The Pro-Incumbent Bias in the 1982 National Election Study

In this article we demonstrate that there is a systematic bias in the 1982 National Election Study (NES). The analysis shows that the NES survey instrument generates a context effect that prompts respondents to report they voted for the incumbent in House elections. The bias inflates the survey estimate of the vote for incumbents by about 9 percentage points. In an earlier study we found the same phenomenon in the 1978 and 1980 NESs. The nonrandom nature of the bias, and its persistence in three consecutive NESs means that researchers may attribute substantive meaning to an effect that is created by the NES survey instrument.