“How Trust Matters: The Changing Political Relevance of Political Trust”
Marc J. Hetherington
Professor
Department of Political Science
VanderbiltUniversity
and
Jason A. Husser
Ph.D. Candidate
Department of Political Science
VanderbiltUniversity
The authors thank Jon Hurwitz, Tom Rudolph, David Lewis, Suzanne Globetti, Cindy Kam, Liz Zechmeister, Larry Evans, and John Hudak for their helpful suggestions and assistance.
How Trust Matters: The Changing Political Relevance of Political Trust
Abstract
Americans most often thinkabout government in terms of its ability to grapple with issues of redistribution and race. However, the September 11 terrorist attacksled to a massive increase in media attention to foreign affairs, which causedpeople to think about the government in terms of defense and foreign policy. We demonstrate that this dramatic change in issue salience alteredthepolicy preferences that political trust shapedfor a time. Specifically, we show that trustdid not affect attitudes about the race-targeted programsas it usually does in 2004, but instead affected a range of foreign policy and national defense preferences. By merging survey data gathered from 1980 through 2004 with data from media content analyses, we show that, in general, trust’s effectson defense and racialpolicy preferences, respectively, increaseas the media focus more attention in these areas and decrease when that attention ebbs.
The scholarly consensus about political trust has evolved over time.For three decades after trust began to plummet in the late 1960s, scholars found little evidence to suggest that its decline hada meaningful impact on American public opinion, electoral behavior, or policy outputs (e.g. Citrin 1974; Citrin and Green 1986).More recently, however, scholars have demonstrated that declining trust has had a range of effects.Most central to our purposes, less trust in government hasled to decreasing support forgovernment spending onredistributive policies(Hetherington and Globetti 2002; Rudolph and Evans 2005), a less liberal policy mood (Chanley, Rudolph, and Rahn 2000), and the implementation of fewer liberal public policies (Hetherington 2005).
Although we, too, argue that trust matters, we believe that the revisionist understanding is incomplete.Using individual level data, we show that theparts of government thatbecome salient at particular points in timecan affecthow trust matters – or, more precisely, which types of opinions political trust affects.Usingdata gatheredmostly from the 1980s through 2000, previous scholarshipprovides strong evidence that political trust shapespublic support for liberal big government initiatives and a range of race targeted programs, at least among those asked to make material (Hetherington and Globetti 2002; Hetherington 2005) or ideological sacrifices (Rudolph and Evans 2005).These findings make theoretical sense because the size of government was the dominant political cleavageof that period and the racialization of several programs was central to Americans’ understanding of government (Gilens 1999; Jacoby 1994).When people conjured an image of the federal government during this period, they disproportionately thought about welfare recipients, rather thanolder people, the foreign policy establishment, or the military (Hetherington 2005, Chapter 2).
Because “the government” is large and amorphous, however, the public’sperception of what it is and what it doescanchange over time in response to changing stimuli.Most relevant to our study, the September 11, 2001 terrorist attacksushered in a fundamentally different understanding of government’s main role from that of the 1990s.As evidence, the percentage of people identifying a foreign policy issue as the country’s most important problem went from the low single digits in the months before September 11 to nearly 90 percent in the months after.[1]That percentage fluctuated mostly between 40 and 60 percent for most of the Bush years.These data suggest that the public began to think ofgovernment less as redistributingscarce resourcesand more as protecting the safety and security of Americans.
In this paper, we demonstrate that such changes in issue salience can affectthepolicy preferences that political trust shapes.First, we examine data from the recent past to show that, after the 9/11 attacks,political trust ceasedto affect preferences about redistribution and race-targeted programs for a time and began to affect defense and foreignpolicy preferences.Next, we merge data from media content analyses with survey data collected between 1980 and 2004 to show that the magnitude of trust’s effect in a policy domain depends on how much attention the media provide that issue domain.When race and redistribution receive more media coverage, as was the case in the late 1980s and 1990s, the effect of political trust on race-targeted policy preferencesis at its largest.Similarly, when media attention on defense and foreign policy is high, as was the case in the mid-1980s and in 2004, trust’s effect on defense spending preferences is at its largest. When these domains receive less coverage, the magnitude of trust’s effect on them diminishes in kind.
How and When Political Trust Matters
We follow the lead of most empirically minded scholars in defining political trust as the ratio of people’s evaluation of government performance relative to their normative expectations of how government ought to perform(Stokes 1962; Miller 1974; Coleman 1990; Hetherington 2005).This conceptual definition squares well with findings that suggest changes in trust are most often a function of changes in perceived performance on important problems like the economy (Hetherington 1998; Citrin and Green 1986; Weatherford 1984) and the incidence of scandals (Chanley, Rudolph, and Rahn 2000; Keele 2007), although some longer term factors, namely social trust, have contributed to fluctuations in political trust as well (Keele 2007).
