Suchergebnisse

The doctoral dissertation often shapes the career that follows it, influencing both opportunities encountered and research conducted. This article describes the ways this has been true for me and then argues that, given the dissertation's importance, graduate programs do not focus sufficiently on strategies for conceiving research. As a result, many students flounder at the dissertation proposal stage. Drawing on the role of doubt in my career and in science more generally, I propose changes in doctoral programs to reduce the problem.

A report on the results of two split-sample experiments designed to examine whether question context can reduce vote overreporting in surveys. The first experiment (N = 964 adults) tested the effect of preceding the turnout item with a question about the location of the polling place. In a test of the hypothesis that if respondents who overreport do not know where their polling place is, they should be less apt to report incorrectly about turnout after being asked the location of the poll. The second experiment (N = 705 adults) tested the effects of preceding the vote item with a question about lifetime electoral behavior: if inaccuracy is due to respondents' wanting to present themselves in a favorable light, the opportunity to report past or usual good citizenship should reduce the pressure to claim participation in the last election. Findings fail to support either hypothesis. 3 Tables, 12 References. AA

The assumption that reporting errors are uncorrelated across survey items was tested using data from the 1949 Denver (Colo) Community Survey. R reports (N = 920) to 14 questions in that study were later validated with official records. Inaccuracy was found to be item-specific for questions about 7 generally unrelated subjects. By contrast, for 7 questions on electoral behavior, all of which were significantly associated, the assumption of uncorrelated errors was clearly violated: Rs inaccurate on 1 of the 7 were disproportionately inaccurate on each of the other 6. This held for groups defined by sex, age, education, & political interest. The correlated error term, combined with a tendency for levels of inaccuracy to be greater for those who express higher political interest, substantially increased the size of the r between voting & political interest. The implications of these results for survey research are discussed. 5 Tables, 1 Appendix, 19 References. Modified AA.