say as they perceive the party they support to be unpopular. Once called ‘Shy Tories’ they have manifested themselves as ‘Shy Labour’ in the dog days of Gordon Brown’s premiership, and we might suspect that Liberal Democrats currently have their own particular level of bashfulness. This phenomenon was behind the 1992 failure of the polls to predict a Conservative victory one day before it happened.
In accepting a need to weight polling data to an election result, we must consider which one to weight to; in Scotland we have two recent – but very different - election results to refer back to. In 2010 the General Election saw Scottish Labour secure twice as many votes as the SNP, while a year on, the Scottish Parliament first vote reversed the situation, with the SNP taking 12-points greater share than Labour. Given an obvious implication that SNP supporters are disproportionately likely to vote Yes, the chosen weighting scheme could have a major impact on the results.
But it doesn’t. At least, not in the case of this poll.
ICM has tested the data by weighting it in a number of different ways – to both 2010 and 2011 election results – to find precious little movement in the headline findings. The poll is internally consistent – securing approximately 60% support for the No campaign irrespective of how we weight it. Given this, our approach is simply defined in terms of the quality of recall produced for each election. The Holyrood election was more recent, and the recall of voting in it much more closely matches the actual result, compared to the 2010 General Election recall. This made the decision to tie our poll back to the most recent Holyrood election (Constituency vote only) an easy one to make, but ICM will continuously monitor this as we track views on independence over the next year, and will always attempt to make the poll as reliable as possible.