Fluent Essays

article summary 1) Write a summary of the key points presented by the author; 2) Express your thoughts on the topic Summary should be no longer than 3 typewritten, double-spaced pages. Most social scientists who study public opinion and public policy in democratic countries agree that (1) public opinion influences public policy; (2) the more salient an issue to the public, the stronger the rela- tionship is likely to be; and (3) the relationship is threatened by the power of interest organizations,' political parties, and economic elites (see, e.g., Aldrich 1995; Dahl 1989; Mueller 1999; Stimson, MacKuen, and Erikson 1995; Page and Shapiro 1983; Smith 2000). There would be much less consensus, however, on the answers to five follow-up questions widely seen as impor- tant but seldom addressed directly: 1. How much impact does public opinion have on public policy? 2. How much does the impact of opinion on policy increase as the importance of an issue to the public increases? 3. To what extent do interest groups, social movement organizations, political parties, and elites influence policy even when opposed by public opinion? 4. Has government responsiveness to public opinion changed over time? 5. How generalizable are our findings about the impact of opinion on policy? This article distills considerable research directed at these questions. It is not, however, a literature review in the usual sense. Rather than summarizing publications in a con- ventional narrative, I use each publication as a source of data, tabulating the issues and countries studied, and the authors' predictions, variables, and findings. The analysis will provide the publications' collective answer to each question, and, at times, show how little evidence is avail- able. Highlighting how little we know on some issues will point to an agenda for future research. It turns out that public opinion influences policy most of the time, often strongly Responsiveness appears to increase with salience, and public opinion matters even in the face of activities by interest organizations, political par- ties, and political and economic elites. Claims that respon- siveness is changing over time or varies across issues rest on very little evidence. The next section describes issues that arise in attempts to answer the questions. This is followed by a description of the data, presentation of findings, and conclusion. ISSUES AND CONTROVERSIES The Impact of Public Opinion on Public Policy No one believes that public opinion always determines public policy; few believe it never does. Even dedicated pro- ponents of democratic theory acknowledge that democratic governments sometimes ignore the public (e.g., Page and Shapiro 1983: 189); those whose theories attribute little power to the public concede that governments sometimes follow public opinion (e.g., Block 1987: 66; Domhoff 1998: 301; Korpi 1989: 313). What distinguishes those who believe democracy gives citizens genuine control over their government from those who believe it does not, is thus dis- agreement over matters of degree: how much impact does public opinion have on public policy? This disagreement is an old one, and one might think it had been resolved, or at least narrowed substantially. 1 The term "interest organization" encompasses both interest groups and social movement organizations; for the rationale for treating them together, see Burstein 1998a. NOTE: I would like to thank William Domhoff, Kim Quaile Hill, Lawrence Jacobs, Florence Katz, and Alan Monroe for helpful advice and comments. This study was partially supported by NSF grant SES-0001509. Political Research Quarterly, Vol. 56, No. 1 (March 2003): pp. 29-40 29 This content downloaded from 99.172.120.36 on Wed, 08 Sep 2021 15:44:35 UTC All use subject to htt 30 POLITICAL RESEARCH QUARTERLY But this is not the case. Indeed, it may be argued that the range of predictions about impact based on democratic theory has widened in the past 20 years, not narrowed, and that researchers are no closer to consensus now than they were then. A good place to begin is Page and Shapiro's (1983) clas- sic article, "Effects of Opinion on Policy." They begin con- ventionally, delineating theoretical controversies about the impact of opinion on policy: some theories (particularly economists' on electoral competition) predict "a high degree of responsiveness" (175), while others (notably those attributing great power to interest groups) predict much less. Their empirical conclusions are presented in a conven- tional way as well: on the one hand, the evidence supports one side ("opinion changes are important causes of policy change" [189]), but, on the other hand, problems in the research require make them hesitate to accept their own conclusion-it would be "unwise to draw normative con- clusions about the extent of democratic responsiveness in policymaking" (ibid). What has happened