Determinants of the opinion gap between the elites and the public in the United States*

ABSTRACT

Recent scholarship indicates that elites possess disproportionate power in the policy-making process in the United States. The degree to which elite preferences trump the preferences of non-elite Americans raises questions about American democracy, and even indicates the nation exhibits oligarchic tendencies. This paper seeks to further our understanding of when or how elite preferences differ from those of the general public. We utilize the unique survey data that ask identical questions both to the elites and to the general public, and present a quantitative model in which the opinion gap between elites and non-elites is the dependent variable. Our results indicate that elites are particularly likely to diverge from the rest of the population on issues related to economic and domestic policy. The preference gap is typically smaller on issues related to international affairs.

KEYWORDS:

• Public opinion
• Elites

Notes

¹ But he did find that this consistency decreased after 1980.

² In terms of actual policies, Bartels (2008) found that on four key roll-call votes dealing with the minimum wage, civil rights, and the budget, Senators were influenced by the views of high-income constituents and but not low-income constituents. He found similar results when examining key votes on abortion policy.

³ According to Ellis (2012), although income is correlated with other attributes such as race and education, money is the best predictor of whether a person’s views will be translated into policy.

⁴ An example survey question for the economic issue is, “Do you favor or oppose the US government giving subsidies to small farmers, who work farms less than 500 acres?”

⁵ An example survey question for the domestic issue is, “Officials in Washington have proposed a plan that would allow foreigners who have jobs but are staying illegally in the United States to apply for legal, temporary-worker status. Do you favor or oppose this proposal?”

⁶ An example survey question for the foreign issue is, “Before the U.S. withdraws from Iraq, do you think it is or is not necessary to have a democratically-elected government?”

⁷ A notable exception would be Oldendick and Bardes (1982). They argue that people tend to have strong preferences on domestic and economic issues as opposed to on other issues.

⁸ One can easily make an argument for the very opposite prediction. For example, if the public generally cares more about domestic and economic issues as compared to foreign policy issues, they may pay more attention to these policy areas. The public will also pay more attention to how elites (and representatives in governments) represent their opinions. In this case, similar to our issue importance argument, there will be a smaller gap between elite and non-elite opinions for domestic and economics issues than for foreign policy issues.

⁹ Unfortunately, the CCFR has not conducted such a survey since then, so the 2004 survey is the most recent data. The Appendix provides detailed information on the two surveys. However, refer to http://www.thechicagocouncil.org for more information on the CCFR survey and the council in general (they have changed their name to Chicago Council of Global Affairs in 2006).

¹⁰ In this sense, the meaning of elites here is not always same as that in the previous literature. Moreover, one can wonder whether or not different types of elites (e.g., policymakers vs. other elites) show distinct patterns regarding opinion gap. While it would be both interesting and theoretically relevant to analyze definition of elites, it is not technically feasible here. CCFR does not provide respondent-level survey results, so we are not able to aggregate the survey responses at different levels. We leave it for future research.

¹¹ We exclude “don’t know” and “not sure” responses not only because it can lead to potentially biased results but also because the proportion of “don’t know” and “not sure” responses varies with the 2002 and 2004 surveys.

¹² The reason why we take the absolute value is straightforward: the direction of opinion gap does not matter. For example, the negative opinion gap for controlling and reducing illegal immigration as an important foreign policy goal is exactly same as the positive opinion gap for favoring for allowing illegal immigrant workers to apply for legal, temporary-worker status.

¹³ Of course, there are alternative ways to measure issue importance. The first alternative is to assign scores for each question item in accordance with the percentage of respondents who feels the given issue as the biggest problem facing the country today. But, the problem is that we have this survey question only in 2002. In addition, even in the 2002 data, there are too many answers in a very few particular issues. The second alternative is to make an index from the media coverage for a group of issues. The problem here is that it is not entirely feasible for this research project.

¹⁴ One could argue that elites are uncertain about the issue because it is less important, thus raising the multi-collinearity concern. In the data, there is certainly a negative correlation (r = −0.23) between the two variables, but this is not the level that requires a statistical attention.

¹⁵ Appendix A also provides 30 more survey questions that have opinion gap of 20–30% points in 2004.

¹⁶ A direct comparison between our analysis and theirs is somewhat tricky. Nevertheless, we can recognize the overall trend from their analysis and compare this general trend with the 2004 case. The fact that our paper and their analysis use the same CCFR survey data also gives at least some justifications for the comparison.

¹⁷ It is interesting to see that education and healthcare issues tend to have rather narrower opinion gap, although their effects do not reach to the conventional 0.05 significance level.

¹⁸ Here, we report the multivariate regression results for each year separately. However, it is possible to run the regression in which the year dummy is interacted with major independent variables. This interaction model would effectively reveal information on whether or not (or how much) the opinion gap changed from 2002 to 2004, which is not the main concern in this paper. Appendix A includes the results from interaction models.

¹⁹ An access to respondent-level survey responses, both at the elite and the mass surveys, would possibly help here. We unfortunately could not obtain them from CCFR.

²⁰ Given that immigration and trade issues have wider opinion gap between elites and the public (see Table 2), the recent Trump phenomenon does have some foundations.