As we enter 2015, electioneering in the UK has already started, up front of May's general election. Is this going to be a difficult time for the pollsters, especially following on from their perceived performance in predicting the outcome of last September's referendum in Scotland on independence?
All the signs point to a complex situation, with the possible annihilation of the Liberals; the rise of UKIP; the increasing support for the SNP in Scotland, possibly causing Labour a lot of grief; the role of the Green vote. Forecasting the likely outcome looks to be more problematic than has been the case for many years.
One issue that was discussed in the context of the polls conducted leading up to the referendum vote in Scotland last autumn was the impact of the 'spiral of silence' which may have contributed towards the 'no' vote having been understated in the polls. So, what exactly is the 'spiral of silence' effect, and are there ways in which pollsters can try to account for its influence when predicting voting intentions?
Luckily, we have published a paper examining this phenomenon, authored by that notable pollster, and twice winner of the MRS Silver Medal, Nick Sparrow (then at ICM) in collaboration with John Turner (Oxford Brookes University) – a paper that won the MRS Silver Medal in 1996.
The authors describe the origin in social psychological analysis 'of the way in which human behaviour exerts an influence on people through the general desire for approval, especially in relation to reference groups and within the context of a particular socio-political environment'. As the authors describe, citing evidence from other studies, 'individuals will refrain from expressing their perceived minority views in the face of a hostile climate of opinion'.
Sparrow and Turner then consider how this effect plays out in the context of measuring voting intentions, especially when individuals feel that their views are in conflict with what they perceive to be the prevailing social consensus – for example, being seen as unpatriotic by expressing an intention to vote for any party other than the SNP. In such situations people turn elsewhere for 'help' and become more responsive to messages in the media – the reticence to discuss political issues during the coal dispute in 1984 being the subject of one cited study in this field of research.
As the authors describe, this effect unwinds as the day for casting the vote approaches, citing evidence from the infamous 1992 situation where the polls got it badly wrong. The paper contains a detailed analysis of ICM polling data to identify the effects of this issue, and describes ICM's method for accounting for the effect. They present evidence from subsequent polling occasions, and how party policies come into play. Finally, they discuss the implications for the marketing strategies of political parties, advocating programmes of market testing for policies as used by consumer goods and service companies in developing their offerings.
It does make me wonder what the impact of this effect might be in France on attitudes to free speech in the aftermath of the 'Charlie Hebdo' shootings. There is also the promised referendum in the UK on EU membership if the Conservatives are in power after the election which is likely to generate intense emotions that may also generate a spiral of silence effect.
If you want another, very different, perspective on this topic, read 'The spiral of silence in election campaigns in a post-communist society: the case of Belarus', Oleg Manaev et al, IJMR Vol.52 Issue 3, 2010. This describes the legacy of authoritarianism and its continuing influence on expressing opinions during election campaigns.
Finally, there will be a debate on polling chaired by Deborah Mattinson at this year's MRS annual conference in March. For further details click here.
This post was first published on the International Journal of Market Research website.