OXFORD JOURNALS OXFORD UNIVERSITY PRESS American Association for Public Opinion Research Public Opinion and Public Policy, 1980-1993 Author(s): Alan D. Monroe Source: The Public Opinion Quarterly, Vol. 62, No. 1 (Spring, 1998), pp. 6-28 Published by: Oxford University Press on behalf of the American Association for Public Opinion Research Stable URL: http://www.jstor.org/stable/2749715 Accessed: 19/09/2009 13:37 Your use of the JSTOR archive indicates your acceptance of JSTOR's Terms and Conditions of Use, available at http://www.jstor.org/page/info/about/policies/terms.jsp. JSTOR's Terms and Conditions of Use provides, in part, that unless you have obtained prior permission, you may not download an entire issue of a journal or multiple copies of articles, and you may use content in the JSTOR archive only for your personal, non-commercial use. Please contact the publisher regarding any further use of this work. Publisher contact information may be obtained at http://www.jstor.org/action/showPublisher?publisherCode=aapor. Each copy of any part of a JSTOR transmission must contain the same copyright notice that appears on the screen or printed page of such transmission. JSTOR is a not-for-profit organization founded in 1995 to build trusted digital archives for scholarship. We work with the scholarly community to preserve their work and the materials they rely upon, and to build a common research platform that promotes the discovery and use of these resources. For more information about JSTOR, please contact support@jstor.org. 9 STOR American Association for Public Opinion Research and Oxford University Press are collaborating with JSTOR to digitize, preserve and extend access to The Public Opinion Quarterly. http://www.jstor.org PUBLIC OPINION AND PUBLIC POLICY, 1980-1993 ALAN D. MONROE Abstract To what extent have the policy decisions of the U.S. government been consistent with the preferences of the public? Using results of national surveys, public opinion was compared with actual policy outcomes on over 500 issues from 1980 through 1993. Overall, policy outcomes were consistent with the preferences of public majorities on 55 percent of the cases, representing a decline from 63 percent during the 1960-79 period. While this decline extended to almost all substantive policy topic categories analyzed, foreign policy decisions tended to be among the most consistent (67 percent), as in the previous two decades. Similar to the analysis of the earlier period, it appeared that a key reason for policy not being more consistent with public opinion was an inherent bias against change in the U.S. political process, a tendency that has increased over time. The reasons for this increase in bias against change, and the corresponding decline in consistency, are not entirely clear, but may be rooted in divided control of government combined with increased ideological conflict. It is interesting that when issues were categorized in terms of their salience to the public, it appeared that opinion/policy consistency was greater (and bias against change less) on those issues of highest public salience. What is the relationship between public opinion and public policy? More specifically, to what extent have the policy decisions of the U.S. government been consistent with the preferences of the American public? This research will attempt to answer that question for the years 1980-93, with comparisons to findings for the previous two decades. Review of the Literature While a good part of all of the research done on American national politics bears some relevance to the broad question of democratic responsiveness, alan d. monroe is professor in the Department of Political Science at Illinois State University. An earlier version of this article was presented to the Midwest Association for Public Opinion Research, Chicago, November 17-18, 1995. A list of the survey items and coding of policy outcomes used is available from the author, 4600 Political Science, Illinois State University, Normal, IL 61790-4600. Public Opinion Quarterly Volume 62:6-28 © 1998 by the American Association for Public Opinion Research All rights reserved. 0033-362X/98/6201-0001$02.50 Public Opinion and Public Policy 7 most of it provides only possible evidence as to whether various potential links between the people and their government are capable of performing that function. This review will omit attention to such possible links as voting behavior, political parties, interest groups, and the mass media. Rather, we will consider only those types of studies that make some attempt to measure the degree of correspondence between preferences of the mass public and actual policy decisions. Such attempts can largely be divided into ' 'dyadic'' and ' 'collective'' analyses (Weissberg 1978). Dyadic studies are those in which the behavior of an individual decision maker (almost always a legislator) is correlated with some kind of measure of public preference in his or her constituency. While there are inherent practical problems in obtaining direct survey-based measures of opinion for large numbers of constituencies—Warren Miller and Donald Stokes's (1963) classic article is one of the major exceptions—a variety of other strategies have been used to operationalize constituency variables, including demographic data (Erikson 1978), referenda results (Kuklinski 1979), and legislators' perceptions (Kingdon 1973). The results of these and many other types of studies of legislative behavior strongly suggest that the key variable in responsiveness to constituency is the salience of the issue. The more important an issue is to a legislator's constituency, the greater the relationship between constituency measures and the legislator's decisions (Hinckley 1983). While dyadic analyses are important to our understanding of legislative behavior, they are inevitably limited in their ability to measure the actual extent of correspondence between public opinion and public policy for two reasons. First, legislators' individual votes on bills do not equal policy. It is quite possible that even if most representatives do what a majority of their constituents would wish, the resulting outcomes would not be what most of the public as a whole would wish. This is particularly possible in the complex world of American national politics with divided partisan control of government, powerful congressional committees, a bicameral legislature, presidential vetoes, equal representation of states in the Senate, and decisions made by judges and bureaucrats who lack identifiable elective or geographic constituencies. Second, dyadic analysis as usually conducted is sensitive only to those issues where mass preferences vary between constituencies. If public opinion is roughly the same in most districts (as is often the case), then it can have little explanatory power for legislative behavior. Therefore, dyadic studies can tell us little about the extent of correspondence of mass opinion with actual policy. Collective analyses use actual policy decisions—for example, laws passed—in comparison to measures of opinion for an entire political system or subsystems. One way that this has been done is to compare different parts of a larger system (e.g., states within the United States). This has been done both by disaggregating national surveys