If democracy is to be representative, then public policies should reflect what the public wants. But how do we determine whether or not this is the case? Christopher Wlezien takes a close look at policy representation, arguing that it can be very hard to assess given that support for and opposition to a specific policy may not reflect what the public actually wants. Public preferences for “more” policy (spending for example) also may not tell us much about what the public wants. He warns that in some policy areas, expressed preferences for more spending are completely unrelated to the policy status quo and so tell us nothing about whether the public even wants more or less.
The congruence between public preferences and public policy is of special importance in representative democracies. We want to know whether the public is getting the policies it wants and, if not, whose preferences are being represented. To directly evaluate this congruence, scholars need to measure what the public wants in a particular policy area and then correctly match it to policy in that area. This is difficult to do. Not surprisingly, while much scholarship examines the congruence of positions, little research examines actual policy congruence.
There is research on policy representation, of course. Much of this work concentrates on the correspondence between opinion and policy across different contexts. Other scholars assess the consistency between preferences for policy change and actual policy change, which is one manifestation of what Weissberg (1976) referred to as “majoritarian congruence.”
This research shows that there is a relationship between public opinion and policy but it tells us little about congruence. Only a tiny fraction of research actually attempts to equate the policy preferences of the public and policy decisions, and here Lax and Phillips’s (2012) groundbreaking work stands out. These scholars directly compare public support for specific policies in US states and the corresponding state policy decisions. Their analysis encompasses 39 specific policies in eight issue areas – abortion, education, electoral reform, gaming, gay and lesbian rights, healthcare, immigration, and law enforcement. The results show a great deal of responsiveness of policy to public opinion – that the two are correlated – but that congruence between policy adoption and majoritarian public support is evident in only about half of the cases.
What Lax and Phillips’ have done is to be commended not only for what it shows but for the example it sets to scholars of policy representation. Importantly, even this research and approach is limited in what it reveals about congruence. There are three particular ways.
First, support for and opposition to a particular policy can be deceiving about the public’s preferences for the policy. Consider Obamacare, which has received minority public support in the polls partly because a significant percentage of opponents actually favors a larger government role in health care, not less. In other words, preferences for a segment of the population seemingly have been fairly stable, i.e., they prefer Obamacare to the status quo but not to a more expansive health care policy. This can be important for the kind of inferences we draw. For instance, taking the public’s underlying preferences into account, Obamacare appears to be a case of congruence, not incongruence.
Second, and relatedly, expressed public support for particular policies does not reveal actual preferred levels. That is, while we can tell from responses whether people favor a particular policy, we cannot determine how much policy people want, which may be more or less than support for any particular policy. For instance, returning to the case of Obamacare, it may be that a person supports the plan but favors more, perhaps much more, of a government role in health care. Thus, while responses to survey questions asking about particular policies make it possible to assess whether the public is represented on those policies, they do not allow us to see whether people are getting what they really want. This is important because the match between policy and preferred levels of policy is what opinion-policy congruence ultimately is about.
Third, although we can tap preferred levels for specific policies in some areas, such as gay marriage, it is difficult to do in others. This is true for even for many seemingly binary policies, as others discuss at some length. It is truer still on more general policies. Consider what would happen were we to ask people how much spending they want on defense or welfare, let alone the many less salient areas in which government spends money and makes (and implements) other policy decisions. Of course, we can ask people what they want in different areas; it just is not clear what responses would reveal. This is mostly because we do not expect people to have specific preferred levels of policy in mind, and so responses are not likely to be very meaningful. Not surprisingly, survey organizations rarely ask respondents about their preferred levels of policy and usually ask about relative preferences, e.g., whether we are “spending too little” or whether we are “doing too much.”
Relative Preferences and Congruence
While it may be difficult to measure the public’s preferred levels of policy in many areas, there may be indirect ways to assess congruence. Indeed, measures of relative preferences for policy themselves may be informative. Consider that those questions in effect ask people whether policy matches their preferred levels, and so it may be that responses tell us something about opinion-policy congruence. If people say “about right,” after all, they seemingly are happy with the status quo. By contrast, if they say “too little” or “too much,” they are indicating that they prefer policy change. This is the assumption underlying work on “consistency” mentioned earlier and provides the basis for Bartels’ (2015) assessment of the social welfare “deficit.” But does the assumption hold?
Let us take a look at some data. Figure 1 shows net support for various spending preference items in the US, drawing on questions from the General Social Survey (GSS) between 1973 and 2012 about whether spending is “too little,” “about right” and “too much” in a set of very broad domains: crime, defense, education, the environment, healthcare, and welfare. Specifically it shows net support for spending, which represents the percentage saying “too little” minus the percentage saying “too much.” The spending categories are quite broad, and so do not provide very specific guidance to policymakers or scholars. That said, research has shown that responses do appear to be informative about spending in general areas, as they adjust to changes in spending over time and, in turn, are effectively represented in spending itself.
Figure 1 – Net support for Spending, 1973-2014 (in percentages)
Source: General Social Survey.
