Measuring expectations: Comparing alternative approaches

The paper compares three alternative approaches employed by the Canadian Election Study to measure voters' perceptions of parties' chances of winning in their local constituency. The first approach, used in 2000, consists of asking respondents to rate parties' chances on a 0 to 100 scale in a random sequence. The second, used in 2004, entails first asking whether each party had a chance of winning and then inviting people to rate the chances. In the third approach, adopted in 2006, respondents are first asked which two parties had the best chance of winning and, then, if any other party has a chance, before requesting that they rate the mentioned parties. By comparing “objective” and “perceived” chances of winning, the paper concludes that the third approach provides a more valid measure of voters' expectations. The paper discusses the implications for the measurement of expectations in different types of electoral systems.