to the state level 8 Alan D. Monroe (Erikson 1976; Joslyn and Ricci 1980) and by using the simulation technique originally devised by Ithiel de Sola Pool, Robert Ableson, and Samuel Popkin (1965) to make estimates of the distribution of public opinion for states or other units (Cook 1977; Edwards and Sharkansky 1978; Schumaker and Getter 1977; Sutton 1973; Weber and Shaffer 1972). When studies such as these compare their public opinion measures with policy decisions, they generally find a relationship, though one that varies with the type of issue and other variables. While the results of this kind of collective analysis is of greater relevance to the present inquiry, its utility for direct comparison is also constrained by the limited number of issues that can be studied this way, the same need for geographic variation noted previously, and the fact that our focus is on national policy decisions, not state and local questions. The relatively small number of studies that compare direct measures of national public opinion with actual policy outcomes on a number of issues employ two slightly different methodologies. Most are studies of "congruence," which require surveys using the same question wording to be conducted more than one time. The direction of opinion change is then compared to the direction of change in public policy. If the two changes are in the same direction, opinion and policy are said to be congruent. Perhaps the first use of this approach was that of Donald Devine (1970), who compared the preferences of the general and "attentive" publics and found that changes in public policy shifts were more responsive to the latter. Robert Weissberg (1976) used the congruence approach to study several areas of policy and found considerable variation between different topics. The most extensive congruence study is that of Robert Page and Benjamin Shapiro (1983) who looked at over three hundred issues drawn from surveys taken between 1935 and 1979. On those questions where there was some significant opinion change, policy change was in the same direction—and hence congruent—66 percent of the time. When cases are excluded for a variety of reasons, for example, opinion fluctuation, con-gruity reached 90 percent. Page and Shapiro suggest that congruence is higher on salient issues, among other patterns. When their data were analyzed in the same way as those in Alan Monroe's (1979) initial study, consistency was 68 percent, demonstrating a considerable similarity to Monroe's 63 percent figure, particularly considering the difference in time periods and specific issues included. There have been several later studies that have used the congruence approach to analyze more limited ranges of issues, particularly federal spending, for which survey data are available over the years and for which there are easily quantifiable policy decisions on an annual basis (see, e.g., Hinton-Andersson 1993; Wlezien 1995). A variation on the congruence approach is used by William Mishler and Reginald Sheehan (1993) who Public Opinion and Public Policy 9 correlate James Stimson's (1991) "public mood" measure with the ideology displayed in Supreme Court decisions and find evidence of responsiveness to public opinion. James Stimson, Michael MacKuen, and Robert Erikson's (1995) analysis of "dynamic representation" uses time-series analysis to investigate the relationship of the public mood data to a variety of summary indicators of policy decisions, finding evidence of a "strong and resilient link between public and policy." The use of summary measures of both opinion and policy makes possible sophisticated analysis, but it does raise the question of whether the issues on which the public was, for example, becoming more liberal in a given time period were the same issues on which policy decisions changed. The congruence approach has several theoretical advantages. By introducing a time dimension into both the opinion and policy variables, confidence in drawing conclusions about causality from data on consistency is increased. The problem of measurement validity, always a difficulty in this kind of research, is also somewhat alleviated. As John Mueller (1993) points out in his case study of public opinion and the Gulf War, different question wordings produce different levels of support for a policy, but comparisons of the same wording across time may provide a consistent measure of change. However, there are several drawbacks to the congruence method. First, since it requires results from identical survey questions asked at more than one time, some issues for which these data are not available must be excluded. Second, if there is no change in the distribution of opinion over time, it is not possible to determine congruence. As Page and Shapiro (1982) demonstrate, statistically significant change occurred over time in only about half of the issues they studied. There are also problems in assessing policy change; while some policies such as spending levels can increase or decrease each year, others, such as adoption of a constitutional amendment or passage of many regulative policies, are dichotomous in nature. Comparison of change in both opinions and policies raises a more fundamental problem from a democratic theory standpoint. Suppose that public support for increasing some government activity drops from 90 percent to 80 percent. A reduction in that activity would be the congruent response, even though that would be at odds with the preferences of a large popular majority. As a consequence of these limitations, the present research uses a ' 'consistency' ' approach, as described in detail in the next section of this article. This method takes surveys done at one point in time and compares the distribution of public opinion with the policy outcome. Thomas Marshall (1989) used this approach to evaluate 146 U.S. Supreme Court decisions from 1934 through 1986 in comparison to survey results and found that the judicial decisions were consistent with popular majorities in 62-66 10 Alan D. Monroe percent of the cases. Aside from Marshall's work and an early effort by Vance Russell Tiede (1974), which considered only issues on which Congress and the president disagreed, the main applications of this approach have been the earlier studies by Monroe previously cited. Methodology Data collection proceeded as follows. Major sources of national survey data from 1980 through 1991 were inspected and results for "usable" policy questions were recorded. (Policy outcomes were considered through 1993.) To be selected, a survey question had to deal with a question of potential federal policy, specific enough that it might be possible to determine whether or not federal policy complied with the terms of the question. This criterion excluded most survey items, including voting intentions, perceptions of parties and candidates, expectations about future conditions, and policy items too general in nature (i.e., "Should we get tougher with the Soviet Union?"). The primary data sources included the American Public Opinion Index and accompanying American Public Opinion Data, a microfiche collection published annually since 1981 that contains survey reports from a wide variety of sources; the Gallup Report (monthly); and Hastings and Hastings's annual Index to International Public Opinion. Most