Vous êtes sur la page 1sur 21

Parties, Institutional Control, and Trust in Government n

Joseph Gershtenson, Eastern Kentucky University Jeffrey Ladewig, University of Connecticut Dennis L. Plane, Juniata College
Objectives. Partisanship should affect evaluations of Congress just as it affects evaluations of the president, and these institutional evaluations should affect political trust. We argue that the relationship between partisanship and trust is dependent on partisan control of Congress and that much of party identications inuence on trust occurs indirectly through approval of governmental institutions. Methods. Using data collected before and after the 2002 congressional elections by the Center for Survey Research and Analysis at the University of Connecticut, we examine changes in frequency distributions and mean values for trust and institutional approval. We use multivariate regression models and a path model to estimate the causes of political trust and self-perceived change in trust. Results. We nd evidence that party control of government and party identication are important in explaining trust and institutional approval. The Republican takeover of the Senate led Republicans to evaluate the Senate more favorably and to become more trusting of the government, while having the opposite effect on Democrats. Conclusions. The changes in approval and trust resulting from the 2002 elections suggest that at least some segment of the population is cognizant of changes in the political environment and updates its views of government when the political environment changes.

Levels of trust in the national government have wide-reaching implications for the vitality of American democracy. Trust in government not only affects which candidates citizens vote for (Hetherington, 1999), but also affects whether citizens cast a ballot at all (Hibbing and Theiss-Morse, 2002). In addition, trust inuences citizens policy preferences and support for government activities on particular issues (Hetherington, 2001; Hetherington and Globetti, 2002), their compliance with the law (Scholz and Lubell, 1998), and their assessments of politicians and political institutions

n Direct correspondence, including requests for data and coding information necessary to replicate the study, to Joseph Gershtenson, Eastern Kentucky University, Department of Government, 113 McCreary Hall, 521 Lancaster Ave., Richmond, KY h joe.gershtenson@ eku.edu i . The authors names appear in alphabetical order. The authors thank Ken Dautrich, Sam Best, and the Center for Survey Research and Analysis at the University of Connecticut.

SOCIAL SCIENCE QUARTERLY, Volume 87, Number 4, December 2006 r2006 Southwestern Social Science Association

Parties, Institutional Control, and Trust in Government

883

(Chanley, Rudolph, and Rahn, 2000; Hetherington, 1998). A lack of political trust undermines support for democratic government (Gamson, 1968; Hetherington, 1998), which in turn raises questions about a governments legitimacy (Easton, 1965). In sum, trust affects the ability of politicians to govern effectively. Because the level of political trust carries important consequences, scholars have focused considerable attention on its causes. Without denying the role of other determinants of trust, we concentrate on the role of partisanship in the trust calculus. Specifically, we argue that citizens who share party identication with the party controlling a political institution should be more satised with its performance and more trusting of government. Indeed, previous research has shown that citizens are more trusting when a member of their party is in the White House (Citrin, 1974; Schaffner and Clark, 2004). On the other hand, the relationship between political trust and partisan control of Congress remains relatively unexploreda gap this research seeks to bridge. We conrm ndings by others (e.g., Citrin, 1974; Citrin and Luks, 2001) that evaluations of Congress are signicant in the trust calculus. However, we move beyond congressional approval by examining approval of the two chambers independently and by outlining how partisanship may have both direct and indirect effects on congressional approval and trust. Specifically, we argue that party identication affects citizens evaluations of political institutions, which in turn affect trust.1 Finally, we examine individuals self-reported changes in trust, something previous research designs do not permit. The ndings are consistent with our hypotheses and carry important implications not only about political trust and partisanship, but also about the nature of citizens political thinking and the signicance of divided government.
The Causes of Trust

Extant research has not yielded a general agreement on the underlying causes of political trust. One reason for the lack of consensus about the causes of political trust is a lack of agreement on what exactly constitutes trust. Most definitions involve evaluations of government relative to normative expectations (Miller, 1974). Of course, normative expectations involve a range of possibilities. To begin, citizens may desire that politicians behave ethically. Thus, mistrust may be traced to citizen perceptions that politicians are dishonest or lack integrity (Lipset and Schneider, 1987; Black and Black, 1994). The close scrutiny politicians currently receive from the media may feed this type of mistrust (Patterson, 1993; Orren, 1997).
1 To be certain, previous research (e.g., Citrin, 1974; Citrin and Luks, 2001) demonstrating the effects of congressional approval is suggestive regarding indirect effects of partisanship. However, the full linkage mechanisms between party identication and trust are not outlined and tested directly as we do here.

884

Social Science Quarterly

In addition to considering the integrity of politicians, citizen perceptions of governmental inefciency and waste are also cited as reasons for governmental mistrust (Blendon et al., 1997). More generally, citizen evaluations of government may also be inuenced by perceptions of how it operates and the relative strength of various political actors. For example, the belief that special interests improperly dominate political debate spurs mistrust (Blendon et al., 1997). Hibbing and Theiss-Morse (1995, 2001, 2002) have extended this line of reasoning, focusing on the role of governmental processes as the root of citizen mistrust. They claim that most Americans do not value democratic processes such as deliberation and compromise. Political debate is thought to be little more than petty arguments that hinder governments ability to act; compromise is equated with selling out on principles. The visibility of the political process, combined with a belief that Americans policy preferences are thwarted by special interests, makes the public increasingly wary of government (Nelson, 2003). Trust may also be inuenced by policy outcomes. For example, the actual and perceived conditions of the economy are often used to measure government performance and have been found to affect trust (Citrin and Green, 1986; Hetherington, 1998; Miller, 1983; but see Lawrence, 1997). Government actions on other policy issues important to citizens also inuence trust (Craig, 1996). More generally, citizens are more trusting of government when they believe it is pursuing policies that reect their own preferences. For example, greater continuity between citizen policy preferences and government outcomes contributes to higher levels of trust (Citrin, 1974; Miller, 1974; Citrin and Green, 1986; Miller and Borrelli, 1991; Hetherington, 1998; Kimball and Patterson, 1997). Even if policy outcomes are not reective of citizens preferences, citizens may nonetheless be more trusting when the party with which they identify has similar policy preferences (King, 1997).
Partisanship, Congress, and Trust

