Log in Page Discussion History Go to the site toolbox

501 Readings Week 13

From BluWiki

Benjamin Page and Robert Shapiro

“Effects of Public Opinion on Policy”

Page and Shapiro find “considerable congruence between changes in preferences and in policies, especially for large, stable opinion changes on salient issues” (175). They present evidence that public opinion is often a “proximate” cause of policy, affecting it more than policy influences public opinion.


They begin their article with a comparison of theories of democratic responsiveness. Economist predict high degree of responsiveness due to the nature of preferences, while interest group theorists suggest that the “will of the general public may be obstructed because the system of pressure group politics is biased in favor of well-organized business and professional groups” (175). Those who see government as a “problem solver” do not expect to see any close relationship between public preferences and policy decisions while certain other scholars do “anticipate agreement between opinion and policy, [but] they see the causal relationship as partly or wholly reversed: politicians and policies themselves affect public opinion” (175). Finally, the authors claim that the relationship could simply be spurious due to some outside factor and that these relationships are not mutually exclusive.


Page and Shapiro also reject the principal data used for most of the previous studies examining policy and opinion, claiming that the data is based on “small and unrepresentative district samples and are susceptible to varying interpretations” (176). Furthermore, the causal direction of these processes remains ambiguous.

The approach in this study employs a macrolevel, aggregate design to examine the “relationships between changes in preferences and changes in policy in the United States” (176). They specify congruence as defining the relationship that occurs when policy moves in the same direction as opinion. Page and Shapiro examine samples from 1935 and 1979 to find 357 instances of significant change in policy preferences. The preferences entailed foreign and domestic, spending, taxation, regulation, military action, trade, and diplomacy policies between federal, state, and local governments. Page and Shapiro measure the policy outputs occurring two years before the initial survey and ending four years after the final survey.

“Congruent changes in policy were clearly much more frequent than noncongruent changes” (177), with 33% of the cases demonstrating no policy change. These cases of no change would, at first, suggest that, while the public’s preferences changed significantly, policy did not. However, the authors find congruence “of a different kind” for many of these cases. First, these cases may represent policies that have reached floors or ceilings in which further policy change was impossible. Also, some of these cases did show congruence, but only after the period of time described in the study. Finally, small opinion movements would “not necessarily be expected to yield policy changes unless opinion happened to surpass some threshold” (178). Placing these cases aside, the authors find “substantial congruence” between opinion and policy (179).

Looking more closely at the cases of noncongruence, almost one fourth of these cases vanish when lags longer than the specified year are allowed for. Noncongruence is much more frequent when opinion change is small. So, when there is opinion change of 20 percentage points or more, policy change is congruent 90 percent of the time (180). Also, more sustained opinions result in policy congruence. However, the majority of American public policy tends to be congruent with changes in policy preferences.

Page and Shapiro then list some examples of noncongruence, such as the public disapproval of US aid to Eastern Europe, while that aid actually increased and the increase of public support for sending economic aid to allies while that economic assistance actually decreased. So, all in all, responsiveness to public opinion is not perfect.

Page and Shapiro also find that “the extent of congruence appears to vary according to the political institutions and the types of issues involved” (181). When public opinion change is large and stable, policy tends to move in the same direction as opinion, and tends to do so more often. Furthermore, policy issues with more salience lead to higher policy congruence. These findings suggest validity for Schattschneider’s argument, which says that “on issues about which the public ahs m roe well-defined opinions and shows more concern, where the scope of conflict is broad, policy tends to move in harmony with public opinion” (182). So, when the public has firm opinions with substantial changes that endure throughout time, political systems will be more responsive to the public’s preferences.

Page and Shapiro then examine how the type of policy issue affects this relationship. They find little difference between foreign and domestic issues but they do find more congruence for large-change social issues (such as abortion, civil liberties, and civil rights) than for economic of welfare policies (including national defense, political reform, and foreign aid). The authors then proceed to an examination of political institution, finding little difference among the executive branch, federal courts, and Congress. However, state policies turned out to be the most congruent , possibly because state issues are typically more salient (ie: capital punishment and abortion). So, policy is more responsive to the public on issues of high salience and visibility (183).

Turning to ideology, Page and Shapiro find that there is a “strong tendency for policy to move congruently with public opinion more often when opinion changed in a liberal direction” (183). However, this could simply reflect the timing of this study, as many of the most salient issues during the time period involved sweeping liberal trends (civil rights, civil liberties, and abortion). The fact that liberal changes tend to be large, salient, and stable could be responsible for the apparent effect of liberal opinion change upon the degree of congruence (185).

Page and Shapiro then turn to examine the causal direction between public policy and public opinion. In half or more of their cases of congruence, they rule out the possibility that policy affected opinion (186). One alternative conclusion they address is the issue of spuriousness. World events, elite leadership, interest groups, technology, or some other exogenous factor could be affecting the relationship, but they believe that opinion changes did affect policy (186).