of the survey items came from well-known sources such as Gallup, the Harris Poll, surveys conducted for the several print and network alliances, and the annual surveys of the National Opinion Research Center, though results from a number of less prominent agencies were also included. After elimination of duplications, 566 cases remained, considerably more than those available (a total of 327) for the 1960-79 period. The specific topics of these issues are reported in Monroe (1995). In deciding which of several questions dealing with a particular issue to use, whether of the same or different wording, several considerations came into play. If there was a substantial difference in question wording, a judgment as to which was best had to be made. In general, specific items were preferred over general, while items that included some potentially biasing language (e.g., "in order to balance the budget, would you favor . . . ?") were avoided if there were alternatives. In the case of identical items, the results from the earliest survey were used in order to gain the longest time perspective and allow the greatest amount of time for policy decisions to occur. The choice of specific item, time, or survey rarely made much difference. Opinion distribution almost always remained fairly constant over time and even different versions of questions on the same policies. More than one use of questions on the same policy alternatives were used in only two circumstances. First, if the net distribution shifted from Public Opinion and Public Policy a plurality on one side of the issue to the other, both distributions were used and the issue then created two different cases. This happened only a handful of times. Second, if a substantial policy change intervened, then the issue could be used more than once. For example, a question about changing income tax rates asked in the early 1980s is really a different issue than when the same question was asked in 1991 because of the changes in taxation that occurred in 1986. However, this sort of circumstance allowed for duplicate use in only a few cases. Another complicating factor in the selection process was the question of what constitutes federal policy questions as opposed to state and local issues. The problem is that a number of topics deal with functions carried out primarily or entirely at the state level, including education; criminal law, trial, and punishment; surface transportation; and some aspects of welfare. But to exclude such topics entirely would be to ignore the realities of contemporary federalism as it is almost always possible that federal legislation, regulations, or court decisions could make policy changes implied in questions about these state functions. Indeed, national survey questions about state functions are typically asked because the issue has been raised at the federal level (e.g., introduced in a congressional bill). Therefore, some of these state issues were included, but only federal policy decisions were considered in the outcome. Once the survey items had been selected, it was necessary to determine what the state of policy was at the time of the survey and what federal actions, if any, took place afterward. The record of federal actions was considered from when the first survey was taken through the end of 1993. The major sources of outcome data were the annual Congressional Quarterly Almanac and Weekly Report, with other news and documentary sources consulted as needed. The task of determining policy outcomes is not as formidable or problematic as one might suppose because most of the survey items dealt with what proved to be specific policy proposals that were under consideration. Others that were not actively under consideration included a number where it was very apparent that the implied policy change had not occurred (e.g., proposals that would require a constitutional amendment). The basic dimension used to categorize both public opinion and policy outcomes was that of maintenance of the status quo versus policy change. Public opinion was measured in terms of the proportions of those holding opinions (excluding "no opinion," "don't know," and similar categories) who favored policy change. In the analysis presented here, this was dichotomized as to whether the majority of opinion holders favored the status quo or policy change. Similarly, policy outcomes were dichotomized into either maintenance of the status quo or policy change. It should be noted that while most outcomes were relatively unambiguous, there were a few cases where a decision had to made as to whether 12 Alan D. Monroe policy changes complied adequately with the terms of the survey item. In most cases, these were situations where a change in an amount of money, a tax rate, or some other numerical quality shifted in the implied direction, but not to the extent specified. In these cases, a conclusion of policy change was made if the numerical change was over half the amount specified. In most of those instances, it was considerably greater. The few cases where it was not possible to satisfactorily determine the outcome were excluded from the study. An important theoretical disclaimer should be included in this section. The method used here, based on analysis of public opinion measured at one point in time cannot, fundamentally, prove that public opinion caused or even influenced a policy outcome. The most basic problem is the lack of a guarantee of temporal priority as some surveys were conducted after a policy decision was made. However, in over 90 percent of the cases, the survey data preceded any policy decision. ("Retroactive" approval cases are discussed later in the article.) More difficult to assess is the causality criterion of nonspuriousness; neither this method nor the alternatives discussed earlier can completely rule out the possibility that some other variables may have caused any apparent relationship between opinions and policies. A number of controls will be used to rule out some possible factors. The possibility still remains that other forces, for example, elite control of the news media, might have conceivably manipulated both public preferences and government decisions. Page and Shapiro (1989) suggest that this may have occurred in some instances. On the other hand, if there is a lack of consistency (i.e., covariation) between mass public opinion and the actions of government, one could reasonably conclude that popular influence on government decisions was lacking. A final methodological comment is that because these issues do not represent any sort of a random sample, no significance tests are used in the analysis of the data. Findings Table 1 presents the overall findings between public opinion, dichotomized into majorities for the status quo versus majorities for change, and actual policy outcomes for the 1980s data as compared to that for the earlier two decades. The consistency percentages are simply the proportion of all cases in the table that fall into the ' 'consistent'' categories on the two variables, that is, "status quo/status quo" or "change/change." Consistency obviously dropped from the earlier period. While a majority of all issues in the later period, 55 percent, show consistency between majority opinion and actual policy, this certainly suggests less responsiveness