Given low levels of information among the American public about specific political issues (Sniderman, Brody, and Tetlock, 1991; Lupia and McCubbins, 1998) and politicians stances on issues (e.g., Delli Carpini and Keeter, 1996), it is perhaps somewhat surprising to nd consistent relationships between policy preferences and trust. Although citizens inattention to politics (Converse, 1964; Luskin, 1987) may be rational (Downs, 1957), their lack of information means that cues and heuristics are important determinants of political behavior (Sniderman, Brody, and Tetlock, 1991; Popkin, 1994). Among political shortcuts used by citizens, party labels are paramount. Quite simply, party labels often guide Americans understanding of the political world and their interaction with that world. Party identication strongly inuences the likelihood that citizens will vote (Rosenstone and Hansen, 1993) and for whom they cast their ballots

Parties, Institutional Control, and Trust in Government

885

(Bartels, 2000; Hager and Talbert, 2000; Lawrence, 2001). In addition, partisanship affects citizens political attitudes, including trust (Citrin and Luks, 2001; King, 1997; Pew Research Center, 1998, 2003). The effects of partisanship on trust, however, are perhaps less straightforward than might be anticipated. Stronger identication with parties encourages political participation (Rosenstone and Hansen, 1993). In a similar vein, King (1997) nds a direct link between strength of party identication and faith in the government: regardless of the party with which they afliate, strong partisans tend to be more trusting than are individuals with weaker party attachments. That strength of partisanship promotes arguably desirable citizen attitudes is not entirely surprising. On the other hand, one might also anticipate differences between Democrats and Republicans within the ranks of strong (or weak or leaning) partisans. Identiers are generally more supportive of politicians who share their afliation and should be more trusting of a government controlled by their own party. Indeed, Citrin and Luks (2001) and the Pew Research Center (1998, 2003) nd citizens to be more trusting when the president shares their party identication (see also Schaffner and Clark, 2004). This relationship between party identication and the presidents party afliation suggests that partisanship operates indirectly on trust. Specifically, party identication should inuence individuals evaluations of the president and those evaluations should inuence trust.2 Although Citrin et al.s analyses (Citrin and Green, 1986; Citrin and Luks, 2001) demonstrate the importance of presidential evaluations for trust (and they hint at the connection between partisanship and assessments of the president), they do not incorporate the indirect inuence of party afliation. In addition, we examine the roles of party control of Congress and congressional approval. As with the president, there is evidence that citizens assessments of Congress affect their trust in government (Hetherington, 1998; Hibbing and Theiss-Morse, 1995; Williams, 1985), and that the magnitude of congressional approvals inuence rivals, or exceeds, that of presidential approval (Citrin and Luks, 2001; Hetherington, 1998). As with presidential evaluations, however, previous research does not investigate the indirect effects of partisanship through congressional assessments. Furthermore, previous research treats Congress as unicameral by using a single congressional approval variable. Congress is bicameral, of course, and there is little reason to believe a priori that attitudes toward the House and Senate have the same effect on trust. Perhaps more importantly, the bicameral nature of Congress can complicate the relationship between partisanship and trust. When the same party controls both chambers, citizens identifying with that party should be more supportive of both chambers and therefore of
2 It is perhaps more common to view trust as dependent on evaluations of individuals and institutions, but trust may also inuence those evaluations (Chanley, Rudolph, and Rahn, 2000; Hetherington, 1998). We explore this relationship more fully in the proceeding discussion.

886

Social Science Quarterly

Congress as a whole. On the other hand, when the House and Senate are not controlled by the same party, examining the inuence of party identication on trust through congressional approval may be misleading. In such a situation, partisanships effect on congressional approval will be pulled in opposite directions by House and Senate approvals. More generally, partisan control of institutions should be an important determinant of citizen trust. Given this, we also maintain that changes in party control of political institutions will produce signicant changes in citizens attitudes toward government. When a political party gains control of an institution, individuals identifying with that party should become more approving of the institution and more trusting of government. Conversely, citizens identifying with the party losing control should exhibit less support for that institution and should be less trusting of government.

Data and Methods

As a midterm year, the 2002 elections afforded citizens the opportunity to vote for members of Congress but not the president. Although majorityparty status remained unchanged in the House (and, by design, the presidency), the Republicans captured the Senate. This change in party control of only one institution makes the 2002 elections particularly well suited for testing our hypotheses regarding the relationships between partisanship, party control of institutions, and trust in government. After the elections, Democrats should be less approving of the Senate and less trusting of government. Republicans should exhibit the reverse: increased Senate approval and greater trust. To test these hypotheses, we use data from two independent cross-sectional surveys of national random samples conducted by the Center for Survey Research and Analysis at the University of Connecticut before and after the 2002 congressional elections.3 We follow several recent scholars (Alford, 2001; Citrin and Luks, 2001; Hibbing and Smith, 2003) by measuring political trust using the following question: How much of the time do you think you can trust the government in Washington to do what is right? Although the questions used in the pre- and postsurveys were the same, the response options differed slightly. Specifically, the preelection survey asks respondents if they trust the government just about always, most of the time, or only some of the time. The postelection survey contains an additional almost never response option, which we have combined with the only some of the time responses for our analyses. Also central to our analysis is party identication. Here we construct the standard seven-point partisanship scale ranging from strong Democrat to
3 The preelection survey was conducted October 914, 2002 and the postelection survey was conducted April 14May 28, 2003. Both used national random-digit-dial samples.