To conclude, Page and Shapiro claim that when “American’s policy preferences shift, it is likely that congruent changes in policy will follow” (189), but they list several reasons to be cautious of assuming pervasive democratic responsiveness : 1) Their findgins concern policies “on which public opinion has been judged by survey researchers to exist and to be worth measuring at two or more points in time”. So, their cases are relatively broad and of high general salience 2) There are some cases among those in which policy does not change at all after opinion changes = nonresponsive 3) Few clear cases exist of noncongruence in which policy mioves in the opposite direction from opinion 4) “Their findings are consistent with policy affecting opinion in a substantial number of cases of congruence” (189) 5) There could be major biases in the system, as in, policy might always move only half as far as opinion, so it may move, but not as much as the public desires The authors seem confident only in suggesting that public opinion is a factor that genuinely affects policies in the U.S.




Larry M. Bartels

“Constituency Opinion and Congressional Policy Making: The Reagan Defense Buildup

Bartels attempts to examine the link between representatives’ roll call votes and constituency opinions on defense spending during the 1980 election campaign. He argues that public opinion for increased defense spending added $17 billion to the FY 1982 Pentagon appropriation. Excluding factors such as partisan turnover, intense district-level competition, and strong presidential coattails, Bartels finds that the majority of the appropriations ($16 billion) came from “across-the-board responsiveness by even the most safely incumbent representatives” (457)


Bartels’ aim is to provide a “direct estimate of the relationship between constituency opinion and specific legislative outcomes” (457). The first defense appropriations bill of the Reagan administration began a serious of annual appropriations resulting in a 40% real increase in defense spending during Reagan’s first five years in office, in line with strong public demand for defense increases.


Modern emphasis for the transformation of public opinion into public policy resides in the “importance of electoral competition for inducing political responsiveness, either through actual partisan turnover or through anticipation by incumbents of constituents’ policy demands” (458). However, Bartels claims that the combination of prospective electoral competition and widespread partisan turnover are too weak to ensure democratic accountability. He argues that the public demand for increases in defense spending in 1981 spread through all parts of the campaign cycle, including those campaigns conducted by securely incumbent representatives. To Bartels, it appears that “important policy changes can and do occur even in the absence of significant congressional turnover” (458).


Bartels begins by discussing the relevance of increased public demand for defense spending by outlining the Iranian hostage crisis and the Soviet invasion of Afghanistan. These two main issues (along with the Republican claim that President Carter’s defense appropriations were too small) led to a strong public demand for defense buildup. One analysis contends that the public demand added as much as 7.6 percentage points to the Reagan margin of victory in the 1980 election. Once in office, Reagan satisfied the public demand by proposing significant increases in the U.S. defense buildup. With this, Bartels examines the responsiveness of individual representatives, and the House as a whole, to these political forces.


He begins by estimating “representatives’ positions on an underlying policy dimension by relating characteristics of the representatives and their districts to observed behavior on a sequence of related roll call votes” (462). Three stages of the FY 1982 appropriations process are examined: an amendment to reduce the amount of money appropriated for weapons procurement and R&D, the bill itself, and the conference report reconciling appropriations figures (462). Bartels claims that an observance of individual representatives’ positions on these votes, it is possible to estimate their underlying preferences regarding overall levels of defense spending. He then examines three characteristics that affect his experiment: constituency opinion, economic interests, and partisan political factors.


Constituency opinion:” how the impact of constituency opinion varied with the electoral competitiveness of specific congressional districts. The idea is that representatives elected by relatively narrow margins might be especially sensitive to the policy demands of their constituents” (464).


Economic interests: “district-level variation in the economic costs and benefits of various Pentagon activities. Other things being equal, we expect representatives…. To support programs that bring their constituents contracts and salaries and to oppose programs that cost their constituents tax dollars, even when those costs and benefits are not entirely reflected in aggregate constituency opinion.” (464).


Partisanship: Bartels attempts to “characterize the real political impact of partisan turnover by estimating the effect of a representative’s party affiliation on his preferred level of defense appropriations after statistically controlling for relevant characteristics of his constituency.” (465). In his examination, Bartels found that Republicans wanted, on average, $3.9 billion more for Pentagon appropriations than Democratic counterparts.


Presidential influence: Bartels also measures Reagan’s potential influence over potentially unwilling representatives.


Bartels estimates the “total impact of constituency demand by multiplying the observed mean national position on the NES defense spending scale (1.235) by the estimated effect of constituency opinion on representatives’ appropriations preferences ($12.87 billion) to produce an estimated aggregate impact of almost $16 billion” (466). This figure significantly understates the actual magnitude of congressional responsiveness in four ways. 1) Congress had already “responded to constituency demands for Pentagon spending by voting huge increases in defense appropriations for FY 1981.” (466) 2) When these roll call votes were being cast, much of the public’s enthusiasm for defense increases had waned by late 1981 3) Representatives in late 1981 were responding to other public demands calling for social programs, tax reduction, and fiscal responsibility, limiting congressional ability to respond to each of these demands. 4) The increase in FY 1982 understates the long-run commitment to increased defense spending in the Pentagon appropriations votes.