in the 20 years since the publica- tion of "Effects of Opinion on Policy"? Theoretically, those expecting responsiveness to be low have generally held fast to their ideas, but the paths of those initially identified with the high responsiveness view have diverged. Some (e.g., Stimson, Mackuen, and Erikson 1995) still argue that democracy works much as it is supposed to, with public officials consistently responding to shifts in public opinion. Others have come to claim, however, that the complexity of modern politics makes responsiveness problematic. Demo- cratic institutions may link opinion and policy on issues that are especially important, relatively simple, and addressed by legislatures straightforwardly, but such issues are few. Jones (1994) argues that inherent limitations in both the cognitive capacities of individuals and the organi- zational capabilities of Congress mean that responsiveness is likely on only the few issues that the public cares about a great deal at any given time. Zaller (1992) and others (see Glynn et al. 1999: ch. 8) contend that on many issues the public cannot be said to have meaningful political opinions, so policy must be the product of other forces. And Arnold (1990: 271-72) suggests that many issues are so complex, and the legislative process so arcane, that most citizens are unable to ascertain whether their interests are being served. Thus, predictions about the impact of opinion on policy range from its having a very substantial influence (Stimson MacKuen, and Erikson 1995) to its keeping policy, rather vaguely, "in bounds" in its distance from public opinion (Jones 1994: 238). Increasing theoretical sophistication about opinion and policy has not narrowed the predictions; instead, they have become more diffuse. One might hope that 20 years of research would enhance the credibility of some theories and reduce that of others. But this does not seem to have happened, partly for a reason rarely discussed: researchers regularly describe their conclusions in terms too vague to be very useful. For example, Wlezien (1996: 81) writes that research "generally corroborates a linkage between public preferences and policy;" Page (1994: 25) that evidence shows "substantial empirical relationships" between opinion and policy; S. Hays, Esler, and C. Hays (1996: 58) that state environmen- tal regulation is "quite responsive" to public opinion, and Erikson, Wright, and McIver (1993: 80) that the relation- ship between opinion and policy in American states is "awe- some." Are they agreeing with each other about the impact of opinion on policy? Or disagreeing? Faced with this conundrum, a recent review (Glynn et al. 1999: 301) decides to "let the cases and data speak for themselves, so that the reader may judge." This does not seem very satisfactory. Thus, our first task is to develop a way to report findings consistently, so that we can address the first question: what does the evidence show about how much impact public opinion has on policy? What has happened in the 20 years since the publica- tion of "Effects of Opinion on Policy"? Theoretically, those expecting responsiveness to be low have generally held fast to their ideas, but the paths of those initially identified with the high responsiveness view have diverged. Some (e.g., Stimson, Mackuen, and Erikson 1995) still argue that democracy works much as it is supposed to, with public officials consistently responding to shifts in public opinion. Others have come to claim, however, that the complexity of modern politics makes responsiveness problematic. Demo- cratic institutions may link opinion and policy on issues that are especially important, relatively simple, and addressed by legislatures straightforwardly, but such issues are few. Jones (1994) argues that inherent limitations in both the cognitive capacities of individuals and the organi- zational capabilities of Congress mean that responsiveness is likely on only the few issues that the public cares about a great deal at any given time. Zaller (1992) and others (see Glynn et al. 1999: ch. 8) contend that on many issues the public cannot be said to have meaningful political opinions, so policy must be the product of other forces. And Arnold (1990: 271-72) suggests that many issues are so complex, and the legislative process so arcane, that most citizens are unable to ascertain whether their interests are being served. Thus, predictions about the impact of opinion on policy range from its having a very substantial influence (Stimson MacKuen, and Erikson 1995) to its keeping policy, rather vaguely, "in bounds" in its distance from public opinion (Jones 1994: 238). Increasing theoretical sophistication about opinion and policy has not narrowed the predictions; instead, they have become more diffuse. One might hope that 20 years of research would enhance the credibility of some theories and reduce that of others. But this