Parties, Institutional Control, and Trust in Government

887

strong Republican. From the seven-point scale we also distinguish between Democrats, independents, and Republicans (with strong identiers, weak identiers, and independent leaners classied as partisans) and use this three-point classication in our bivariate analyses. Finally, our argument that partisanship inuences trust through party control of political institutions makes evaluations of such institutions critical. Specifically, we anticipate that evaluations of the Senate should be affected by the change in majority-party status, whereas evaluations of the House and of the presidency should exhibit little or no change since control of those institutions remained constant. In both surveys, respondents were asked to indicate approval or disapproval of the performance of George W. Bush, the Senate, and the House of Representatives. We create three-point scales with those who disapprove scored as 1, those who dont know scored as 2, and those who approve scored as 3.
Trust and Approval Before and After the 2002 Elections

Although the terrorist attacks of 9/11 caused an uptick in political trust, by the time of the 2002 elections, those effects appear to have dissipated (Hetherington and Globetti, 2003). Indeed, the preelection survey reveals political trust to be consistent with the generally low levels observed in recent years. By October 2002, only 5.9 percent of citizens reported trusting the government almost all the time, while 57.6 percent of citizens reported trusting the government only some of the time. Signicant variations across party afliates exist in the level of political trust observed prior to the elections. As Table 1 shows, more Democrats (62.9 percent) were highly skeptical of government preceding the 2002 elections than were Republicans (49.4 percent). As the Pew Research Center (2003) notes, higher levels of trust by Republicans are a departure from patterns observed in the mid-1990s. Presumably, even with the Democrats controlling the Senate by virtue of Jim Jeffordss defection from the Republican Party, Republican control of both the presidency and the House gave Republican identiers greater faith in government. Independents were even less trusting than were Democrats, a nding consistent with governmental dominance by the two major parties. Given Republican control of the House and presidency, higher Republican trust is consistent with our hypotheses. Yet our model also suggests that the Republican capture of the Senate in the 2002 elections should yield changes in citizens levels of trust according to their partisanship. Table 1 shows that Democrats and Republicans did, in fact, exhibit distinctive reactions to the elections. The percentage of Democrats reporting the highest levels of trust declined considerably (from 6.0 before the elections to 2.9 afterward), while the percentage reporting the lowest trust levels increased from 62.9 to 71.9. (Both changes are statistically signicant at the 0.05

888
TABLE 1

Social Science Quarterly

Trust in Government Before and After 2002 Elections by Party Identication Trust Government in Washington to Do What is Right Just about always Most of the time Only some of the timew Democrats Before 6.0% 31.1% 62.9% After 2.9% n 25.2% n 71.9% n n Independents Before 5.9% 30.4% 63.7% After 9.4% 30.6% 60.0% Republicans Before 5.9% 44.7% 49.4% After 8.7% n 45.4% 45.9%

n Difference in proportion between preelection and postelection signicant at 0.05 level in a onetailed z test. nn Difference in proportion between preelection and postelection signicant at 0.01 level in a one-tailed z test. w For the postelection survey almost never responses are included here. This was not a response option in the preelection Survey. NOTE: Cell entries are column percentages.

level.) As expected, Republicans became more trusting following the elections, although the magnitude of the change was modest. The share of Republicans almost always trusting the government increased a statistically signicant 2.8 percentage points from 5.9 to 8.7 percent. The rst row of Table 2 contains the mean values for political trust before and after the elections and offers further evidence of the elections effects on political trust. Indeed, Democrats trust levels were signicantly lower following the elections. On the other hand, the increases in mean values observed with regard to Republicans and independents in Table 2 are not

TABLE 2
Trust and Institutional Approval Before and After 2002 Elections by Party Identication Democrats Before Trust government in Washington Approve of the Senate Approve of the House Approve of President Bush 1.431 2.188 2.083 1.818 After 1.309 2.057 n 2.038 1.853
nn

Independents Before 1.422 2.076 2.076 2.162 After 1.494 2.388 n n 2.209 2.424 n

Republicans Before 1.565 2.282 2.484 2.807 After 1.628 2.396 n 2.566 2.883 n

n Difference between preelection and postelection mean values signicant at 0.05 level in a two-tailed t test (one-tailed test for Senate approval for Democrats and Republicans). nn Difference between preelection and postelection mean values signicant at 0.01 level in a two-tailed t test (one-tailed test for Senate approval for Democrats and Republicans). NOTE: Cell entries are mean values.

Parties, Institutional Control, and Trust in Government

889

statistically signicant. Nonetheless, differing reactions to the 2002 elections did serve to widen the partisan gap in trust. Although Republicans mean trust was 0.134 greater than that of Democrats in the preelection survey, the difference increased to 0.319 following the elections. We contend that citizens are more trusting of government when it is controlled by their preferred party. Therefore, we expect partisans to evaluate political institutions differently in the wake of the 2002 elections. Specifically, Democrats should become less approving of the Senate due to the Republican ascension to majority status while Republicans should become more approving. On the other hand, the lack of change in the party controlling the House and the presidency suggest that approval levels for those institutions should be relatively stable. Evidence in Table 2 generally supports these hypotheses. For both Democrats and Republicans, approval of the Senate is signicantly different following the Republican takeover: Democrats became less approving while Republicans became more so.4 Furthermore, the lack of change in approval of the House bolsters the case that institutional approval is sensitive to party control. It also suggests that at least some citizens have a reasonable understanding of the bicameral structure of Congress and are able to evaluate the two chambers separately. Somewhat surprisingly, presidential approval among Republicans and independents increased in the postelection surveys despite the presidency not being on the ballot. We believe the war in Iraq is mostly responsible. The postelection survey was conducted between April 14 and May 28, 2003 and likely captures reactions to the most successful portions of the war. Baghdad fell on April 9, before the rst postelection interviews were conducted. On May 1, while the survey was in the eld, President Bush declared the end of major combat operations. As a result, Republicans and independents may have rallied around the president and government in general. This rally effect should be greatest among independents because they were not predisposed to be as supportive of the president. Indeed, Table 2 shows that trust and institutional approval increased the most for independents. Even among independents, however, the increases in trust and House approval were not statistically signicant. Combined with the failure of Democrats to jump on the war-in-Iraq bandwagon, this indicates the absence of an undifferentiated, across-the-board increase in citizen assessments of the government.

