Polling errors could decide Swedish election
The Local · 11 Apr 2006, 18:20
Published: 11 Apr 2006 18:20 GMT+02:00
Unrepresentative opinion polls could decide the outcome of the Swedish election in September, according to two political scientists who have investigated the way polling companies measure voters' intentions.
The Swedish rule that a party must win 4% of the vote in order to take its seats in parliament means that tactical voting could play a role in the results, particularly for the conservative opposition parties.
But when Sören Holmberg and Olof Petersson investigated the way the pollsters conduct their interviews they found several failings - and that means that tactical voters could be basing their actions on incorrect information.
It is not uncommon for 50% of those who are meant to be interviewed by telephone to be out, or to refuse to take part in the survey. The polling companies adjust the figures to account for this drop-off rate but Holmberg and Petersson say that the end result is a tendency to underestimate the support for the Christian Democrats and to exaggerate the support for the Moderates, two of Sweden's opposition parties.
For the Christian Democrats, and other parties which are currently hovering around the 4% mark, these estimates could be critical.
If conservative voters think that the Christian Democrats are under the 4% limit, they are likely to tactically shift their support from the more popular Moderates to help bring the smaller party into parliament and ensuring a conservative majority.
But if the poll evidence suggests that the Christian Democrats are safely above the 4% limit, voters will stick with their usual party - and if that decision is based on incorrect information, the result could see the Christian Democrats out of parliament.
"It's very important for the small parties to get an accurate idea of how things are, because that influences many people with their tactical votes," said Sören Holmberg.
Both he and Olof Petersson admit that it has become harder to measure party support before elections. Increasing numbers of voters are deciding late, perhaps even as late as the day of the election, which party they will vote for.
But both men said that the variations in voters' intentions are not as wild as the polls would suggest.
"I think that has more to do with uncertainty in the measuring. Today they're measuring a bird in flight instead of one sitting still. That's harder," said Holmberg.