Trust has important implications for policy preferences insofar as it can act as a valuable heuristic. Since the actual inner-workings of government are opaque in most areas and since developing constrained belief systems is too taxing for many (e.g. Converse 1964), trust provides a simple decision rule. If a person trusts government – the entity that produces and administers public policies – it follows that he or she would be more likely to support more government involvement; if not, then less. The need for trust isparticularly important theoretically when people are asked to make sacrifices for programs from which they do not perceive they benefit.To support more government spending or intervention in these areas, people must trustthe government to think itsprograms will produce societal benefits andnot waste resources(see Hetherington and Globetti 2002; Hetherington 2005; Rudolph and Evans 2005).
Our work here is complementary but more clearly elucidatesthe theoretical foundations of trust’s effects. We, too, view trust in government as a powerful heuristic, yet we argue that this heuristic can mean different things at different times. To understand how trust matters, it is critical to understand what parts of government are salient to the public, which, in turn, influences the specific preferences for government action that trust affects.Since government is vast and performs a wide array of tasks, it is particularly susceptible to framing effects (see e.g. Gamson and Modigliani 1989; Jacoby 2000; Kellstedt 2000)[2]. By framing effects we mean “different presentations of an issue [that] generate different reactions among those who are exposed” (Jacoby 2000, 751). Depending on the nature of events, the media and political elites could highlight a nearly limitless range of frames about the federal government – the provider of welfare state programs like food stamps, the arbiter of conflicts between business and labor, the root of broad based programs like Social Security, or the engine behind the military’s efforts to ensure Americans’ safety and security to name just a few.
Theories of information processing suggest that, when people are asked to consider how much they trust the government, they will draw upon the material that different frames make accessible in memory (Fazio 1986; Zaller 1992). Elite emphasis on the government as redistributor of wealth ought to make accessible constructs about “big government” and race (Gilens 1999), whereas emphasis on the government as protector of Americans’ safety ought to make accessible constructs about the defense and foreign policy apparatus, and so forth.[3] When people evaluate government, they sample from all these potential considerations, drawing disproportionately on things that are most accessible (see e.g. Zaller 1992).
Indeed, some existing evidence suggests that elite frames mightaffect the ingredients that make up people’s evaluations of government. Using a range of feeling thermometers about politicians, institutions, and politically relevant groups, Hetherington (2005, Chapter 2) estimated a model of people’s feelings about “the federal government in Washington” using data from 1996. Feelings about the president, Congress, and the Supreme Court had the largest effects by far. But, controlling for these factors, feelings about people on welfare, labor unions, and big business – all traditional New Deal, Great Society groups – also had significant effects. These results are consistent with the notion that traditional New Deal and Great Society considerations were most salient in people’s understanding of government in the 1990s. Accordingly, and as our theory predicts, scholars have shown that political trust affects preferences in this policy domain (Hetherington and Globetti 2002; Rudolph and Evans 2005; Hetherington 2005).
Importantly, these studies focused primarily on the 1980s and 1990s, a time when race-targeted and redistributive programs represented the dominant political cleavage (Carmines and Stimson 1989). As a result, race and redistribution were consistently salient considerations. When health care became similarly salient in 1994 with the introduction of the Clinton-style reform effort, however, political trust profoundly affected preferences in this domain as well (Hetherington 2005, Chapter 7), which is also consistent with the issue salience hypothesis. Further suggesting the importance of salience, trust ceased to have this effect on health care preferences two years later after the issue disappeared from the public agenda.[4]
Like the brief but intense emergence of health care reform during the Clinton years, September 11 caused a profound change in issue salience, which ought to affect which considerations are most available to people when they are asked to evaluate the government. Evidence suggests it did. When we replicated Hetherington’s feeling thermometer model using data from 2004, we found that feelings about people on welfare, labor unions, and big business were no longer statistically significant as they were in 1996. The two variables that exhibit substantially stronger effects on evaluations of the federal government in 2004 relative to 1996 are feelings about the military and feelings about liberals, a group that was perceived as staunchly anti-war by November 2004.[5] Furthermore, it stands to reason that political trust would come to affect policy preferences connected to the parts of government people are thinking about when they evaluate it.