Bartels’ analysis “suggests that public opinion was a powerful force for policy change in the realm of defense spending in the first year of the Reagan administration. Moreover, the impact of constituency opinion appears to have been remarkably broad-based, influencing all sorts of representatives across a wide spectrum of specific defense spending issues” (467.) So, Congress can produce substantial policy changes even without significant turnover and this congressional responsiveness depends on a variety of complex motives. The pattern of results reported by Bartels cries for “further exploration of the actual political processes relating constituency opinion and legislative behavior” (468).


Erikson, Robert S. 1978. Constituency Opinion and Congressional Behavior: A Reexamination of the Miller-Stokes Representation Data. American Journal of Political Science 22(3): 511-535.

Erikson re-explores the relationship between constituency opinion and voting behavior by their Congressional representatives originally written on by Miller and Stokes (1958). Identifying a potential measurement error in Miller and Stokes original study – small n – the author re-measures constituency opinions through an index of constituency characteristics that correlate with opinion. The study follows Miller and Stokes original study in a focus on the issues of social welfare, civil rights and foreign policy.

The data will be compared against both the opinion of Representatives and their perception of constituency opinions. This assumes that these two do not influence each other – which Erikson identifies is not necessarily true.

The study seeks to answer the question of whether Representative’s roll call votes reflect their own attitudes or their perceptions of the attitudes of the constituency. He founds that both influence roll call votes on the issues of civil rights and foreign policy, but that for social welfare votes the representative’s own opinions are the dominant influence.

How the constituency influences Representatives is the next issue explored by Erikson. Generally it is hypothesized that congruency between Representative votes and constituency opinion can be explained through:

1. That the election results in the selection of the candidate with the most similar preferences with the constituency, or

2. Candidates have congruent opinions b/c they are selected from their populations.

Erikson examines both. For the first he compared the opinions of the candidate elected with that of the one not elected to the constituency and found that the one elected did correlate more closely. For the second, candidate and constituency views were compared, however the results were mixed.

Overall, Erikson’s measure significantly increased the correlations between constituency opinion and Representative roll call voting compared to Miller and Stokes’ study. The source of the representation can be seen in both perception of constituency opinion as well as constituencies influencing their representatives both indirectly (b/c from same population) and directly (through vote selection).

Hill, Kim Qualie and Patricia A. Hurley. “Dyadic Representation Reappraised” American Journal of Political Science, Vol. 43, No. 1, January 1999. 109-137.

Long standing assumption exists within the literature that public opinion is an exogenous variable on representation. This tendency seems to exist due to a normative view of representation that desires constituency preference as the causal explanation for legislator action. Hill and Hurley argue that this assumption needs to be questioned. None of the current research looks that the circumstances under which certain patterns of elite/mass interaction should take place even though the literature is recognizing the endogenous characteristics of public opinion.

Argue that “issues that function as main lines of cleavage between competing political parities (easy issues) should be characterized by reciprocal linkages between mass and elite preferences, while highly complex issues on which party distinctions are unclear (hard issues) should be characterized by no linkages between mass and elite” (109). Other influences can exist between mass and elite policy preferences including attitude sharing, outright influence efforts, and various election dynamics.

Most fruitful method would be to do a longitudinal study that looks at policy change, elite behavior and public opinion over time but there is insufficient data to conduct this kind of study. Instead another method commonly used is a cross sectional analysis of particular policy issues.

Tries to look at specific relationships of dyadic representation that are missing in Miller and Stokes model of representation (See Model on Page 111). Uncertainty in their model “about the direction of the relationship between the representatives own attitude and his or her perception of the district’s attitude” (111)

Looks at one easy issue: social welfare, one hard issue: foreign policy, and one ambiguous issue: civil rights. (These issues were chosen from the issues of the 1950’s, which is where the data is from).

Data comes from the American Representation Study from 1958, which is characterized by small mass respondent samples. To make their results as valid as possible Hill and Hurley “take the original, individual level responses to each separate question posed to the mass respondents about their preferences in a particular policy area and calculate the arithmetic means of these responses within each district”, use “LISREL structural equation models for latent variables to analyze these data and test hypotheses” and use “LISREL confirmatory factor analysis procedures to investigate the validity and reliability of several observed indicators” (119).

First two hypotheses for social welfare and civil rights are confirmed with and acceptable fit (although not necessarily determinative). Third hypothesis of no relationship holds but lacks sufficient data to inspect validity.

In their conclusion they argue that these three types of relationships included in the typology are not exhaustive. Argue that their results are important as “they demonstrate that dynamic analysis is not the only way to establish the direction of the linkages between elite and mass opinion” (127) and has implications for democratic theory and the models of representation (128). Each hypothesis test confirms the dominance of the linkage from elite to mass but shows that linkage from mass to elite is larger than once thought. Argue that shared beliefs may account for the dominance of this second linkage.

Site Toolbox:

Personal tools
GNU Free Documentation License 1.2
This page was last modified on 26 April 2010, at 22:27.
Disclaimers - About BluWiki