AN ANALYSIS OF THE ACCURACY OF "TRIAL HEAT" POLLS DURING THE 1992 PRESIDENTIAL ELECTION

There is little systematic research on the multitude of factorsthat influence the accuracy of poll results. This article examinessix methodological factors directly under the survey researcher'scontrol and two exogenous factors concerning the nature of publicopinion as sources of survey error. Data for this study comefrom 56 "trial heat" polls conducted during the last month ofthe 1992 presidential election. The most important variablesinfluencing survey accuracy were the number of days a poll isin the field, which increased total accuracy one-half of a percentagepoint per day; conducting interviews only on weekdays (and thusonly during evening hours), which reduced overall accuracy ratesby more than 1 percentage point; and conducting a "tracking"poll, which increased accuracy by about 1.5 points. Sample sizewas not related to accuracy rates. Results also indicated thatsampling frames of "likely voters" (relative to "registeredvoters") tended to overestimate support for George Bush andunderestimate support for Ross Perot, that interviewing onlyon weekdays led to overestimates of support for Bush, and thatstrict methods of defining a respondent as "supporting" a candidatehurt the two newcomers, Perot and Bill Clinton, more than Bush.In light of these data it is recommended that the common practiceof reporting "margins of error" based solely on sample sizesbe abandoned as misleading and replaced by a more empiricallyjustifiable measure based more on response rates.