Intentions vs expectations: A better predictor of outcomes

A study by economists David Rothschild and Justin Wolfers looked at more than 50 years of data on voter intentions and voter expectations for US presidential races before 2008. They identified times when the voter intention and expectation questions predicted different candidates and tested which one was more accurate. It turned out that the question on voter expectations was correct in 78 per cent of cases when expectations and intentions disagreed. When researchers looked at Australian elections, they found polls of expected winners were right nine times out of ten, whereas surveys of voting intention only picked the winner four times out of ten. So why is asking people the question "who will win?" a more powerful predictor than "who will you vote for?" The researchers argue it's because asking voters for a prediction has the effect of greatly broadening the sample. It allows survey respondents to consider not only their views but also the attitudes of those around them. It's like taking "a personal poll of approximately 20 friends, family, and co-workers". (VIEW LINK)