does not seem to have happened, partly for a reason rarely discussed: researchers regularly describe their conclusions in terms too vague to be very useful. For example, Wlezien (1996: 81) writes that research "generally corroborates a linkage between public preferences and policy;" Page (1994: 25) that evidence shows "substantial empirical relationships" between opinion and policy; S. Hays, Esler, and C. Hays (1996: 58) that state environmen- tal regulation is "quite responsive" to public opinion, and Erikson, Wright, and McIver (1993: 80) that the relation- ship between opinion and policy in American states is "awe- some." Are they agreeing with each other about the impact of opinion on policy? Or disagreeing? Faced with this conundrum, a recent review (Glynn et al. 1999: 301) decides to "let the cases and data speak for themselves, so that the reader may judge." This does not seem very satisfactory. Thus, our first task is to develop a way to report findings consistently, so that we can address the first question: what does the evidence show about how much impact public opinion has on policy? Issue Salience and Government Responsiveness Issue salience has long been seen as a key element of democratic responsiveness. Citizens who care about an issue are especially likely to take elected officials' actions on that issue into account on election day (Arnold 1990: ch. 6; Jones 1994; see also Lindaman and Haider-Markel 2002). This leads elected officials to be particularly responsive on highly salient issues. The impact of salience on responsiveness has implica tions not only for particular issues, but for overall govern- ment responsiveness as well. If only a few issues at a time can be salient to the public and the legislature, and if responsiveness is high primarily when salience is high, then responsiveness will be high on only those few issues (Jones 1994: ch. 10). Policy would be kept from drifting too far from public opinion on low-salience issues mainly by elected officials' realization that their salience might increase at some future date. These arguments about overall responsiveness presume that salience has a powerful impact on responsiveness. But does it? Our second question: How much does the impact of opinion on policy increase as an issue's salience to the public increases? Interest Organizations, Political Parties, and Elites vs. the Public The most common objection to the claim that public opinion influences public policy is that policy is really determined by interest organizations, political parties, and elites, particularly economic elites. The resources available to interest organizations and elites may enable them to get what they want, even in opposition to public opinion (Domhoff 1998; Wilson 1990; Wright 1996), and political parties may, when in office, enact policies favored by their most ardent supporters rather than the general public (Aldrich 1995). Even when opinion and policy are highly correlated, the public's power may be more apparent than real; citizens may have been persuaded that they are getting This content downloaded from 99.172.120.36 on Wed, 08 Sep 2021 15:44:35 UTC All use subject to htt THE IMPACT OF PUBLIC OPINION ON PUBLIC POLICY 31 what they want, while effective power lies elsewhere (Mar- golis and Mauser 1989; Page and Shapiro 1992: ch. 9). These points seem obvious to most people, but social scientists have developed important alternative points of view. Many think interest organizations cannot get what they want against the wishes of constituents, who can defeat elected officials who ignore them. As Lohmann (1993: 319) writes, "it is puzzling that rational political leaders with majoritarian incentives would ever respond to political action" by interest organizations. Even if interest organiza- tions may be influential, their political activities may be most effective when consistent with public opinion (Denzau and Munger 1986; Kollman 1998). Indeed, some political scientists argue that interest organizations don't impede responsiveness, they enhance it. Hansen (1991: 227-30), for example, suggests that interest organizations may be influential, in part, because they pro- vide information useful to legislators, including information about what the public wants, serving as useful intermedi- aries between the public and the government. They repre- sent some groups better than others (see also Baumgartner and Leech 1998: ch. 6), but overall may enhance the impact of public opinion on public policy Denzau and Munger (1986: 103) argue that it makes sense for interest groups to focus their efforts on legislators whose constituents are divided, ignorant, or indifferent, because it is too costly to influence legislators whose constituents are informed and clearly on one side or the other. The latter group of con- stituents winds up being effectively represented by their leg- islators, even if they are