1. Introduction
Securing the trust of constituents can facilitate the delivery of high-quality public services and government has a normative expectation to act in a trustworthy manner. Citizen trust in government, however, can be difficult to obtain and easy to lose (Kampen et al., 2006). The virtuous circle between a trusting citizenry and high-quality public service provision is fragile and can quickly be replaced with a vicious one, where a distrusting citizenry withholds voluntary compliance and public service provision suffers as a consequence (Van de Walle & Bouckaert, 2003). While there is an inherent relationship between service and policy performance and citizen trust, the relationship is not simple (Van de Walle & Bouckaert, 2003; Yang & Holzer, 2006), and nor can government performance be reduced to efficiency or effectiveness (Campbell, 2021; Jørgensen & Bozeman, 2007; Rainey & Steinbauer, 1999). Instead, public managers need to consider a range of values throughout the public service delivery process including equity, accountability, and transparency. Even if public service quality is high, if decisions are taken without accountability and according only to the logic of efficiency, trust in government may suffer. In terms of policymaking, even high-performance decisions can face legitimacy challenges if procedural expectations are disregarded, and the manner and conditions under which public policy is formulated can shape citizen evaluations of government performance regardless of outcomes.
To increase the legitimacy of decisions made about important policy or public service provision questions, public managers may include citizens in the decision process itself (Campbell & Im, 2016; Yang & Pandey, 2011). In other words, even though public participation in the policymaking process can improve policy performance, it can also contribute to policy legitimacy (Ahn & Campbell, 2022; D.-H. Kim & Campbell, 2014; Scharpf, 1999). Through involving the public meaningfully in decisions about policy and public service, public managers can use participation strategically to not only capitalize on the distributed knowledge of constituents, but also to shore up support for difficult policy decisions and ultimately increase trust in the policymaking process (Irvin & Stansbury, 2004). At the same time, meaningful public participation entails costs and carries risks for managers and bureaucrats. For instance, public managers may become beholden to poor decisions that reflect particular interests rather than the broader interests of the community, and moreover, the participatory process itself can consume scarce resources that could otherwise be used to support the mission of the organization (Irvin & Stansbury, 2004; Moynihan, 2003). As such, although public participation can benefit public managers it can also harm them and the decision to adopt a participatory mode of policymaking should be weighed carefully.
Although there is an intuitive link between participation and trust in government, the relationship can be tested empirically. Additionally, while the theoretical literature suggests that participation can increase the legitimacy of the policymaking process, it is not clear whether this additional legitimacy is itself a substantive enough benefit in all cases to justify the costs associated with public involvement. This study uses a two-by-two vignette-based experimental design to test the relationship between public participation in the policymaking process, policy performance, and trust in government. Below, I discuss the relationship between government performance and trust in government, arguing that an efficiency or effectiveness model alone is unlikely to be the sole source of trust. I then discuss the possible role that public participation plays in generating trust in government. After staking out my theoretical position, I describe the data and the experimental methodology I use to test the hypotheses. After presenting the results of the empirical analysis, I discuss their implications.
2. Literature Review and Hypotheses
Policy Performance, Public Participation, and Trust in Government
Identifying the foundations of citizen trust in government is a longstanding goal (Van de Walle & Bouckaert, 2003). While trust in government facilitates government functionality, in turn, the behavior of government - for instance, whether it efficiently and cleanly provides high quality public services, implements public policy effectively, or is seen as contributing to economic development - acts as an input to trust (Zhang et al., 2022). In this sense, trust and performance are intimately linked. Moreover, the conceptual composition of the trust concept itself suggests a link between performance and trust. Typically, trust is conceptualized as having a three-dimensional structure consisting of benevolence, integrity, and competence (Grimmelikhuijsen & Knies, 2017; Mayer et al., 1995). Whereas benevolence and integrity capture qualities related to the intentions and ethical orientation of the object of trust, competence relates to the perceived ability of that object to fulfill its promises. In this sense, performance (or the potential to perform) is a conceptual component of trust, and previous research has found a robust (if culturally conditioned) relationship between the two constructs (Mizrahi et al., 2010; Zhang et al., 2022).
