Big difference in county polls
The Polls are so diverging that it is hard to say anything specific regarding the popularity of the parties in each county, says Johan Giertsen in Poll of Polls.
– The county polls point in different directions and convey completely different messages, says Giertsen to NTB.
He describes the situation in a post on the Poll of Polls website and gives three explanations as to why the picture is so unclear, all related to particular circumstances at one Poll station, namely InFact.
A current average from eight counties made by Response, Sentio and Market Info provides a majority of 86 mandates for Frp, Conservatives, KrF and Liberals against 80 for the three red-green alternative.
But the polls made by InFact in 15 counties in August paints a completely different picture. There, AP, SV and Sp have a solid majority of 91 mandates against 74 for the four bourgeois partners.
– How the situation really is, is unclear. It really down to which poll you rely the most on, says Giertsen.
According to the law professor and the statistics enthusiast, the fact that the picture is so unclear relies on three facts: InFact weighs their responses to the municipal elections in 2015 higher than the parliamentary elections of 2013, InFact relies on automated calls instead of telephone interviews and that the agency has a low proportion of undecided among their respondents.
– These skeins are also found in the polls carried out on a national basis, but the differences are much more pronounced in the county polls. Making an average of all county polls has little value, because the preconditions are so diverging, he says.
For the strategically important western counties Rogaland and Hordaland, the discrepancies are particularly large.
– It’s only InFact that measures the Conservatives with a big decline in some counties. If we remove InFact’s polls, results from county polls are more in line with the average for the country as a whole, says TV 2 journalist and numerator, Kjetil Løset.
The theme was highlighted when Prime Minister Erna Solberg (H) on Monday participated in a poltical debate organized by Dagbladet. She was then confronted with results for the Conservatives in her hometown of Bergen.
– There are two different institutes that have polled in Bergen. The one showed us with 29.9 per cent support, the other at 21.5. We are probably be somewhere in between, Solberg said , pointing out that InFact, weighs heavily against 2015.
Giertsen has on several occasions urged InFact, which supplies measurements to VG and Amedia, to show openness about its raw data. Such insight could help clarify the causes of the large-scale agencies, Giertsen says
He has urged VG to submit the figures, but the newspaper says it is up to InFact to decide this.
– We will reassess this together with VG after the election. But we are not alone in the industry to refrain from delivering raw numbers. The reason is that we can not be good at how these numbers are being used, says head of department in InFact Knut Weberg to NTB.
The InFact boss believes it is right to weigh the results against the previous election simply because people remember better what they voted two years ago than four.
He rejects that the use of automatic calls makes InFact’s polls less credible and indicates that the polls from 2015 were close to the final outcome.
– We are looking forward to the Election Day September 11. When the exit poll is coming, we will see with what precision we hit the estimate, Weberg says.
© NTB Scanpix / Norway Today