On the myth of reported precision in public opinion polls

Abstract

Many public opinion polls report a mythical margin of error. These reported margins of error are calculated on the basis of a nonexistent (fictitious) dichotomous poll question and are completely unrelated to the actual question responses in a poll. Because it can convey a false sense of poll precision, a mythical margin of error can lead to unsupported and even incorrect inferences regarding the precision of a public opinion poll. This research note documents the existence and implications of mythical margins of error and offers suggestions for addressing them.