unorganized. Similar arguments have been made about political par- ties. They may want to serve the interests of their most ardent supporters rather than the public, but electoral com- petition often mandates responsiveness to the public. They may have some flexibility in how they do this, but inter- party competition may actually increase the impact of opin- ion on policy (see, e.g., Blais, Blake, and Dion 1993; Burstein 1998b: ch. 5; Kitschelt 1994: ch. 7). Thus, discovering a relationship between opinion and policy is only a first step toward ascertaining how much power the public has. We also need to know the answer to the third question: To what extent do interest organizations, political parties, and political and economic elites influence policy even when opposed by public opinion? Trends in Responsiveness The struggle for democratic responsiveness never ends. There is a long history of institutional reforms intended to increase responsiveness, including extending the suffrage, regulating campaign contributions, nominating candidates through primary elections, and instituting referenda and initiatives. To the extent that such institutional changes have the effects their proponents intend, government responsive- ness to the public should increase (see, e.g., Garrow 1978; Rueschemeyer, E. Stephens, and J. Stephens 1992; Haskell 2001; Lijphart and Grofman 1984). Responsiveness might increase for other reasons as well. Improvements in communications, transportation, and information processing may enhance citizens' connections to their elected officials (Clemens 1997; Hansen 1991; Walker 1991: ch. 1). Public opinion polls may increase politicians' knowledge of citizens' preferences (Geer 1991). And the rise of interest groups may have enhanced respon- siveness as well (Clemens 1997). Increasing responsiveness is hardly inevitable, however. Attempts to reduce the public's influence on policy have occurred often (Markoff 1996)-some blatant (such as deny- ing effective suffrage to blacks after Reconstruction) and others subtle. Jacobs and Shapiro (2000: xvi) recently claimed that in the U.S. "the influence of public opinion on government policy is less than it has been in the past" (emphasis in origi- nal; also see pp. 326-27), largely because politicians have dis- covered how to avoid accountability to voters. A "growing body of evidence," they write (4), "suggests that since the 1970s the policy decisions of presidents and members of Con- gress have become less responsive to the substantive policy preferences of the average American." Both television and new strategies developed by interest organizations have been described as reducing responsiveness (Iyengar 1991: 42-43; Haskell 2001), and it has been suggested that it is reduced responsiveness that has led to the drastic decline in Americans' trust in government over the last 30 years (Bok 1997). Thus, there is real disagreement about whether changes in politics and society have increased responsiveness or decreased it. Hence, our fourth question: are democratic gov- ernments getting more responsive to public opinion, or less? Generalizing across Issues and Polities Theories about the impact of opinion on policy are typ- ically stated in general terms, and hypotheses about partic- ular aspects of the opinion-policy relationship are supposed to be derived from general theoretical propositions. For example, the hypothesis that responsiveness will be lower on foreign policy issues than on domestic issues is based on the general propositions that responsiveness increases with salience and with how well informed people are, together with the fact that foreign policy issues are usually of low salience to a poorly informed public (Jones 1994; Kollman 1998; Page and Shapiro 1983). The way research is usually designed and implemented presents at least a couple of impediments to hypothesis test- ing and generalization. First, researchers have limited resources and typically devote them to studying one issue they are particularly interested in, making generalization very problematic. Potentially, researchers could accomplish collectively what they could not as individuals, studying enough issues and circumstances to make hypothesis testing and generalization possible. Even collectively, however- here is the second possible impediment-the entire set of issues studied may be so small that it is unrepresentative of the set of all issues and an inadequate basis for generaliza- tion (Wittman 1995: ch. 13; cf. Page and Shapiro 1983). This content downloaded from 99.172.120.36 on Wed, 08 Sep 2021 15:44:35 UTC All use subject to htt 32 POLITICAL RESEARCH QUARTERLY Thus, the fifth and final question: what does the evi- dence show about our ability to generalize across issues and polities? DATA Data Sources This article presents no new data, instead drawing on the work of others. But it is not a conventional literature