Abstract
Three surveys compared the accuracy of predictions based on forced-choice and subjective probability scales. The latter produced significantly more accurate election predictions and significantly reduced the percentage of undecided, or “Don't Know” responses, compared to forced-choice scales in all three surveys. Analysis indicates subjective probability scales decrease sampling error and confirms there is an inherent source of error in traditional forced-choice questions about voting intentions not attributable to sampling error. The results are discussed with respect to (1) sampling and measurement errors in forced-choice and subjective probability scales measuring behavioral intentions, (2) their practical application, and (3) cognitive theory, especially support theory.
Get full access to this article
View all access options for this article.