The importance about issue salience is not novel to political trust. Scholars have shown that issue salience conditions the effectsthat other symbolic attitudes, most notably partisanship and ideology, have on policy preferences. For example, Carmines and Stimson (1989, Chapter 6) demonstrate that partisanship’s effect onpeople’s preferences for New Deal programs decreased markedly in the 1950s and 1960s, while, at the same time,its effect on preferences for racial desegregation increased dramatically. During this period, of course, American politics was experiencing an “issue evolution” in which race was challenging the New Deal dimension as the dominant issue cleavage. Layman (2001) shows a similar change in the strength of the relationship between partisanship and preferences on moral issues as they became more salient in the 1980s and 1990s (see also Hetherington and Weiler 2009).Stenner (2005) reveals similar changes in the effects of ideology. When economic issues become more salient, the relationship between ideology and economic policy preferences grows. When morally traditional or authoritarian issues become more salient, however, ideology is more strongly related to preferences in these areas. Our thinking about trust in government follows the same arc. If people are thinking about the government in terms of keeping the country safe from world terrorism, as was apparently the case in the early 2000s, trust in government ought to affectpreferences connected to that task morestrongly than it affects preferences about race or redistribution.[6]
Theoretically speaking, the effect of trust on foreign and defense policy preferences, specifically,ought to be particularly strong when this domain becomes salient. Foreign affairs are especially remote from most citizens’ personal experience (Lippmann 1922), and most Americans possess little information about the world abroad (Delli Carpini and Keeter 1996). Lacking personal experience or significant information about foreign policy, preferences about it are plausibly more uncertain than for other domains (Hurwitz and Peffley 1987, Lavine et al 1996). Such uncertainty ought to make a commodity like political trust particularly important, given the role that trust can play in overcoming uncertainty (see Yamagishi, Cook, and Watabe 1998). For example, very few Americans would have the access necessary to assess conditions in Iraq and, in turn, whether the costs of military intervention there have been worth the benefits. We suspect that the more they trust the government, the more inclined they will be to express support for the government’s evaluation of conditions abroad along with its favored initiatives. Conversely, if people do not trust the government, then they ought to be less likely to adopt the government’s evaluation and its preferred policy course.
Finally, although perceptions ofwho sacrifices and who benefits play an important conditional role in the recent studies of political trust’s effects (Hetherington and Globetti 2002; Hetherington 2005; Rudolph and Evans 2005), this approach maps less well onto the foreign policy realm than for welfare state policies.All Americans benefit from the government protecting its citizens.That said, the effect of political trust still might be conditional regarding national defense.Davis and Silver (2004) find that the effect of trust on support for civil liberties after September 11 was contingent on the amount of threat people perceived from terrorism.Those who perceived less threat needed to trust the government to part with certain civil liberties.Among those who felt more threat, however, trust had no effect.Indeed, the logic is complementary to the logic regarding sacrifice in the racial policy and redistributive realm.If people are overwhelmed with worry that terrorism will strike again, they will not need to trust government to support government initiatives that are designed to combat it because alternatives to the government’s role are practically unavailable. This is akin to benefitting from and not sacrificing for a program.Trust would, however, remain important for those who feelless threatened.[7]
Data
Wefirst draw on the 2000-2004 National Election Study (NES 2004a) panel survey to demonstrate that the causal flow in our nexus of variables runs from political trust to foreign policy preferences rather than vice versa.Next, we turn to the 2004 NES cross-section (NES 2004b) to showthat political trust affected a wide range of foreign and defense policy preferences when foreign policy concerns became salient.Finally, weuse the NES’s Cumulative data file (NES 2004c) to demonstrate the relationship between issue salience and the effect of political trust on defense and racial policy preferences over time.
Foreign/Military Policy Preferences
The 2004 NES includes a range of items that allow us to test our hypotheses.In all, we employ five dependent variables.Two deal specifically with Iraq.Respondents were asked whether the war in Iraq was worth the cost and whether they thought the war reduced the threat of terrorism. In addition, respondents were asked whether they thought spending on the war on terror ought to be increased, decreased, or kept about the same and the degree to which they thought combating international terrorism was an important foreign policy priority.Finally, the NES asked people to place themselves on a seven point defense spending scale.[8]
The level of measurement of these dependent variables differs, requiring different estimation techniques.Assessments of whether the war in Iraqwas worth the cost are dichotomous, so we estimate this model using logistic regression.Preferences about spending on the war on terrorism, whether the war in Iraq reduced the threat of terrorism, and whether or not combating international terrorism ought to be a foreign policy priority are measured on three point ordinal scales.For these variables, we use ordered probit to estimate our models.Lastly, support for defense spending is measured on a seven point scale, which approximates an interval scale.Hence we use ordinary least squares to estimate this model.
Explaining Foreign Policy Preferences
Attitudes about foreign policy are a function of more than just political trust.Although the scholarly conventional wisdom once suggested that foreign policy attitudes were not firmly held and ostensibly arrived at random (see e.g. Almond 1950; Converse 1964; Gamson and Modigliani 1989), more recent research suggests that opinions on these matters are structured in predictable ways (Hurwitz and Peffley 1987) and are reasonably stable over time (Achen 1975).