A New Normative Ideal and Empirical Examination of Citizens’ Opinion Stability

• In Event: Methodological Innovations in the Study of Public Opinion

Abstract

*** Beyond black and white: A new normative ideal and empirical examination of citizens' opinion stability ***



Citizens' lack of ability to form stable opinions have frustrated political scientists for decades. The ideal opinion, as expressed by Converse (1964, 49) is "well crystallized and perfectly stable over time." Empirical examinations of opinion stability, however, consistently show that citizens fall well short of this benchmark. This seeming erratic quality of citizens' opinions raises important questions about citizens' political competence and whether they are fit for their democratic responsibilities.

In this paper, we argue that the normative ideal of opinion stability partly misguided, and that unstable opinions do not necessarily indicate a lack of voter competence. Building on Zaller and Feldman's (1992, 612) idea that "most opinions on most issues have both a central tendency and a variance", we propose a new normative ideal: whether citizens are able to quantify their own opinion ambiguity. This contrasts with the current ideal that citizens should be absolutely certain about what they think about all the issues of today and that they should think the same way tomorrow. Instead, we ask that they should be able to characterize how uncertain they are of their opinions, and that their opinion stability is predicted by their self-reported opinion ambiguity.

In order to test the core idea that citizens’ opinions both have a mean and a variance, we construct a simple but novel survey measurement which allows citizens to express their opinion ambiguity. Specifically, we follow up the conventional Likert scale opinion question with an almost identical question that allows the respondents to indicate all response options that represent their view to at least some extent. Using a panel survey with over 1000 respondents and multiple waves spanning a year, we show that individuals have more stable opinions on issues where their initial opinion ambiguity is lower. We also show that our measure can not be reduced to regular “uncertainty” and “importance” indicators, and that measurement error is not a plausible alternative explanation for our results. Our results imply that opinion instability in large part arises from meaningful uncertainty over different issues. In this sense, opinion instability does not imply non-opinions nor random responding.

Our paper makes several important contributions to the public opinion literature. Theoretically, we show how conceptualizing opinions as distributions has important implications for the normative ideal of opinion stability. Empirically, we show that we can empirically characterize citizens' opinions not only as point estimates but as distributions and that our measure of attitude ambiguity is a strong predictor of opinion stability. Yet, the perhaps most important implication of this paper is that citizens will only appear as competent as our measurement strategies allow. Consequently, when we move away from very simplistic ways of measuring citizens' political opinions and allow citizens to express their opinions in a more nuanced manner, citizens' views also appear more nuanced and sophisticated.


References

Converse, Philip. 1964. The Nature of Belief Systems in Mass Publics. In Ideology
and Its Discontent, ed. D.E. Apter. New York: Free Press of Glencoe pp. 206–261.

Zaller, John and Stanley Feldman. 1992. “A Simple Theory of the Survey Response:
Answering Questions versus Revealing Preferences.” American Journal of Political
Science 36(3):579–516.

Authors

• Mattias Agerberg, University of Gothenburg

• Love Christensen, Aarhus University