Predicting No Opinion in the Polls |
| |
| Authors: | Converse Jean M. |
| |
| Affiliation: | Jean M. Converse is a Research Associate in the Survey Research Center, Institute for Social Research, University of Michigan. An earlier version of this article was presented informally at a roundtable discussion of the Thirtieth Annual Conference of the American Association of Public Opinion Research, 1975. This research was carried out within the project on Survey Question Wording, supported by grants from the National Science Foundation (SOC 73-05602) and the National Institute of Mental Health (MH 24266). The author is indebted to Howard Schuman, director of the project, for advice and encouragement, and to Laura Klein, Jane Fountain, Stanley Presser, and Charles F. Cannell for their good counsel on statistical and editorial matters. |
| |
| Abstract: | On poll questions, levels of expressed public ignorance or indifference—NoOpinion or Don't Know—can be explained in some part bycertain properties of the questions pollsters ask, althoughthe educational level of respondents is the single best predictor.No Opinion levels are analyzed in two large sets of recent pollquestions published by Gallup and Harris. A measure of the languagecomplexity of the questions shows no relationship to DK. Ofthe three other question predictors assessed, it is questioncontent which best illuminates levels of No Opinion in bothpolls and points to some unique characteristics of each. Theimportance of question content is demonstrated in two additionalsets of Gallup and Harris data. The more difficult kind of questioncontent dominates in all four sets of poll questions examined. |
| |
| Keywords: | |
| 本文献已被 Oxford 等数据库收录！ |
|
