Why are political opinion polls getting it wrong more often? (2)
I wrote a blog posting recently about the growing difficulty for opinion pollsters in forecasting accurately the result of elections. I particularly referenced the recent failure of all the pollsters in Australia to forecast the victory of the National-Liberal coalition.
I wrote: “I think that what we are seeing is more voter fluidity. Class used to be the major determinant of voting behaviour and class does not change quickly, but class seems no longer to be the dominant factor that it was. Voters seem to be more willing to change support from election to election and even, in the course of the campaign, from week to week and day to day.”
We now have more evidence of this – least as regards Australia – from a new poll asking voters when they decided how they would cast their vote.
Almost half of voters, 48%, had made their choice about which party they were voting for well before the election was called. But, more than a quarter of voters in the sample, 26%, had not yet made up their minds as the federal campaign entered its closing weeks. That number was still 11% by polling day, with those voters making their decision on the day they cast their ballots.
There may be a special factor at play in Australia which has mandatory voting. Maybe, in countries without mandatory voting, those who have still not made up their mind by polling day do not actually vote. But I suspect that in many countries a significant proportion of voters only make a decision once they go to the polling station.