THE IMPACT OF PARTISAN SPONSORSHIP ON POLITICAL SURVEYS

Although it is widely believed that revealing the topic or sponsor of a survey to potential respondents can produce large nonresponse biases, measurement errors, or both, recent research has shown that the effects of the framing of the survey request are often quite modest. However, research has not experimented with partisan or candidate sponsors, which are most likely to produce large effects. We carried out three experiments with political surveys that varied whether a partisan sponsor was identified (and in one of the experiments also varied the identity of that sponsor), expecting that this would have dramatic effects. Instead, all three studies found only minor effects. Thus, even in the context of partisan election surveys, sponsorship may not be the powerful cue it is often thought to be.