than was the case in the previous two decades. Measures of asso- Public Opinion and Public Policy 13 Table I. Majority Preference and Policy Outcome Majority Preference (%) Policy Outcome Status Quo Change 1960-79: Status quo 82 49 Change 18 51 Total 100 100 N 125 202 Consistent 63 Gamma + .66 1980-93: Status quo 70 55 Change 30 45 Total 100 100 N 230 336 Consistent 55 Gamma + .30 ciation, such as gamma, between the two dichotomized variables are much lower after 1980. A principle goal in analysis of the earlier data was to try to explain why consistency was not higher. Here we have not only that question, but also that of why it has declined. Some insight may be gained by breaking the cases down by substantive area of policy, as is done in table 2. This set of policy categories was used for the 1980-93 issues only because it had been used for the earlier data. "Vietnam" was used a separate category for the 1960-79 data, but there was no analogous topic of similar duration or magnitude in the years since. Questions concerning the Gulf War and other U.S. military involvements were assigned to the "Foreign Policy" and "Defense" categories. As table 2 shows, consistency between majority opinion and policy outcomes declined in all but one policy area, the exception being defense issues. There appears to be no simple answer for these changes, but some observations are in order. First, there were substantial declines in consistency for the two largest policy categories, ' 'Economic and Labor Policy'' and "Foreign Policy" (minus 17 and 18 percentage points, respectively). It should be noted that the economic and labor policy area in the 1980s included a very large number of questions dealing with taxation as a consequence of several attempts, particularly in 1985-86, to revise the 14 Alan D. Monroe Table 2. Opinion/Policy Consistency by Policy Area 1960-79 1981-93 Consistent (%) N Consistent (%) N All cases 63 327 55 566 By policy area: Social welfare 63 51 51 45 Economic and labor 67 46 51 156 Defense 52 21 61 49 Foreign policy 84 38 67 150 Civil rights/liberties 59 39 56 61 Energy and environment 72 36 67 27 Political reform 41 34 17 23 Vietnam 71 35 . • . • • • Miscellaneous 74 27 40 55 structure of the U.S. tax system. Another reason for the increase in the relative number of cases in this category are a wide variety of proposals, mostly regarding spending cuts, which appeared on surveys in the context of possible ways to deal with budgetary and deficit problems. It is possible that this mix included a much greater number of inconsistent issues than was the case in past years. It should be emphasized that foreign policy issues since 1980 had a consistency level (67 percent) that was above average. That these issues showed a relatively great decline is a function of the surprisingly high levels of consistency in the pre-1980 years. Indeed, in the original study, covering 1960-76, opinion/policy consistency in foreign policy was 92 percent. Possible reasons for this decline are suggested in the concluding section of this article. Greater declines in consistency occurred in the policy areas of ' 'Political Reform" (minus 24 percent) and the "Miscellaneous" category (minus 34 percent). Reform is a somewhat peculiar category; it includes both proposals for fundamental structural changes in government, for example, abolition of the electoral college or a single 6-year term for presidents, as well as proposals dealing with subjects like political campaign finance. In the pre-1980 analysis, it was clearly the least consistent category, in part because many of its proposals would have required constitutional amendments to accomplish the changes. This difficulty in bringing about change was even greater in the past decade. Of the 23 issues covered, there were no outcomes resulting in the implied policy change. (There Public Opinion and Public Policy 15 was some consistency because in two of the cases, the public majority opposed change.) The "Miscellaneous" category is inherently difficult to describe. However, since 1980, in addition to somewhat obscure questions such as daylight savings time and the metric system, which typified it earlier, it now includes controversial topics such as gun control and drug abuse, and a number of crime and punishment issues that did not fit well into the "Civil Rights and Liberties" category. As noted previously, these categories, which proved adequate for the pre-1980 data, may not be the best for the past decade. In particular, large numbers of items were found relating to single, if complex, issues. Examples of this were the 1985-86 tax reforms, which elicited a total of 31 separate items included in the study, or the Gulf War, whose relatively brief history generated 28. Even though these represent major issues, including each separate question may exaggerate their importance and impact. In addition, decisions as to which substantive category an issue should be assigned are rather problematical as many contemporary questions transcend traditional boundaries. For instance, many items, such as arms sales, arms limitations agreements, troop deployments, and military aid, could legitimately be considered either foreign policy or defense. Foreign trade issues are both economic and foreign policy, while energy taxation could fall into both economic and energy categories. For these reasons, each issue was coded into at least one and possibly two "issue complexes." A total of 56 such complexes were established, and every issue was assigned to at least one. It should be noted that since 47 percent of the issues were coded into two complexes, this method, in a sense, overrepresents those questions. Table 3 reports the majority Table 3. Opinion-Policy Consistency for Leading Issue Complexes, 1980-93 Issue Complex Consistent (%) N Social Security and Medicare 61 23 Taxes, 1981-82 60 38 Taxes, 1985-86 58 31 Other taxes 18 22 Spending 51 55 Troop deployment 67 36 Gulf War 61 28 Foreign trade 57 30 Relations with Soviet Union 81 26 Energy 56 27 16 Alan D. Monroe opinion/policy outcome consistency figures for the 10 issue complexes that include 20 or more different issues. See appendix table A1 for a complete report. Most of the opinion/policy consistency levels for the groupings in table 3 are reasonably close to the overall level of 55 percent, suggesting that overrepresentation of survey questions on some topics did not greatly bias the total result. Most domestic policy complexes, that is, social security, taxation in the 1981-82 and 1985-86 periods, and energy, are generally in the 50-60 percent consistency range. Given the large number of taxation issues in the data set, they were divided into three periods. The most numerous came at the start of the Reagan years and included that administration's initial tax change proposals as well as those from other sources. The 1985-86 group dealt with the many questions over deductions, exemptions, brackets, and other details of the 1986 Tax Reform Act. ' 'Other'' taxes included various proposals in the late 1980s and early 1990s, as well as miscellaneous questions (e.g., various excise taxes, from the whole period). That these included a number of usually unpopular tax increase proposals, as well as several Bush probusiness proposals that would have decreased taxes, may account, in part, for