Both the trust and approval variables in Table 2 are ordinal level and strictly speaking, therefore, calculating and comparing mean values is not appropriate. We have chosen to do so for ease of presentation and because this is common practice in political science. We note, however, that we did also examine changes in trust and approval by comparing the proportions of observations in the three response categories and all conclusions from such analyses are substantively the same as those presented here.

890
Explaining Trust and Approval

Social Science Quarterly

The preceding analyses indicate a relationship between partisanship, party control of institutions, and evaluations of government. But to what extent can trust in government be explained by partisanship? To answer this question, we pool respondents from the two cross-sectional surveys and use ordinary least squares regression to estimate a model of political trust. Partisanship is central in our examination, and our model includes the traditional seven-point party identication item, with 7 indicating a strong Republican. The ndings above suggest that party identication should positively affect trust, largely through institutional evaluations. To test for change in the relationship between partisanship and trust before and after the election, we include a dummy variable for postelection respondents and an interaction term between party identication and the postelection dummy. Regardless of partisanship, approval of President Bush and the two chambers of Congress should promote citizen trust. Finally, we include education, income, race, and gender as control variables. Consistent with our expectations, the rst column of Table 3 shows that presidential approval and assessments of the House and Senate have positive effects on trust. Furthermore, the magnitude of the Senate approval coefcient is approximately the same as that of the presidential approval variable and, together, House and Senate approval has a greater effect on trust than does presidential approval. Party identications effect is more complicated. As indicated by the partisanship variable and its interaction with the postelection dummy, there does not appear to be a direct effect of partisanship on trust before the election. On the other hand, the impact of party identication differs for the postelection respondents. Specifically, the coefcient estimate for the interaction term indicates that partisanship emerged as signicantly more important following the elections, with a direct link between strong Republican identication and higher trust. We suspect that the change from divided to unied government may help explain this. Gaining control of an institution should foster greater approval of the institution from fellow partisans; this, in turn, should promote trust. However, the nature of the American political system as one of separated institutions with shared powers complicates this relationship. Although identiers of the party gaining control of an institution should become more trustful, this effect may be tempered when the change in partisan control of that institution does not accompany a shift from divided to unied government. For example, consider an election where the Republicans maintain control of the presidency, the Democrats maintain control of Senate, and the Republicans wrest control of the House of Representatives from the Democrats. Republican identiers should gain faith in the House and the government as a whole. Yet, increased trust is mitigated by the recognition that Democrats in the Senate still have the ability to thwart Republican initiatives. Thus, party identication may have

Parties, Institutional Control, and Trust in Government


TABLE 3
Explaining Trust in Government Pooled Constant Party identication (7-point scale) Postelection Postelection Party identication Bush approval House approval Senate approval Education Income Race (1 5 white) Gender (1 5 male) Support war in Iraq Approve of process Total knowledge N R2 0.833 n n (0.080) 0.002 (0.010) 0.170 n n (0.059) 0.028 n (0.013) 0.113 n n (0.020) 0.080 n n (0.021) 0.111 n n (0.020) 0.013 (0.013) 0.021 (0.013) 0.001 (0.035) 0.004 (0.028) 1,660 0.15 Pooled 1.472 n n (0.070) 0.029 n n (0.010) 0.231 n n (0.062) 0.043 n n (0.013) 0.038 n n (0.014) 0.028 n (0.013) 0.028 (0.037) 0.002 (0.029) 1,674 0.05 Postelection 0.443 n n (0.108) 0.025 n (0.011) 0.065 n (0.032) 0.079 n n (0.030) 0.056 n (0.028) 0.002 (0.020) 0.018 (0.016) 0.028 (0.051) 0.003 (0.038) 0.023 (0.017) 0.205 n n (0.024) 0.189 n n (0.073) 816 0.27

891

Postelection 0.615 n n (0.106) 0.044 n n (0.010) 0.003 (0.020) 0.020 (0.016) 0.018 (0.052) 0.002 (0.039) 0.059 n n (0.015) 0.249 n n (0.024) 0.283 n n (0.072) 822 0.23

n Signicant at 0.05 level; n nsignicant at 0.01 level (two-tailed tests). NOTE: Cell entries are unstandardized coefcient estimates with standard errors in parentheses.

a direct effect on trust under unied government but not under divided government. Our model assumes that institutional approval affects political trust. However, some earlier research suggests that institutional approval may also be affected by trust (Chanley, Rudolph, and Rahn, 2000; Hetherington, 1998). This possibility of reciprocal causality thus calls into question our model specication. Unfortunately, data limitations do not permit the estimation of a system of simultaneous equations and full exploration of the causal relationships between these variables. Instead, we reestimate our original equation without the potentially endogenous approval variables and

