Conflicted Conservatives: The Politics of Ideological Contradiction in the United States

By Christopher Ellis

The American public calls itself conservative, but its opinions on issues are predominantly liberal. When asked about policies to provide public goods or benefit specific social groups, Americans generally want government to do more rather than less, spend more rather than less, and play a more rather than a less active role in regulating business. But when asked about their ideology, far more citizens choose the label “conservative” than “liberal.”

To scholars of public opinion, it is not new to suggest that Americans do not “get” ideological thinking—the macro-level disconnect between operational and ideological opinions has been observed since at least the 1960s. What is not generally recognized is that there is a large segment of the population that systematically identifies itself as ideologically conservative for reasons that have little to do with the positions it takes on important political conflicts.

We argue that the reasons for this are best understood not in terms of issue politics, but instead in the ways in which citizens approach the labels “liberal” and “conservative” themselves. Because of its nonpolitical connotations and the way in which it is used by political elites, the label “conservative” is both more popular and more multidimensional than the term “liberal.” Conservatives can thus approach conservative self-identification for reasons that have little to do with preferences on the main dimension of political conflict. But even though these identifications are not explicitly political, they can still be meaningful—and can still be used to guide political choices.

To understand the relationship between political issues and ideological self-identification, we need to do more than look at the electorate as a whole; we need to understand the preferences of individuals at particular points in time. We need to understand on what issues, in broad terms, citizens can hold “operationally” liberal or conservative preferences, and use this information as a basis for exploring the aggregate-level paradox.

To approach this problem, we conducted and interpreted a simple exploratory factor analysis of the thirty-five policy preference questions asked of all respondents in the 2004 American National Election Studies (NES). This analysis suggests the existence of two coherent dimensions along which the public arranges issues.

The first (and by far the strongest) dimension consists of preferences for most everything related to the size and scope of the federal government (see Table 1). “New Deal”-type policies pertaining to distribution and redistribution of wealth fall on this dimension, as do preferences for the death penalty and gun control, and the tradeoffs involved in taxing, providing services, and balancing the federal budget. NES respondents, it appears, largely followed the lead of party elites in collapsing most policy conflict onto a single, dominant dimension.

The second, smaller, dimension, shown in Table 2, contains issues that are not economic. But neither are they the social dimension broadly defined. Instead, they relate narrowly to matters of traditional, religious morality. In 2004, this largely meant abortion and gay rights.

We can classify citizens as either operational liberals or conservatives on these dimensions. Individuals are considered operational conservatives on a particular dimension if they hold more conservative preferences than liberal ones, and vice versa for operational liberals. As we see in Figure 1, preferences on the dominant “scope of government” dimension reflect what we have long known: Americans generally express preferences for more programs, more spending, and more redistribution of wealth. Nearly 70 percent of NES respondents held predominantly liberal views on these issues. Although liberalism was not as overwhelming on traditional moral issues, a plurality held liberal views here as well.

But what happens when we consider the relationship between ideological self-identification and policy preferences? We find self-identified liberals, by and large, uninteresting (see Figure 2). Sixty-three percent of all ideological liberals are consistent liberals—operationally liberal on both issue dimensions. In total, 87 percent held liberal preferences on the dominant “scope of government” dimension, and only 4 percent were not operationally liberal on either issue dimension. To call oneself a liberal almost universally means to be one.

The story for conservatives is far different (see Figure 3). The largest group of self-identified conservatives in the sample was made up of those who held conservative issue preferences only on the narrow dimension of traditional morality—we can classify them as “conflicted” conservatives. The second largest group (“moral” conservatives) were conservative on neither issue dimension. It was only the third largest group (“consistent” conservatives) that held consistently conservative preferences on both issue dimensions. Put another way, nearly three-quarters of self-identified conservatives are not conservative on at least one issue dimension, and considerably more than half hold liberal preferences on the dominant dimension of conflict over the size and scope of government. Simply put, many conservatives are not very conservative.