the low consistency here. Spending items include all those that asked specifically about whether federal spending for specific programs and purposes should be changed in some way. As noted earlier, this included administration proposals as well as several sets of question lists that asked about possible cuts to balance the budget. For these items, policy outcomes were limited to changes in the fiscal year under consideration at the time of the survey in comparison to the current year with no adjustment for inflation or the size of the budget. What happened, of course, is that in the great bulk of the cases, spending increased or remained the same whether or not the public supported cuts, with a resulting consistency of 51 percent. The several categories more concerned with foreign policy showed somewhat greater levels of opinion/policy consistency. The Gulf War (which for all its positive effect on presidential popularity was not an event supported by a national consensus) was only a little above the overall average. Troop deployment, a category that included questions of sending forces to a variety of hot spots all over the globe as well as more traditional questions of maintaining troops and bases in established locations, was somewhat higher at 67 percent. The Soviet relations items were much higher than any other category at 81 percent, reminiscent of the overall foreign policy dimension before 1980. Of the issue complexes analyzed here, this represents the issues where presidential discretion may be the greatest; it also includes a number of instances where the public was asked to give retroactive approval to actions already taken, including a number of responses to the Soviet invasion of Afghanistan promulgated Public Opinion and Public Policy 17 by President Carter. As was argued in the analysis of the pre-1980 data (Monroe 1979), the ability to make decisions—any decisions—with greater ease may tend to increase consistency of policy with public opinion. This ability in military and foreign policy making, undoubtedly diminished since the 1970s, may still play a role, as discussed in the final section of this article. There are two structural features that apply to relatively few cases, but whose effects on consistency should be acknowledged. The first is whether or not a constitutional amendment would be required to accomplish the potential policy change. There were 13 such issues and most of them were coded into the ' 'Political Reform'' policy area, though a few, such as the Equal Rights Amendment, were classified elsewhere. Since the public favored 11 of these, and since none were adopted, the inclusion of such proposed changes requiring an unusually high level of political support for success, might have had a negative effect on overall consistency, but did not change the level by more than a percentage point. The other factor is theoretically more interesting: retroactive approval. There were 44 such cases in which the survey question was asked after the policy decision in question had already been made, most often in issues related to foreign affairs. Opinion/policy consistency was higher here than overall at 75 percent. Some of this higher figure is due to the fact that a policy change was made. As has been argued in earlier studies and will be discussed again, the bias of the political process against change appears to be an important reason government is not more responsive to public opinion. There is, however, also the possibility that official pronouncements of actions already taken may manipulate the public, particularly in situations of limited information, as in foreign policy (Page and Shapiro 1989). That public majorities favored such previously accomplished changes on 93 percent of such issues (compared to 64 percent of nonretroactive issues) may be a measure of this effect. Such is not always the case of course; when President Bush received strong approval for having (finally) banned the importation of assault weapons, one can hardly assume that the public would not have supported such an action without his already having taken it. In any event, exclusion of all retroactive approval cases from the analysis decreases overall opinion/policy consistency by only two percentage points to 53 percent. As mentioned several times, analysis of the pre-1980 data suggested that a major reason for lack of greater consistency of national policy with public preferences was the difficulty of government in making policy changes. In doing the case studies necessary to determine policy outcomes, it was striking how many times proposed changes had not only public support, but also apparent support by presidents and/or congressional leaders in one or both parties, but seemed to be long delayed and ultimately lost in the legislative labyrinth. In order to measure the extent 18 Alan D. Monroe Table 4. Index of "Bias against Change" by Policy Areas and Issue Complexes Bias Index 1980-93 1960-79 All cases 25 17 Policy area: Social welfare 66 33 Economic and labor 14 27 Defense 26 29 Foreign policy -16 12 Civil rights and liberties 57 -54 Energy and environment 37 25 Political reform 100 77 Vietnam -10 Miscellaneous 55 21 Issue complex: Social Security and Medicare 44 Taxes, 1981-82 -10 Taxes, 1985-86 -16 Other taxes 36 Spending 40 Troop deployment -7 Gulf War -27 Foreign trade -18 Relations with Soviet Union -40 Energy 25 of this bias for various groupings of cases, an index of bias against change was devised. This is a simple measure computed by taking the percentage of times a public majority who favored the status quo received that result and subtracting from it the percentage of times in which the public favored change actually experienced that policy outcome. In tables such as table 1, it means taking the percentage in the upper left cell and from it subtracting the column percentage in the lower right. This gives a relative measure of the probability a public majority has in getting what it wants when it wants to maintain the status quo than when it desires something new. This figure can be negative, indicating a bias in favor of change. This bias index is reported in table 4, overall and within policy areas, and within the issue complexes previously discussed. Comparison with the pre-1980 data are also presented. When figures for the earlier time period are compared to consistency figures, it appeared that the higher Public Opinion and Public Policy 19 the bias, the less the consistency, part of the basis for the argument that bias reduces responsiveness. For the 1980-91 data, there is a decidedly negative relationship between those two measures for both sets of categories, albeit imperfect and irregular. This is particularly seen by comparing the extremes. Political reform has a 100 percent bias figure (no change outcomes) and only 17 percent opinion/policy consistency, while foreign policy has a —16 percent bias (i.e., a bias in favor of change) and is tied for the highest consistency with 67 percent. Among the issue complexes, relations with the Soviet Union has a strong negative bias, —40 percent, and the highest consistency, 81 percent. Thus, the earlier conclusion that bias against change leads to lower consistency with public opinion seems to be supported. And as before, policy categories involving military and foreign policy