892

Social Science Quarterly

evaluate the robustness of the results, which can be found in the second column of Table 3. The results suggest some important differences. Party identication becomes signicant, once again suggesting that some of partisanships effect on trust operates through evaluations of institutions. Perhaps most important, however, is what does not change: party identication continues to wield signicantly greater inuence on trust following the elections. Pooling the pre- and postelection surveys required estimating a model for which data were available in both cross-sections. As a consequence, another potential pitfall in our initial model is the exclusion of other variables potentially affecting trust. The most obvious candidate here is support for the war in Iraq, a variable not available in the preelection survey. To examine the possibility that these ndings are a byproduct of omitted variables, we estimate an expanded version of the trust model for the postelection survey by including support for the war in Iraq and two other variables available only in the postelection survey: approval of the policy-making process and political knowledge.5 Previous research suggests that political information has a negative effect on political trust because those who know more about politics also learn about its aws and realize the extent to which democratic government falls short of expectations (Hibbing and TheissMorse, 1995). Conversely, we anticipate both support for the war in Iraq and approval of the policy-making process to have positive effects on trust. The coefcient estimates are presented in the third column of Table 3. Even with the additional variables, partisanship and institutional approval continue to signicantly inuence trust in the expanded model, thereby giving us greater condence in concluding that these variables do indeed operate as anticipated. Although partisanship and institutional approval are still signicant, the new variables add considerably to the explanatory power of the model. Both process approval and political knowledge have signicant effects in the anticipated directions and, as standardized coefcient estimates reveal, the effect of process approval is three times greater than that of party identication and is equal to the combined effects of all three institutional approval variables. Finally, while the coefcient estimate of support for the war in Iraq is positive, it is somewhat surprising that it is not statistically signicant. Due to potential endogeneity, we also reestimate the expanded postelection model without the approval variables. With the exception of support for the war in Iraq (which becomes statistically signicant), the results in the

The political knowledge variable is an index measuring correct responses to the following six items: the ofce held by Dick Cheney; who determines the constitutionality of laws; the size of the majority needed to override a presidential veto; partisan control of the House; partisan control of the Senate; and which party is more conservative. The policy-making process approval variable uses an item asking respondents how much of the time they approve of the process that the government in Washington uses to make policies.

Parties, Institutional Control, and Trust in Government

893

nal column of Table 3 point to the robustness of our ndings. Specifically, party identication, process approval, and political knowledge all continue to operate in the same manner. It is also the case, however, that the results again suggest important indirect effects of party identication on trust. In the absence of the approval variables, the magnitude of the party identication coefcient nearly doubles.

Indirect and Total Effects of Partisanship on Trust: A Path Analysis

The regression models estimated above indicate the direct impact of party identication on trust in government in the aftermath of the 2002 elections. However, we argue that much of partisanships effect is channeled through approval of political institutions; differences between models with and without the approval variables suggest this to be the case. To more directly examine this, we turn to path analysis using the postelection data. The model employs the same variables used in the expanded regression analysis as depicted in Figure 1. The model includes approval of President Bush, the House, and the Senate as dependent variables inuenced by party identication, attitudes toward the war, approval of the political process, and political knowledge.6 Finally, the model includes the war in Iraq as both a direct and indirect inuence on trust and as a dependent variable of party identication. As with the earlier regression models, the path model depicts causality between approval and trust as owing in only one direction, thereby ignoring potential endogeneity. Thus, our results must be interpreted with a note of caution. Nevertheless, robustness of our earlier ndings to model specication suggests that our path results should be relatively stable. Table 4 displays effects for the path model. The rst three columns show that approval of the political process, political knowledge, and attitudes toward the war all signicantly inuence approval of the president and both chambers of Congress. Party identication is also a signicant determinant of institutional approval. As citizens move from strong Democrat to strong Republican, they are more likely to approve of the president, the Senate, and the Houseall of which were controlled by the Republican Party. The fourth column shows that party identication is also a signicant predictor of attitudes toward the war in Iraq, with Republicans exhibiting greater approval than Democrats. Furthermore, the fth column reveals that party identication, process approval, and institutional approval (but not the demographic controls) directly affect trust.
6 The model for approval of President Bush also includes income. There is reason to suspect that those with higher income would be more approving of the president and the inclusion of income in this equation allows for the model to be identied.

894
FIGURE 1

Social Science Quarterly

Path Model Showing the Relationship Between Party Identication and Trust
Party identification

House approval

Senate approval War in Iraq Bush approval

Trust

Process approval Political knowledge

Income

Education Race Gender

In addition to direct effects, party identication exerts important indirect effects on trust through attitudes toward the war and through institutional approval, as shown in the sixth, seventh, eighth, and ninth columns of Table 4. The last column presents the total effects (the sum of the direct and indirect effects) of each of the variables on political trust. By comparing the total effect of partisanship on trust with its direct effect, we see that over 50 percent of partisanships total effect does, indeed, take place behind the scenes. At the same time that party identication is important in explaining trust, we see that a full understanding also requires taking into account citizens attitudes toward the current political environment such as the war in Iraq. The last column also underscores the importance of underlying assessments of the political process. The total effects of process approval are considerably greater than those of party identication and are equal to the combined effects of partisanship and support for the war.

