Tag Archives: Polling

On opinion polls

Posted by Euan Bennet on 11/02/2013

Much has been written, and will be written, on the interpretation and analysis of opinion polls. The constant caveat that must always be kept in mind is that polls provide a snapshot of the electorate at a particular instant, and the samples are unlikely to be truly representative. Even a poll with sample size 1,000 is only expected to be within 3% of the truth when extrapolated to the entire electorate. Each individual polling company has their own nuances and ways of interpreting the data, which adds further complications when comparing different polls. Referendum polling particularly suffers from methodological difficulties since none of the polling companies are familiar with one-off votes, and as such still depend heavily on party identification to weight responses: a practice that is as best risky and at worst outright misleading.

Ultimately as the saying goes, the only poll that matters is the one on the day of the election/referendum. The only real value of opinion polls is to look at trends over time.┬áThe mainstream media doesn’t really understand this, hence why we have had more than a year of headlines saying “polls static!”, “Support for Independence stuck at around 1/3!” etc etc. This suits their narrative but it doesn’t really resemble reality. Here is reality (in the polls!):

Image

The above figure was prepared by “scottish_skier”, one of the regular commenters on Wings over Scotland, see his comment in context here. You can judge for yourself if that looks anything like what the media have been reporting over the past few months.

Polls taken since October are particularly interesting, since the fieldwork would have been done after the publication of the Scottish Government’s White Paper. The trend since then has been only in one direction: Yes rising and No falling. As noted on Wings over Scotland, the notable thing that’s happening at the moment is that all polling companies are converging to show a gap of around 14%, plus or minus 3% – if accurate then a 7% swing is all that is required for Yes to take the lead. In my opinion the polls are way behind reality anyway, and canvassers are reporting far more optimistic results from the ground. But when the polls finally reflect this, we really will be living in interesting times. Given the panic-stricken tone of the No campaign and the mainstream media when No were “winning”, I wonder where they can possibly go next.