decisions had less bias against change. Comparison of the categories in tables 2-4 suggests how the Reagan-Bush agenda may have interacted with the political situation to produce this pattern of policy outcomes. In domestic affairs, the general social welfare area had somewhat higher bias and slightly lower consistency in comparison to other areas and to the same area in the previous two decades. This is not surprising as these administrations hardly saw new innovations in welfare policy as a main focus, other than the unsuccessful "New Federalism" proposals. Nor were congressional Democrats in a strong position to initiate new programs. The specific complex of Social Security-related issues had many questions raised in response to concerns about the financial stability of the system in the early 1980s, but relatively few policy changes resulted. On the other hand, taxation questions in both the 1981-82 and 1985-86 periods had a bias in favor of change, reflecting the major overhaul of tax policy that occurred during this decade. Civil rights and liberties issues presented the greatest contrast to earlier findings. These topics had a strong bias in favor of change from 1960 to 1979, reflecting the greatest period of activism by both the judiciary and the elected branches in expanding individual rights in American history. It is hardly surprising that the bias was reversed after 1980, given the ideology of both the executive branch and the Supreme Court. However, it is important to note that the level of consistency with public opinion did not decline, but actually increased slightly. This may demonstrate that government may fail to respond to public majorities by making more policy changes than the public wants as well as by failing to adopt desired changes. The latter pattern, however, is clearly the more common. Political reform issues had a high degree of bias and therefore low opinion/ policy consistency in both eras. That these tendencies were even stronger after 1980 reflects the fact that, for all of the recent criticism of political campaigning, PAC contributions, congressional ethics, and the like, there has not yet been a wave of successful reform legislation comparable to the post-Watergate experience. 20 Alan D. Monroe Foreign policy-related topics showed a great similarity to the pre-1980 patterns. The general foreign policy area and the specific issue complexes of troop deployment, the Gulf War, foreign trade, and relations with the Soviet Union all showed biases toward change and a higher-than-average level of consistency with public opinion. The ability of the executive branch to make many more decisions on its own authority than in domestic policy, as well as to gain the acquiescence of Congress and the public— sometimes after the fact—makes this pattern somewhat inevitable. And the emphasis of the Reagan and Bush administrations on an aggressive foreign policy may have increased the relative number of occurrences of this pattern. In the case of foreign trade issues, which can involve both domestic and foreign policy components, it should be noted that many of the popular policy changes involved varieties of trade sanctions or restrictions against the Soviet Union or other communist countries, rather than major substantive trade issues that would have generated protracted legislative debate. As noted, the degree of salience of an issue to the public has appeared to be related to the opinion/policy relationship. Analysis of the 1960-79 data found a small tendency for greater consistency on more salient issues and Page and Shapiro (1982) report the same relationship. Unfortunately, direct measures of the salience of issues on which surveys ascertain preferences are quite rare. The present research measures salience by reference to the Gallup Poll question "What do you think is the most important problem facing this county today?'' as reported in the annual Gallup Poll volumes. This question was asked from one to six times annually during the years covered here. Cases since were matched with results of those Gallup questions closest in time to the date of the preference question survey. It must be noted that use of the Gallup results are extremely problematic. The biggest difficulty is the categorization of how responses to this open-ended item are reported. Typically, there are major categories such as "Economy (general)" or "Economic Problems" and "International Problems." Sometimes these are broken into their constituent parts—for example, "Unemployment, Recession," "Fear of War"—but often they are not. There is also some variation in how the multiple responses to this question are tabulated. But more fundamental, there are specific issues that some measurable proportion of the public feels are highly salient (e.g., abortion) that never appear as a separate category in the Gallup reports. Yet another practical problem is that salience is measured for the same time period as the original survey item, yet decisions come later, when the public agenda may have changed. Given the measurement problems with this variable, no attempt has yet been made to consider the effects of such shifts in public concern. Given the difficulties with this available measure, it was treated only Public Opinion and Public Policy 21 Table 5. Majority Preference and Policy Outcome, Controlling for Salience, 1980-1993 Top Problem Other Top Five Status Status Policy Outcome Quo Change Quo Change Status quo (%) 67 30 70 49 Change (%) 33 70 30 51 Total 100 100 100 100 N 21 59 86 86 Consistency (%) 69 60 Bias against change (%) -2 19 Gamma .64 .41 Other Ranked Not Ranked Status Status Quo Change Quo Change Status quo (%) 78 60 66 67 Change (%) 22 40 34 33 Total 100 100 100 100 N 40 53 80 130 Consistency (%) 56 46 Bias against change (%) 38 33 Gamma .39 -.02 in terms of ordinal categories: "top problem," referring to the issue being related to the most mentioned problem; "top five," that is, related to one of the other five most mentioned problems; "other ranked," related to some problem appearing in the Gallup responses below fifth in frequency; and "not ranked," not related to any response category reported for the relevant Gallup survey. It should be noted that the ' 'other ranked'' topics were almost always cited by less than 10 percent of respondents as, indeed, some of those in the ' 'top five'' sometimes were. Overall, 14 percent of the cases were related to the "top problem," 30 percent to other "top five" topics, 16 percent to "other ranked," and 37 percent to "not ranked." Table 5 presents the relationship between majority preferences and policy outcomes when salience is controlled. Salience has a decided effect and one consistent with expectations and past findings. The more salient an issue, the greater the degree of consistency of policy with majority 22 Alan D. Monroe preferences, as "top problem" cases had consistency of 69 percent; other "top five," 60 percent; "other ranked," 56 percent; and "not ranked," only 46 percent. Indeed, this is somewhat stronger and more regular than was the case for the 1960-79 analysis where the corresponding figures were 72 percent, 67 percent, 73 percent, and 59 percent, respectively. Correlations between opinions and policy also decline in the same pattern. One possible explanation is suggested by the "bias