TABLE 4

Direct, Indirect, and Total Effects Related to Postelection Trust in Government Direct Effects On Bush Approval On Trust 0.022 0.027 0.028 0.017 0.007 0.014 0.023 0.018 0.330 0.466 0.496 n n 0.081 n n 0.084 n n 0.014
nn nn

Indirect Effects on Trustw On War in Iraq Through Through Through Through War Bush Senate House Approval Approval Approval in Iraqww 0.032 0.047 0.008 0.008 0.001 0.041 Total Effects on Trust 0.193 n n 0.144 n n 0.333 n n 0.137 n n 0.038 0.096 n 0.116 n n 0.085 n 0.003 0.018 0.002

On House Approval
nn

On Senate Approval

Parties, Institutional Control, and Trust in Government

Party identication Support war in Iraq Process approval Total knowledge Income Bush approval House approval Senate approval Education Race (1 5 white) Gender (1 5 male) N R2 819 0.55 819 0.22 0.09 n n 0.055 0.274 n n 0.095 n n 0.037 0.096 n 0.116 n n 0.085 n 0.003 0.018 0.002 816 0.27

0.191 0.234 n n 0.245 n n 0.143 n n

nn

0.086 0.171 n n 0.270 n n 0.213 n n

818 0.25

818 0.19

n Signicant at 0.05; n nsignicant at 0.01 (one-tailed tests); wthe signicance of indirect effects are not generally computed for path models; wwindirect effect of party identication on trust through war in Iraq and then through House approval (0.013), Senate approval (0.007), and Bush approval (0.022). NOTE: Direct effects are standardized coefcient estimates. Indirect effects are the standardized coefcient estimates from institutional approval multiplied by the direct effect of institutional approval on trust. Total effects are the sum of the direct and indirect effects.

895

896

Social Science Quarterly

Partisanship, Party Control, and Self-Reported Change in Trust

We argue that the change in party control of the Senate in 2002 led to differing changes in individuals trust in government as a function of their party identication. By examining pre- and postelection surveys, we uncover evidence consistent with our hypotheses. Because these surveys involve different respondents, however, comparing the results of the surveys does not permit for direct analysis of change at the individual level. Fortunately, the postelection survey contains an additional item that asks respondents whether they trust the government in Washington to do whats right more than you did a year ago, less than you did a year ago, or about the same as you did a year ago. Thus, we have a measure of self-perceived change in trust. Admittedly, there may be potential difculties with such a measure. For example, respondents may have no real recollection of how much they trusted the government a year earlier and simply guess about how their attitudes might have changed. Similarly, they may report changes in trust based on their current trust: if they have low trust now they may be prone to report a decline even if no change occurred. On the other hand, compared to using panel responses, this question may allow for more precise measurement of change. Since the generic trust question offers only three levels of trust (only some of the time, most of the time, almost always), considerable movement may be occurring within each category. For example, someone who trusts the government 30 percent of the time before the elections and 55 percent of the time after the elections has experienced considerable change in trust. However, that same individual would have been likely to answer only some of the time on both occasions in a panel survey. The question measuring self-perceived change in political trust avoids this pitfall by measuring whether any change occurred, be it within or between the broad categories usually employed. To examine the effects of partisanship on self-perceived change in trust, we include all the independent variables used to explain trust in the expanded postelection trust model. We anticipate that Republican identiers experienced an increase in trust given the Republican takeover of the Senate. However, this dynamic should not operate the same for all citizens; political knowledge likely conditions the relationship between partisanship and trust. As Citrin and Luks (2001:19), note, the politically knowledgeable have an image of government that differentiates among specific institutions. Similarly, Bernstein (2001) nds that partisan makeup of institutions is more important in attitudes toward those institutions among the politically sophisticated. Thus, it is for those individuals who follow politics more closely, understand the signicance of the system of separated institutions with shared powers, and know which party currently controls which institutions that the effects of party identication should be most pronounced. Consequently, we anticipate the effects of party identication to be dependent on knowledge of party control, with Republicans knowing that

Parties, Institutional Control, and Trust in Government

897

their party has a majority in the House and the Senate being the most likely to report increased trust and Democrats aware of party control being the least likely to do so. To capture these relationships, we include interaction terms between partisanship and knowledge that the Republican Party controls the Senate and between partisanship and knowledge that the Republican Party controls the House.7 Finally, we include an interaction between overall political knowledge and party identication, suspecting that party differences in perceived change in trust will be most pronounced among individuals who are more politically aware. The results of the regression analysis in the rst column of Table 5 show that stronger identication with the Republican Party led citizens to report greater increases in trust over the previous year. This is clearly consistent with our hypothesis that party control of institutions affects trust. In addition, we see that the relationship between party identication and change in self-perceived trust is mediated by awareness of institutional control. Both the Senate knowledge and House knowledge interaction terms are positive and signicant, indicating that different reactions to the 2002 elections were more pronounced among individuals cognizant of party control of institutions. On the other hand, the total (political) knowledge interaction term is insignicant, thereby underscoring the importance of the distinction between overall knowledge and knowledge of specific institutions (McGraw and Pinney, 1990). As with the models of trust, presidential approval and House approval also positively inuence self-reported change in trust. Somewhat surprisingly, however, Senate approval fails to have a signicant effect, perhaps because the effects of partisanship (and its interaction with knowledge of Republican control of the Senate) simply overshadow it. Finally, attitudes toward the war and approval of the political process also play an important role in helping explain perceived change in trust. The second column of Table 5 presents results for a model excluding the potentially endogenous approval variables. Other than total knowledge becoming statistically signicant, the ndings are substantively the same. These results give us greater condence in our conclusions regarding the role of party identication and its interactions with knowledge of institutional control.

Discussion

The most notable trend with regard to trust in the national government over the past 35 years is its decline (e.g., Lipset and Schneider, 1987; Rosenstone and Hansen, 1993; Hibbing and Theiss-Morse, 1995; Orren, 1997; Hetherington, 1998; Pew Research Center, 1998). At the same time
7 Of the 1,006 respondents in the postelection survey, 72 percent correctly identied the Republicans as having a majority in the House and 61 percent correctly identied the Republicans as having a majority in the Senate.