review, because it is oriented to hypothesis testing, which most such reviews are not. The approach here is a hybrid; others' research is used as data, with their "output" serving as our "input." Creating the new data set based on others' work required decisions about which studies to include, how to code the variables, and which data to include (cf. Baumgartner and Leech 1998, and Burstein 1998c). Any review of past work is necessarily selective; for this article relevant studies were drawn from the bibliographies of two recent, fairly extensive literature reviews (Burstein 1998c; Glynn et al. 1999: ch. 9), the three most prestigious journals in sociology (American Sociological Review, American Journal of Sociology, Social Forces) and political science (Amer- ican Political Science Review, American Journal of Political Sci- ence, Journal of Politics) from 1990 through 2000, and the book in which Jacobs and Shapiro (2000: 4) contend that responsiveness has declined. To be included, a study had to gauge quantitatively (though not necessarily statistically) the relationship between opinion and policy at the aggregate level, utilizing at least one measure of opinion based on a large random (or stratified, random) sample and a clear measure of public policy. Not included were discursive nar- ratives and studies of decisions by individual legislators. There were 30 such studies, listed in the appendix. Because the focus was on major reviews, top journals, and relatively recent works, their quality should be high. The unit of analysis is the effect of a predictor on a dependent variable, a measure of the relationship between opinion and a policy. Thus, if a particular author analyzes the impact of two distinct measures of public opinion on a policy outcome, that would be two effects.2 Studies considering many issues presented a problem. Some (e.g., Erikson, Wright, and McIver 1993; Stimson, MacKuen, and Erikson 1995) combine many issues into a single index (of "policy liberalism," for example). Arguably, such studies should be weighted more heavily, but it is not obvious how much more. Other studies (e.g., Page and Shapiro 1983; Monroe 1998; Brooks 1985) considered hun- dreds of issues separately before reaching an overall conclu- sion about responsiveness. If each issue were counted sepa- rately, those studies would dominate the results of any review like this one. The decision here was to take each study into account along the lines emphasized by their authors, focusing on coefficients for those relying on indexes and overall estimates of responsiveness (e.g., the percentage of issues on which opinion and policy agree) for the multi-issue studies. On this basis, the 30 studies include estimates of 52 effects. These will be called coefficients, even though not all take that form. Gauging Impact Researchers most often describe the impact of inde- pendent variables in two ways: in terms of statistical signif- icance, and of substantive significance. The first is by far the more common in studies of policy change. Its virtues are apparent precision and objectivity. It is difficult to argue with, except on highly technical grounds, and provides an answer to what is often the key question in a piece of research: did a variable have an impact? Statistical significance is not, however, a very satisfac- tory measure of impact (Gill 1999; Lieberson 1992; McCloskey 1998: ch. 9). It tells us whether there is a rela- tionship (with some uncertainty), but not how strong it is or how important in policy terms. It is thus of little help in answering the first question: how much impact does public opinion have on policy? Unfortunately, the studies use many measures of impact, and there is no precise way to compare them. That does not mean that nothing meaningful can be said about substantive significance, however. Each relationship between opinion and policy was coded as: 1 not significantly different from zero; 2 statistically significant, substantive significance not discussed; 3 statistically significant, substantive significance discussed and described as of little policy importance; 4 sta- tistically significant, substantive significance discussed and of considerable policy importance; and 5 ambiguous, some- times statistically significant and sometimes not, in ways unpredicted by the author CONCLUSIONS AND IMPLICATIONS FOR FUTURE RESEARCH This review has shown that: (1) Public opinion affects policy three-quarters of the times its impact is gauged; its effect is of substantial policy importance at least a third of the time, and probably a fair amount more. (2) Salience does affect the impact of public opinion on policy. (3) The impact of opinion on policy remains substantial when the activities of interest organizations, political parties, and elites are taken into account; but the paucity of data on interest organizations and elites mandates great caution when interpreting