against change" index, also reported in table 5. While "top problem" issues had a slight negative bias (i.e., a bias in favor of change), the extent of antichange bias increased for the less salient categories. This is an important finding, for, to the extent that this problematical measure does reflect how important the public feels an issue is, there is a tendency for government to act more expeditiously in responding to those issues of greatest public importance. A more detailed analysis including policy area and theoretical policy "type" tended to confirm this conclusion that salience is an important factor in the opinion/policy relationship (Monroe 1995). As Bryan Jones (1994) concludes, greater public attention to issues can lead to greater responsiveness. Possible Explanations for the Decline in Consistency There are several possible explanations for a decline in consistency between public opinion and public policy since 1980. We must consider them only as potential explanations, both because this type of research does not lend itself well to precise determinations of causality and because even the considerable amount of data here does not cover all of the relevant factors. First, we must consider the possibility that some aspects of the research strategy produced an artifact of lowered consistency. There are several possibilities here. The greater availability of survey data which, even when cleansed of repetitions, resulted in almost three times as many issues per year, may mean that a relatively greater number of obscure suggestions that were never on the public or governmental agenda were included in the analysis. While this is a distinct possibility, it is difficult to see how it could be measured or evaluated. A related concern is that the very intensive concentration of survey questions on some topics led to items raising almost all possibilities in terms of policy options. In some cases, the public may have given approval to several possible steps that were, in fact, mutually exclusive. Second, there is the variable of time. As in the earlier studies, policy outcomes up to the end of the study were considered in relationship to Public Opinion and Public Policy 23 the earliest surveys, giving government more than a decade to act in response to some public preferences. Since the current analysis is restricted to the 1991 period in surveys with policy considered only to 1993, there was less opportunity for the federal government to respond to majority opinions on later issues. However, analysis of policy outcomes in both time periods shows that most relevant policy decisions occur very shortly after the survey. In the 1980-93 data, relevant policy changes occurred in only 42 percent of the cases. Of those instances where change occurred, it happened in the same year or earlier (including a small number of retroactive instances) in 64 percent of cases. When instances of policy change within one year are added, the proportion rises to 80 percent, and to 88 percent within 2 years. Furthermore, the extent of consistency does not appear to be particularly lower for issues surveyed in the past several years than in the early 1980s. Thus, extending the data collection period for policy outcomes is not likely to greatly increase the consistency ratio, particularly since it is always possible that later policy decisions will sometimes be contrary to public preferences. Leaving these methodological questions aside, there are several possible, and often interrelated, substantive explanations as to why opinion/policy consistency may have declined since 1980. There is evidence that an important limitation on the ability of the political process to respond to public opinion is the difficulty in making policy changes. This tendency appeared both before and after 1980 and the extent of this "bias against change" seems to have increased during the Reagan and Bush years. Why would there be greater difficulty in changing federal policy after 1980 than in the preceding two decades? Perhaps the most obvious possible explanation would be that 1981-92 was a period of complete divided partisan control of government. During the 1960-79 period, Democrats controlled the presidency and both houses of Congress for 12 of the 20 years. This explanation seems promising except that analysis of that earlier period showed no particular tendency for consistency to be lower under Republican presidents than Democrats (Monroe 1978). Hinton-Andersson's (1993) analysis of opinion/policy congruence on spending issues found that divided versus unified control made no difference, though it should be noted that she found little correlation of public opinion variables with policy in any event. It should also be noted that with a Republican majority in the Senate during Reagan's first 6 years, control was actually less divided than during under other Republican presidents. A somewhat different version of this partisan explanation would emphasize ideology. During the Reagan years, the nation had a clearly conservative president for the first time in two generations. This is relevant 24 Alan D. Monroe because analysis of the pre-1980 data indicated that public majorities tended to favor the liberal alternative on most issues. In order to consider that possibility, issues were coded as to which alternative (change or the status quo) seemed to represent a more liberal position. It should be emphasized that the researcher was quite cautious in assigning this evaluation. Only 39 percent of the cases were assigned ideological positions, while the rest were left undetermined. As in the earlier data, the public proved to be on the liberal side of issues more frequently than on the conservative side. Of those issues that were coded as to ideology, the public majority favored the liberal position on 67 percent of the issues. This proved to have some impact on policy outcomes, though not necessarily what would have been expected. In cases where the public majority favored the more liberal alternative, policy outcomes were consistent 58 percent of the time. In cases where the public majority favored the more conservative alternative, policy was consistent at a somewhat lower rate of 46 percent. Thus, while one might expect that the combination of a liberal public and a conservative administration would naturally lower opinion/policy consistency, the pattern of policy outcomes when controlled for ideology does not support that result. There are some other possible trends that may help explain a greater difficulty in achieving policy change. Certainly Congress has become less likely to acquiesce to White House desires in foreign policy than before the Vietnam War. Hence, the ease of policy change and resulting consistency has declined from earlier heights. However, change and consistency remain higher in foreign policy than other areas, perhaps reflecting the idea that the ' 'two presidencies'' thesis still holds, but only for Republican presidents (Fleisher and Bond 1988). Another change has been in the behavior of the U.S. Supreme Court, much of whose "Warren Court" activist period was included in the 1960-79 data. Mishler and Sheehan (1993) found that the Court became more conservative after 1980 in contrast to the overall trend in the public mood. One could also note that changes in