898
TABLE 5

Social Science Quarterly


Explaining Self-Perceived Change in Trust

Constant Party identication (7-point scale) PID Senate knowledge PID House knowledge PID Total knowledge Bush approval House approval Senate approval Education Income Race (1 5 white) Gender (1 5 male) Support war in Iraq Approve of process Total knowledge N R2

0.829 n n (0.129) 0.044 n (0.021) 0.029 n n (0.011) 0.030 n n (0.012) 0.046 (0.034) 0.186 n n (0.031) 0.074 n n (0.029) 0.046 (0.027) 0.001 (0.019) 0.005 (0.015) 0.034 (0.049) 0.023 (0.037) 0.081 n n (0.017) 0.067 n n (0.023) 0.147 (0.140) 816 0.38

1.019 n n (0.128) 0.057 n n (0.021) 0.028 n n (0.011) 0.034 n n (0.012) 0.022 (0.035) 0.002 (0.019) 0.006 (0.016) 0.040 (0.050) 0.024 (0.037) 0.141 n n (0.015) 0.090 n n (0.023) 0.305 n (0.141) 821 0.34

n Signicant at 0.05 level; n nsignicant at 0.01 level (two-tailed tests). NOTE: Cell entries are unstandardized coefcient estimates with standard errors in parentheses.

that trends in trust for various social and political groups largely mirror one another, there are also some noticeable differences across groups (Citrin and Luks, 2001). Although our model does nothing to explain the broader decline in trust, it does help account for partisan-related variation in trust, including greater Republican trust relative to Democrats since the mid1990s (Pew Research Center, 2003), and offers clear expectations regarding reactions to future changes in the political environment. Our interest in partisanships inuence on trust is not meant to obscure other determinants of trust. In fact, while our ndings reveal the importance of party identication and party control of institutions, they also indicate

Parties, Institutional Control, and Trust in Government

899

that current political events inuence trust. Even more obviously, our results point to the signicance of citizens attitudes toward the political process. Thus, Hibbing and Theiss-Morses (1995, 2002) attention to process assessments remains central to understanding trust in government and a decrease in process approval may ultimately be the key to accounting for the longer-term decline in Americans trust. Having acknowledged the continued importance of other variables in understanding trust, our research centers around the claim that party identication acts as an important shortcut through which citizens understand government and politics, and that party identication and party control of institutions consequently play a prominent role in the trust calculus. Namely, citizens have greater faith in the party with which they identify and are therefore more approving of governmental institutions when their party controls them and are more trusting of the government as a whole. The 2002 elections provide an excellent case study for examining the relationships between party identication, party control of institutions, institutional approval, and trust, since they brought about a change in party control of the Senate while control of the House and presidency remained the same. As anticipated, Democrats and Republicans reacted differently to the Republican takeover of the Senate in 2002. At the aggregate level, Republicans became more approving of the Senate and more trusting of government while the opposite was true for Democrats. Individual-level analyses reveal the important role of party identication in this process. A citizen is more likely to approve of institutions when they are controlled by the party with which the citizen identies and this institutional approval, in turn, inuences trust. Furthermore, citizens report changes in trust in reaction to signicant developments in the political landscape, changes that vary as a function of their party identication. Party identication did not have a direct effect on trust prior to the 2002 elections, but did exert a direct inuence after the election. Much as Fiorina (1996) argues that some citizens have a preference for divided government as a way of balancing Democrats and Republicans, it appears that some individuals exhibit greater faith in the government when their preferred party controls all governmental institutions. We concur with Citrin and Luks (2001:24) that the role of divided (vs. unied) government on political trust deserves future attention. The changes in approval and trust resulting from the 2002 elections suggest that at least some segment of the population reacts in reasonable ways to the political environment. In addition to the ascension of a new president, some citizens are aware of more subtle changes, such as party control of the House or Senate, and update their views of government accordingly. Our ndings also suggest that some citizens (likely the same individuals responding to changes in institutional control) understand the difference between unied and divided government and that their attitudes toward government depend at least in part on this distinction.

900
REFERENCES

Social Science Quarterly

Alford, John R. 2001. Were All in This Together: The Decline of Trust in Government, 19581996. In John R. Hibbing and Elizabeth Theiss-Morse, eds., What is it about Government that Americans Dislike? New York: Cambridge University Press. Bartels, Larry M. 2000. Partisanship and Voting Behavior, 19521996. American Journal of Political Science 44:3550. Bernstein, Jeffrey L. 2001. Linking Presidential and Congressional Approval During Unied and Divided Governments. In John R. Hibbing and Elizabeth Theiss-Morse, eds., What is it about Government that Americans Dislike? New York: Cambridge University Press. Black, Gordon S., and Benjamin D. Black. 1994. The Politics of American Discontent. New York: Wiley. Blendon, Robert J., John M. Benson, Richard Morin, Drew E. Altman, Mollyann Brodie, Mario Brossard, and Matt James. 1997. Changing Attitudes in America. In Joseph S. Nye, Jr., Philip D. Zelikow, and David C. King, eds., Why People Dont Trust Government. Cambridge, MA: Harvard University Press. Chanley, Virginia A., Thomas J. Rudolph, and Wendy M. Rahn. 2000. The Origins and Consequences of Public Trust in Government: A Time Series Analysis. Public Opinion Quarterly 64:23956. Citrin, Jack. 1974. Comment: The Political Relevance of Trust in Government. American Political Science Review 68:97388. Citrin, Jack, and Donald Philip Green. 1986. Presidential Leadership and the Resurgence of Trust in Government. British Journal of Political Science 16:43153. ja ` Vu All Over Citrin, Jack, and Samantha Luks. 2001. Political Trust Revisited: De Again?. In John R. Hibbing and Elizabeth Theiss-Morse, eds., What is it about Government that Americans Dislike? New York: Cambridge University Press. Converse, Philip E. 1964. The Nature of Belief Systems in Mass Publics. In David Apter, ed., Ideology and Discontent. New York: Free Press. Craig, Stephen C. 1996. Change and the American Electorate. In Stephen C. Craig, ed., Broken Contract: Changing Relationships Between Americans and Their Government. Boulder, CO: Westview Press. Delli Carpini, Michael X., and Scott Keeter. 1996. What Americans Know about Politics and Why it Matters. New Haven, CT: Yale University Press. Downs, Anthony. 1957. An Economic Theory of Democracy. New York: Harper Collins. Easton, David. 1965. A Systems Analysis of Political Life. New York: Wiley. Fiorina, Morris. 1996. Divided Government, 2nd ed. Needham Heights, MA: Allyn and Bacon. Gamson, William A. 1968. Power and Discontent. Homewood, IL: Dorsey Press. Hager, Gregory, and Jeffrey Talbert. 2000. Look for the Party Label: Party Inuences on Voting in the U.S. House. Legislative Studies Quarterly 25:7599. Hetherington, Marc J. 1998. The Political Relevance of Political Trust. American Political Science Review 92:791808. . 1999. The Effect of Political Trust on the Presidential Vote: 19681996. American Political Science Review 93:31126.