I follow previous research and hypothesize a positive relationship between government performance and trust. While performance is relevant to a range of government activities, such as ongoing public service delivery, or the functioning of government in general, it can be difficult at these levels to both measure performance as well as connect individual actions to performance-relevant outcomes (Campbell, 2021). As such, I here conceptualize performance at the policy output level to clearly link the behavior of officials to performance. Here, while (perceived) performance – that is, the achievement of stated policy goals – is likely a necessary and important condition of trust, it is not likely to be the only relevant factor. Public sector performance is inherently multi-dimensional, complex, and linked with diverse and potentially competing values (Campbell, 2020; Jørgensen & Bozeman, 2007). Government actions throughout the policy process should be carried out in a manner consistent with a range of values including accountability, transparency, and ultimately, the rule of law, and thus the procedural dimension of decision-making in the public sector is a critical factor contributing to the legitimacy of outputs (cf. Van Ryzin, 2011). In other words, not only is policy output important for trust, the manner in which performance is achieved is relevant, and, particularly in democratic contexts, policymakers and government officials must behave in ways consistent with procedural norms if their behavior is to retain legitimacy, regardless of the performance achieved.
It is a democratic imperative that those affected by a policy decision have the opportunity to shape it, and public participation in the policymaking process is thus desirable from a normative perspective (Moynihan, 2003). The face-to-face nature of participation builds trust by breaking down barriers to communication, facilitates perceptions of government transparency, and tends to enhance trust when it increases the perception of ethical behavior and public service competence (Ansell & Gash, 2008; S. Kim & Lee, 2019; Wang & Van Wart, 2007). Participation communicates to citizens that decision-making is transparent, that government is open, and that citizens can meaningfully shape policy that affects their lives. Inviting citizens to participate in the policymaking process signals government’s willingness to share power. Other things being equal, it is expected that inviting citizens and other stakeholders to engage in policymaking in a meaningful way will increase citizen trust in government. I propose the following hypotheses.
Hypothesis 1: Policy performance is positively related to trust in government.
Hypothesis 2: Public participation in the policymaking process is positively related to trust in government.
Participation and Trust When Performance Outcomes Vary
Aside from the normative expectation of involving those whom the policy will impact in the policymaking process, participation can be motivated by a desire to improve decision-making (Irvin & Stansbury, 2004). However, involving the public in the policymaking process also has both costs and risks for administrators, which will undoubtedly be taken into account by those tasked with designing public participation initiatives (Campbell & Im, 2016; Moynihan, 2003). From an instrumental perspective, a key question for administrators that have the discretion to invite (or not) the public to a decision-making forum is whether the associated cost to benefit ratio is large enough to justify the risks.
Irvin and Stansbury (2004) suggest that one of the important benefits of public participation is diluting the blame that government decision makers would otherwise exclusively shoulder when a given decision goes wrong. Relatedly, James et al. (2016) looked at the relationship between the extent to which citizens blame local politicians under conditions of public service failure, finding that delegation to public managers reduces blame. This suggests that if a particular policy initiative results in failure and citizens were involved in the decision-making process, the damage to the public’s trust may be less than if the government acted unilaterally. That the decision was made in a more collective, consensus-oriented manner may communicate to citizens that managers recognized the risks the policy entailed and sought to include in the decision-making process stakeholders who were potentially affected by them. Participation may also communicate to citizens the complexity of the issues involved and the corresponding difficulty in producing effective decisions. In short, particularly when policy fails, involving the public in the decision-making process can highlight that steps were taken to come to the best decision possible while taking risks into account.
Participation may thus be particularly important for trust when policy fails. In contrast, when policy is successful, the legitimacy enhancing effect of public participation may be diminished. Although participation is likely to exert a positive influence on trust even when policy outcomes are good, the procedural legitimacy that participation bestows on decision-making may be less salient in these cases (cf. stealth democracy theory (Hibbing & Theiss-Morse, 2002)). Although effectiveness is not the only value against which government performance is judged, it may be the most important and thus when performance is good, the other procedural dimensions of decision-making may carry less weight. Consequently, I test whether the effect of participation on trust in government differs based on policy outcome, hypothesizing that the link is strongest when policy performs poorly.
Hypothesis 3: The positive relationship between public participation in the policymaking process and trust in government is stronger when policy performance is poor.