The data in Figure 1 suggests that there is substantial incongruence in spending domains in the US. With the exception of welfare and defense, the median (and mean) person appears to be underrepresented. One cannot determine from these data how much misrepresentation there is on spending in the US, of course, but we can tell that substantial majorities say the government is spending too little in many domestic domains or, in the case of welfare, too much. These data thus can be taken to imply that there is a substantial democratic deficit, which is potentially damning of substantive political representation in the US and elsewhere where the same is true.
These data also can be used to assess who is being represented. That is, we can see for which groups relative preferences indicate the greatest congruence and which groups the least. Figure 2 displays preferences for spending across income groups, the subject of much political science research on inequality, some of which suggests that the preferences of the rich dominate American politics. There is a spate of research demonstrating that the portrait of oligarchic control is not quite right. Figure 2 shows the mean net support for spending—the percentage who think we are spending too little minus the percentage saying too much—in the different domains by family income group.
Figure 2 – Net Support for Spending, by Income Tercile, 1973-2014 (Percent “too little” minus percent “too much”)
Source: General Social Survey.
Interestingly, the results in Figure 2 imply that the lower third of the income distribution is best represented in most domains. This can be seen from the fact that mean net support over the period is closest to zero for the low income group in four of the six domains — education, the environment, defense and welfare. On average, the mean absolute difference from zero is smallest for the low income group; the means for the middle and high income groups are about 15 percent larger (7 points of net support). These results appear to contrast with much of the recent research on inequality in opinion-policy representation; that is, they imply that the poor are more satisfied on average than the middle and rich!
The most important take-away from Figure 2 may be that there is not much difference in preferences across income levels for the spending domains if they are taken separately or together, particularly given the theoretical range of the variables from -100 to 100. The one exception is welfare, where there are sizable differences, and there the poor seem more satisfied with the policy status quo than the middle and especially the rich.
While intriguing, the analyses in Figures 1 and 2 and the conclusions we might draw from them are deceiving, for at least three reasons.
First, question wording matters a lot, and in two main ways.
- To begin with, the spending categories are important. The archetypal case is spending on welfare versus the poor. If we ask people about “welfare,” a majority of respondents think we are spending too much, as can be seen in Figure 1. If we ask about “the poor,” by contrast, a majority think we are spending too little. So, which is it? Are we spending too much on people in need or too little?
- Budget constraints also are important. The spending question registers people’s unconstrained preferences, where no trade-offs are stipulated between spending on different programs, between spending and taxes and/or deficits. Responses to the question thus are cheap – indeed, costless. Including trade-offs produces very different spending preferences, and greater support for the status quo. The numbers reported in the figures thus are deceiving about public preferences, and it may be that the public is quite satisfied with the spending they are getting in particular domains, at least given fiscal constraints.
Second, even assuming we have the right question wording, to assess congruence we need to match up preferences with spending. Consider welfare. Does spending on it include food stamps? How about housing assistance? What about social security? You or I may be comfortable making these decisions and budget analysts may be too, but we cannot be sure that the public has in mind what we and budget analysts do when we categorize the programs. This is important, because what matters for assessing congruence is not what we have in mind but what is in the minds of the public.
Third, using relative preferences to indicate congruence presumes that the public responds thermostatically to policy. Otherwise measured relative preferences do not tell us anything about the public’s satisfaction with the policy status quo. We often observe such thermostatic responsiveness, but we sometimes do not (Wlezien 1995). Even where we do, the effect size varies owing partly to characteristics of issues and partly to political institutions. In other words, measured relative preferences are not reliable indicators of congruence across domains. In some domains, responses are completely unanchored to the policy status quo and so tell us absolutely nothing about whether the public wants more or less.
On Responsiveness and Congruence
What is a scholar do to? Where possible, we can follow the example of Lax and Phillips (2012), bearing in mind the limitations discussed above as we do. In other areas, we can keep working on ways to measure preferred levels of policy, either directly or indirectly, though this might not be possible for many areas. We also can pay more attention to policy measurement, especially the implementation of decisions, which has received scant attention of scholars studying representation. In the meantime, we need to recognize and accept that there often is little basis for assessing congruence between public preferences and policy in most domains.
This article is based on the paper: ‘Public Opinion and Policy Representation: On Conceptualization, Measurement, and Interpretation’ in the Policy Studies Journal.
Note: This article gives the views of the author, and not the position of USAPP – American Politics and Policy, nor of the London School of Economics.
Shortened URL for this post: http://bit.ly/2om9SJC
Christopher Wlezien – University of Texas at Austin
Christopher Wlezien is Hogg Professor of Government at the University of Texas at Austin. He holds or has held visiting positions at numerous universities in the United States and other countries. His primary, ongoing research develops and tests a “thermostatic” model of public opinion and policy, and his other major project assesses the evolution of voter preferences over the course of the election “timeline.” He has published numerous articles and chapters as well as a number of books, including Degrees of Democracy and Who Gets Represented? and The Timeline of Presidential Elections. Wlezien was founding co-editor of the international Journal of Elections, Public Opinion and Parties. He currently is Associate Editor of Public Opinion Quarterly, Research and Politics, and Parliamentary Affairs and a member of the editorial boards of six other journals.