Parties, Institutional Control, and Trust in Government

901

. 2001. Declining Political Trust and a Shrinking Public Policy Agenda: Why Media Scholars Should Care. In Roderick P. Hart and Daron R. Shaw, eds., Communication in U.S. Elections: New Agendas. New York: Rowman and Littleeld. Hetherington, Marc J., and Suzanne Globetti. 2002 Political Trust and Racial Policy Preferences. American Journal of Political Science 46:25375. . 2003. The President and Political Trust. In Michael Nelson, ed., The Presidency and the Political System, 7th ed. Washington, DC: CQ Press. Hibbing, John R., and James T. Smith. 2003. Is it the Middle that is Frustrated? Americans Ideological Positions and Governmental Trust. Paper presented at the Annual Meeting of the Midwest Political Science Association. Chicago, IL. Hibbing, John R., and Elizabeth Theiss-Morse. 1995. Congress as Public Enemy: Public Attitudes Toward American Political Institutions. New York: Cambridge University Press. . 2001. What is it about Government that Americans Dislike? New York: Cambridge University Press. . 2002. Stealth Democracy: Americans Beliefs about How Government Should Work. New York: Cambridge University Press. Kimball, David C., and Samuel C. Patterson. 1997. Living Up to Expectations: Public Attitudes Toward Congress. Journal of Politics 59:70128. King, David C. 1997. The Polarization of American Parties and Mistrust of Government. In Joseph S. Nye, Jr., Philip D. Zelikow, and David C. King, eds., Why People Dont Trust Government. Cambridge, MA: Harvard University Press. Lawrence, David G. 2001. On the Resurgence of Party Identication in the 1990s. In Jeffrey E. Cohen, Richard Fleisher, and Paul Kantor, eds., American Political Parties: Decline or Resurgence? Washington, DC: CQ Press. Lawrence, Robert Z. 1997. Is it Really the Economy, Stupid? In Joseph S. Nye, Jr., Philip D. Zelikow, and David C. King, eds., Why People Dont Trust Government. Cambridge, MA: Harvard University Press. Lipset, Seymour Martin, and William. Schneider. 1987. The Condence Gap. Baltimore, MD: Johns Hopkins University Press. Lupia, Arthur, and Mathew D. McCubbins. 1998. The Democratic Dilemma: Can Citizens Learn What They Need to Know? New York: Cambridge University Press. Luskin, Robert C. 1987. Measuring Political Sophistication. American Journal of Political Science 31:85699. McGraw, Kathleen, and Neil Pinney. 1990. The Effects of General and Domain-Specific Expertise on Political Memory and Judgement. Social Cognition 8:930. Miller, Arthur H. 1974. Political Issues and Trust in Government, 196470. American Political Science Review 68:95172. . 1983. Is Condence Rebounding? Public Opinion 6:1620. Miller, Arthur H., and Stephen Borrelli. 1991. Condence in Government During the 1980s. American Politics Quarterly 19:14773. Nelson, Michael. 2003. Why Americans Hate Politics and Politicians. In David Canon et al., eds., Enduring Debate: Classic and Contemporary Readings in American Politics, 3rd ed. New York: W. W. Norton.

902

Social Science Quarterly

Orren, Gary. 1997. Fall from Grace: The Publics Loss of Faith in Government. In Joseph S. Nye, Jr., Philip D. Zelikow, and David C. King, eds., Why People Dont Trust Government. Cambridge, MA: Harvard University Press. Patterson, Thomas. 1993. Out of Order. New York: Knopf. Pew Research Center for the People and the Press. 1998. Deconstructing Distrust: How Americans View Government. Washington, DC: Pew Research Center for the People and the Press. . 2003. The 2004 Political Landscape: Evenly Divided and Increasingly Polarized. Washington, DC: Pew Research Center for the People and the Press. Popkin, Samuel. 1994. The Reasoning Voter: Communication and Persuasion in Presidential Campaigns, 2nd ed. Chicago, IL: University of Chicago Press. Rosenstone, Steven J., and John Mark Hansen. 1993. Mobilization, Participation, and Democracy in America. New York: Macmillan. Schaffner, Brian F., and John A. Clark. 2004. MacroTrust: Partisanship and the Dynamics of Trust in Government. Paper presented at the Annual Meeting of the Southern Political Science Association. New Orleans, LA. Scholz, John T., and Mark Lubell. 1998. Trust and Taxpaying: Testing the Heuristic Approach to Collective Action. American Journal of Political Science 42:398417. Sniderman, Paul M., Richard A. Brody, and Philip E. Tetlock. 1991. Reasoning and Choice: Explorations in Political Psychology. New York: Cambridge University Press. Williams, John T. 1985. Systemic Inuences on Political Trust: The Importance of Perceived Institutional Performance. Political Methodology 11:12542.

Vous aimerez peut-être aussi