Predicting an election when voters won’t tell you their voting intentions has been a challenge in Sweden.
UKIP are currently enjoying great success in the polls and judging from their experience at the recent European elections, the polls could be a strong indicator of what’s to come. Accuracy of polls is paramount to their existence but it is not a given. A lot of it has to do with trust. If people do not perceive polling institutions as neutral and objective, then they won’t necessarily be completely honest when interviewed.
The major headache for all pollsters in Sweden at the last election was the Sweden Democrats (SD). SD is a far right party, deeply skeptical of immigrants. Founded back in the 80s, their first success came in 2006 when they got 2.9 percent of the vote. This was not enough to gain access to parliament but it did attract serious attention.
In the 2010 election, they got 5.7 percent of the vote and with that over 20 seats in Parliament. Last year, they achieved 12.9 percent and 49 seats. Obviously the SD has grown enormously in the last couple of years but at the same time pollsters have continued to underestimate them. United Minds (Weber Shandwick’s sister agency in Sweden) has historically been the pollster that has had SD at the highest levels, yet in both 2010 and 2014 even these predictions turned out to be an underestimate.
One explanation for this could be what is known as the Bradley effect, so named after the Los Angeles African-American mayor Tom Bradley who lost out in the 1982 California Governor’s race despite being ahead in the poll. This was put down to the fact that some voters claimed to be voting for him so as not open themselves up to criticism of racial motivation. The Bradley effect in short says that voters are inclined to lie in polls as they know that some choices are not socially accepted. This has been seen as the ‘shy Tory’ phenomenon in the UK, for which pollsters now adjust. The phenomenon should only arise when voters are being interviewed by an actual person (i.e. in a telephone interview). United Minds has pioneered online surveying in Sweden and applied this extensively to political polling, which could be one explanation as to why our predictions were more accurate. Voters responding to online polls may be less inclined to lie about their voting intentions.
That said, this has been a major discussion point in the polling community post-election, not least because even the online polls underestimated the SD. Hence, the Bradley effect matters even when there is not another person asking the question. Thus, the theory is that voters that plan to vote for SD do not trust polling institutions and thus to some extent lie in order not to risk social stigma, no matter how the interview is being conducted. This – of course – is very troubling for anyone who wants to take the temperature of Swedish politics or of any other countries. For those tracking the rise of UKIP in the UK, this could be an interesting trend, particularly if it becomes apparent that voters are ‘shy’ of declaring their support for UKIP.