the results. (4) The hypothesis that gov- ernment responsiveness to the public has changed over time cannot be definitively rejected, because so little evidence is available; but that evidence does not support the hypothe- sis. (5) Our ability to generalize about the impact of opinion on policy is severely compromised by the narrow focus of available work, both geographically and in terms of issues. Overall, the findings about responsiveness seem quite robust, not strongly affected by the activities of political organizations or elites, type of issue, or time. Yet it is also surprising how little has been published in major journals, or referred to in major reviews, about critical topics con- cerning public opinion and public policy The publications reviewed suggest two agendas for future research, one sub- stantive and one methodological. A Substantive Agenda More work is needed on every topic addressed here, but the findings highlight some avenues of research likely to prove especially fruitful. 12 It would be desirable to ascertain whether democratization increases responsiveness. Only two studies looked at this, but both found it did (Erikson, Wright, and McIver 1993: ch. 9; Fording 1997) . This content downloaded from 99.172.120.36 on Wed, 08 Sep 2021 15:44:35 UTC All use subject to htt THE IMPACT OF PUBLIC OPINION ON PUBLIC POLICY 37 It has long been hypothesized that responsiveness varies with salience, and recent theoretical work has emphasized how important salience is to political conflict and overall responsiveness-if the connection between salience and responsiveness is in fact strong. Thus, the magnitude of the impact of salience on responsiveness matters greatly. Simple tests of the hypothesis that salience matters go back decades (e.g., Page and Shapiro 1983), and a great deal of data on salience is available. It therefore seems astonish- ing that only one study (Jones 1994) assesses statistically whether salience affects responsiveness, and only one more comes close to doing so (Burstein 1998b). More research on the relationship between salience and responsiveness is both feasible and urgently needed. Another issue of great theoretical importance is how the relationship between opinion and policy is affected by the activities of interest organizations, political parties, and elites. Again it seems surprising how little relevant research has been done. Studies of the impact of opinion neglect organizations and elites, while studies of the impact of inter- est organizations and parties neglect public opinion (Burstein and Linton 2002). Why this is the case is difficult to surmise. Contributors to each body of work ought to be able to get together with contributors to the other. Progress, though, would not simply be a matter of each set of researchers incorporating the other's variables into their studies. Some political scien- tists (Hansen 1991; Lohmann 1993, 1994; Wright 1996) who study interest organizations, for example, have argued that organizations are most likely to influence elected offi- cials when they provide them with information and resources relevant to their re-election prospects. Yet few studies of organizational influence consider the impact of information, and those that consider resources seldom assess their relevance to re-election (Burstein and Linton 2002). Similarly, with regard to public opinion, if salience is theo- retically important but seldom investigated, progress will be slight if those studying political organizations simply borrow conventional measures from specialists in public opinion. A third concern is generalizability Most studies of opin- ion and policy focus on issues that the researchers find espe- cially important and of interest to them personally. Almost never considered is how the choice of issues affects our abil- ity to generalize about the impact of opinion on policy Even important issues are neglected; perhaps even more critically, issues that don't make the headlines are virtually ignored (except in the studies that address hundreds of issues) even though, in the aggregate, the relevant policies affect the public tremendously. The sample of issues studied is very much biased toward those of relatively high salience; if salience influences responsiveness, current estimates of the strength of the relationship between opinion and policy may be too high. But we won't know if this is the case until we study a much wider range of issues-perhaps even some- thing like a random sample of issues. Another concern about generalizability stems from the exceptionally strong bias in extant work toward studying the United States. Not only does this limit our ability to say much about other long-established democracies, it also may cause us to miss opportunities to study the consequences of democ- ratization. In recent years many countries have democratized their political institutions, including …