committee procedures in the House during the 1970s led to greater decentralization of power, which may have made the adoption of policy change more difficult (Dodd and Oppenheimer 1993). Finally, it could be suggested that the experience of the 1980s was simply a continuation of a trend already in progress. Indeed, in the earlier set of cases, consistency declined slightly and the size of the bias against change increased sharply after 1974. Perhaps it is a combination of all of the factors hypothesized above, as well as the increasing complexity of modern government and continued popular expectations in a context of limited resources. It will certainly be interesting to extend this inquiry through the Clinton administration and into the future to see what the effect of the Republican "contract" and greater party discipline in Congress will be on the opinion/ policy relationship. Appendix Table AI. Opinion/Policy Consistency for all Issue Complexes, 1980-93 Consistent (%) N Social welfare: New federalism proposals 29 1 Social Security/Medicare 61 23 "Welfare" 25 8 AIDS 73 11 Other health issues 61 18 Education 82 11 Other social welfare 43 7 Economic and labor: Taxes, 1981-82 60 38 Taxes, 1985-86 58 31 Other taxes 35 23 Spending 51 55 Employment 46 11 Labor 67 6 Bank/savings and loan regulation 50 10 Other business regulation 46 13 Other economic and labor 37 19 Defense: Military hardware 47 15 Draft, national service 50 8 Space program 80 5 Arms limitation agreements 64 11 Military aid and sales 36 11 Other defense 54 13 Foreign policy: Troop deployment 67 36 Economic aid 63 16 Afghanistan 90 10 Iranian hostage crisis 74 19 Central America 50 16 Panama/Noriega 100 4 Other Latin America 70 10 Gulf War 63 30 Other Middle East 56 16 Foreign trade 57 30 Relations with USSR 81 26 Relations with Asia 72 18 Other foreign policy 50 8 26 Alan D. Monroe Table AI. (Continued) Consistent (%) N Civil rights and liberties: Crime and punishment 50 14 Drugs 80 10 Racial issues 50 10 Gender equity issues 44 9 Religious issues 38 8 Sexual orientation issues 50 4 First amendment issues 55 11 Abortion 33 6 Other civil rights and liberties 55 11 Energy and environment: Environmental protection 83 12 Energy 56 27 Reform: Structure and powers 15 13 Campaign reform 27 11 Other reform 0 3 Miscellaneous: Gun control 25 8 Transportation 41 17 Tobacco and alcohol 55 11 Agriculture 72 9 All other 56 9 References Cook, Beverly B. 1977. "Public Opinion and Federal Judicial Policy." American Journal of Political Science 21:567-600. Devine, Donald J. 1970. The Attentive Public: Polyarchical Democracy. Chicago: Rand McNally. Dodd, Lawrence C, and Bruce I. Oppenheimer. 1993. "Maintaining Order in the House: The Struggle for Institutional Equilibrium." In Congress Reconsidered, 5th ed., ed. Lawrence C. Dodd and Bruce I. Oppenheimer, pp. 41-66. Washington, DC: Congressional Quarterly. Edwards, George, and Ira Sharkansky. 1978. The Policy Predicament. San Francisco: W. H. Freeman. Erikson, Robert S. 1976. "The Relationship between Public Opinion and State Policy: A New Look at Some Forgotten Data." American Journal of Political Science 20: 25-36. -. 1978. "Constituency Opinion and Congressional Behavior." American Journal of Political Science 22:511-35. Public Opinion and Public Policy 27 Fleisher, Richard, and Jon R. Bond. 1988. "Are There Two Presidencies? Yes, but Only for Republicans." Journal of Politics 50:1 Al-61. Hinckley, Barbara. 1983. Stability and Change in Congress. New York: Harper & Row. Hinton-Andersson, Angela. 1993. "Opinion and Policy Congruence during Periods of Divided and Unified Control in the National Government." Paper presented at the annual meeting of the Southwestern Political Science Association, New Orleans, March 18-20. Jones, Bryan D. 1994. Reconceiving Decision-Making in Democratic Politics: Attention, Choice, and Public Policy. Chicago: University of Chicago Press. Joslyn, Richard A., and David Ricci. 1980. "The Relationship between Public Opinion and Public Policy across the American States." Paper presented at the annual meeting of the Midwest Political Science Association, Chicago, April 23-26. Kingdon, John. 1973. Congressmen's Voting Decisions. New York: Harper & Row. Kuklinski, J. 1979. "Representative-Constituency Linkages." Legislative Studies Quarterly 4:121-35. Marshall, Thomas R. 1989. Public Opinion and the Supreme Court. Boston: Unwin Hyman. Miller, Warren E., and Donald E. Stokes. 1963. "Constituency Influence in Congress." American Political Science Review 57:45-56. Mishler, William, and Reginald Sheehan. 1993. "The Supreme Court as a Countermajoritarian Institution? The Impact of Public Opinion on Supreme Court Decisions." American Political Science Review 87:87-101. Monroe, Alan D. 1978. "Public Opinion and Public Policy, 1960-1974." Paper presented at the annual meeting of the American Political Science Association, New York, August 31-September 3. -. 1979. "Consistency between Policy Preferences and National Policy Decisions." American Politics Quarterly 7:3-18. -. 1995. "The Role of Public Opinion in National Policy-Making: Effects of Salience and Policy Type." Paper presented at the annual meeting of the Midwest Political Science Association, Chicago, April 6-8. Mueller, John. 1993. "American Public Opinion and the Gulf War." In The Political Psychology of the Gulf War: Leaders, Publics, and the Process of Conflict, ed. Stanley A. Renshon, pp. 199-226. Pittsburgh: University of Pittsburgh Press. Page, Benjamin I., and Robert Y. Shapiro. 1982. "Changes in Americans' Policy Preferences, 1935-1979." Public Opinion Quarterly 46:24-42. -. 1983. "Effects of Public Opinion on Policy." American Journal of Political Science 77:175-90. -. 1989. "Educating and Manipulating the Public." In Manipulating Public Opinion, ed. Michael Margolis and Gary A. Mauser, pp. 294-320. Pacific Grove, CA: Brooks/Cole. Pool, Ithiel de Sola, Robert P. Abelson, and Samuel L. Popkin. 1965. Candidates, Issues, and Strategies: A Computer Simulation of the I960 and 1964 Presidential Elections. Cambridge, MA: MIT Press, 1965. Schumaker, Paul D., and Russell W. Getter. 1977. "Responsiveness Bias in Fifty-One American Communities." American Journal of Political Science 21:247-81. Stimson, James A. 1991. Public Opinion in America. Boulder, CO: Westview. Stimson, James A., Michael B. MacKuen, and Robert S. Erikson. 1995. "Dynamic Representation." American Journal of Political Science 89:543-65. Sutton, R. L. 1973. "The State and the People: Measuring and Accounting for State Representativeness." Polity 5:451-76. Tiede, Vance Russell. 1974. "Do 'The People' Agree More Often with the President or Congress? A Secondary Analysis of Survey Data." Paper presented at the annual meeting of the American Political Science Association, Chicago, August 29-September 2. 28 Alan D. Monroe Weber, Ronald E., and William R. Shaffer. 1972. "Public Opinion and American State Policy-Making." Midwest Journal of Political Science 16:683-99. Weissberg, Robert. 1976. Public Opinion and Popular Government. Englewood Cliffs, NJ: Prentice-Hall. -. 1978. "Collective versus Dyadic Representation in Congress." American Political Science Review 72:535-47. Wlezien, Christopher. 1995. "The Public as Thermostat: Dynamics of Preferences for Spending." American Journal of Political Science 39:981-1000.