The data analyzed in this paper are from the National Election Studies 1998 Pilot Study. Neither the principal investigators (Virginia Sapiro, Steven J. Rosenstone, and the National Election Studies) nor the producers and distributors (the Center for Political Studies, University of Michigan) bear any responsibility for our analyses and interpretations.
Why Is Research on the Effects of Negative Campaigning So Inconclusive? Understanding Citizens’ Perceptions of Negativity
Version of Record online: 5 FEB 2003
Journal of Politics
Volume 65, Issue 1, pages 142–160, February 2003
How to Cite
Sigelman, L. and Kugler, M. (2003), Why Is Research on the Effects of Negative Campaigning So Inconclusive? Understanding Citizens’ Perceptions of Negativity. Journal of Politics, 65: 142–160. doi: 10.1111/1468-2508.t01-1-00007
1 For year-by-year results, see http://www.umich.ed/~nes/nesguide/toptable/tab6d_7.htm (accessed March 10, 2000).
2 Inter-coder agreement was 95%, and kappa, a measure of the extent to which the observed inter-coder agreement improved on chance agreement, was .91 (s.e.=.09, z=4.5, p < .05).
3 The efficacy items included three randomly ordered statements with which respondents were asked to agree strongly, agree somewhat, neither agree nor disagree, disagree somewhat, or disagree (“People like me don’t have any say about what the government does”; “Sometimes politics and government seem so complicated that a person like me can’t really understand what's going on”; and “Public officials don’t care much what people like me think.”) and two other questions related to their sense of political efficacy (“How much do elections make government pay attention to what the people think: a good deal, some, or not much?”; and “Over the years, how much attention does the government pay to what the people think when it decides what to do: a good deal, some, or not much?”). All five items were coded such that higher scores indicated greater efficacy. Preliminary factor analyses were suggestive of unidimensionality, and coefficient alpha for the scale was .66. The folded party identification measure was based on the standard seven-point branching question (“Generally speaking, do you usually think of yourself as a Republican, a Democrat, an independent, or what?,” followed by either “Would you call yourself a strong Republican [Democrat] or a not very strong Republican [Democrat]?” or “Do you think of yourself as closer to the Republican party or to the Democratic party?”). Pure independents were scored 0, independent leaners 1, weak identifiers 2, and strong identifiers 3. To explore the possibility that independent leaners would have more in common with strong identifiers than with weak identifiers (Keith et al. 1992), we also tried coding the partisanship categories as separate dummy variables, but this yielded results virtually identical to those reported below.
4 The survey question was: “Who do you think you will vote for in the election for governor where the candidates are [the candidate's name], the Democrat, and [the candidate's name], the Republican?” In the model, the omitted or reference category consists of backers of the eventual winner.
5 The political information index was based on five items. The first four tapped interviewees’ command of basic political facts (“Which branch of government decides whether laws are unconstitutional?”; “Which branch of government nominates judges to the federal courts?”; “Which party has more members in the House?”; and “Which party has more members in the Senate?”), with one point awarded for each correct answer. The fifth item was the interviewer's assessment of the interviewee's level of information about politics and public affairs, rescaled to range from 0 to 1. To form the scale, the five items were summed and divided by 5 so that scale scores would range from 0 to 1. The results of a factor analysis of the five items were consistent with our treatment of them as unidimensional. Coefficient alpha for the scale was .68.
6Berry and Mielke's (1992) Index of Ordinal Variation (IOV) ranges from 0 (when all the cases are concentrated in a single category) to 1 (when all the cases are evenly divided between the two polar categories). Thus, 0 would signify total consensus. For the three distributions shown in Figure 1, IOV is between .56 and .60—nowhere near 0.
7 See Houston, Doan, and Roskos-Ewoldsen (1999) for a comparable experimental result.
- Issue online: 5 FEB 2003
- Version of Record online: 5 FEB 2003
Options for accessing this content:
- If you are a society or association member and require assistance with obtaining online access instructions please contact our Journal Customer Services team.
- If your institution does not currently subscribe to this content, please recommend the title to your librarian.
- Login via other institutional login options http://onlinelibrary.wiley.com/login-options.
Login via OpenAthens
Search for your institution's name below to login via Shibboleth.
Registered Users please login:
- Access your saved publications, articles and searches
- Manage your email alerts, orders and subscriptions
- Change your contact information, including your password
Please register to:
- Save publications, articles and searches
- Get email alerts
- Get all the benefits mentioned below!