Hade Person rätt om opinionen?

Kan det vara så att det finns bättre metoder att förutspå valresultat än de opinionsinstituten använder? Kan det vara så att vadslagningen om valresultaten på spelhusen ger en mer exakt bild? Kan det vara så att en grupp människors förutsägelser om hur det ska bli faktiskt är mer korrekt än en mätning av hur de säger att de ska rösta?

Hade Göran Persson rätt när han berättade om sina oortodoxa sätt att avläsa opinionen genom statyerna på Sagerska palatset och konsum-parkeringen i Katrineholm?

Ja, jag vet inte. Men psykologiprofessor Lennart Sjöberg på Handelshögskolan har precis börjat nosa på frågan, och släppt ett working paper.

Abstract:

In this study, 4 groups of people made prognostic judgments of the outcome of the Swedish Parliamentary election in the fall of 2006, about one week before the election. The groups consisted of members of the public (N=123), political scientists (N=53), journalists writing about domestic politics in Swedish daily newspapers (N=32), and journalists who were editing sections of readers’ letters in daily newspapers (N=10). They rated, for each of seven political parties, which percentage of the votes that they believed they would get in the election. They also marked which party they themselves preferred, and answered to a few questions about interest and competence. Data were then obtained on the outcome of the election, and on the two opinions polls closest in time to it.

It was found that all four groups did reasonably well, when average prognostic judgments were compared to the outcome of the election, and better than the opinion polls.

The two last polls overestimated the span between the incumbent government and the victorious opposition by a factor of 2. Wishful thinking was assessed by comparing prognostic judgments for each respondent’s preferred party with his or her judgments of other parties. All groups showed some wishful thinking; the political scientists least and the public most. There were large and consistent individual differences in prognostic ability. Men performed better than women, as did those who expressed more interest and knowledge in politics, but neither level of education nor confidence in making the judgments correlated significantly with performance.

Kanske kan vara läs- och tänkvärt.