3. Data and Methods
Survey
The study uses a sample of university students drawn from a national university in South Korea. Using the university’s communications system, on February 22, 2018, a text message was sent to students in four departments asking them to participate in an online survey, a link to which was posted on the author’s personal website. This page provided information about the purpose of the survey, stated that participation was voluntary and responses anonymous, and provided a channel to contact the author with any concerns. Web analytics collected during this time suggest that about 182 individuals visited the site landing page, and a total of 86 responses were collected.
Although a representative sample of the target population is the gold standard, student samples have been widely used for experimental studies in the social sciences as well as public administration specifically (for instance, Tummers et al. (2016), Kaufmann & Feeney (2014), or Campbell (2020)). While theory confirmation is best done across multiple studies with various samples (Peterson & Merunka, 2014), particularly when testing theory, the use of student samples is not cause for the immediate rejection of a study’s results (Druckman & Kam, 2011; Mullinix et al., 2015; Trottier & Gordon, 2018). Moreover, experimental realism far outweighs mundane realism in securing the validity of an experiment (McDermott, 2002). While the use of a small student sample is of course a limitation of this study, such a limitation suggests the need for future research to validate the findings rather than to prematurely dismiss them.
The survey was organized into three parts. In the first part, respondents provided basic information (sex, major, and year of study) as well as reported their general attitudes about politics and propensity to trust. The second part of the survey consisted of one of four randomly distributed treatments in the form of short scenarios describing a fictional policy at the local government level, described in detail below. After reading one of the four scenarios, respondents reported their level of trust in the public officials described the vignette.
Experimental Design: The Vignette
The objective of this study was twofold. The first objective was to explore the link between citizen participation and trust in government and the second was to determine whether this link varied based on policy outcome. As such, a two-by-two factorial experimental design was adopted, with different treatments corresponding to one of four vignettes.
The experimental vignette approach has methodological and practical advantages. In addition to being flexible in their subject range and convenient to implement (Campbell, 2020), the approach also combines the experimental method’s gold standard level of internal validity with, when the vignette is carefully composed, high levels of experimental realism (Aguinis & Bradley, 2014). The approach has been used previously to study core public administration topics including red tape (Ahn & Campbell, 2022; Kaufmann & Feeney, 2014) and public participation (Kang & Van Ryzin, 2018; Migchelbrink & Van de Walle, 2020).
First, all respondents first read about a situation in which local government officials had decided to redevelop some “unused land” into “greenspace,” the full text of which read:
Access to greenspace is important for the wellbeing of local communities. Recognizing this, the government of LocalTown recently decided that some unused land in the city would be converted into a public park. The redevelopment was projected to be relatively inexpensive and was scheduled to be finished in 18 months.
Following this paragraph, respondents were exposed to one of four treatments, corresponding to whether citizens were involved in the development of the redevelopment plan or not, and secondly whether the redevelopment was successful or not. The participation treatment read as follows:
When creating the redevelopment plan, the government engaged in extensive consultation with citizens and civic groups who raised concerns and made suggestions about the plan. In response to this feedback, the plans were changed to incorporate the opinions of the citizens and civic groups. Then, the redevelopment project began.
Public participate can be shallow or extensive (King et al., 1998), and this paragraph describes both a genuine attempt to engage the public in the decision-making process as well as allowing them to have a meaningful level of influence over outcomes.
Good performance was described in the following text:
The results were good. The redevelopment project was completed on time and within budget. The redevelopment project is widely seen as a success.
Bad performance was described as such:
The results were not good. 3 years after its initiation, the redevelopment was still not finished, and the project had run significantly over budget. The redevelopment project is widely seen as a failure.
Following this, respondents answered a twelve-item inventory of questions about trust in government adopted from Grimmelikhuijsen & Knies (2017). The items measure perceived benevolence, integrity, and competence. I use the full twelve-item scale in the main analysis (Cronbach’s alpha equal to .920). I also use the individual scale components in a post hoc analysis to probe the results.
In the full model, I control for gender, study year, and political orientation (reported on a 3-point scale denoting conservative, centrist, or liberal). Rather than control for each of the 4 majors, I instead use a dummy variable to capture enrollment in the public administration program, given that these individuals may be the most sensitive to cues about participation (robustness checks indicated that controlling for individual majors did not affect the results). I also control for propensity to trust, which was measured by 2 items (alpha = .676) used by Yang (2005) asking respondents whether, “In everyday life, most people are honest” and “In everyday life, I always trust people, even those with whom I am not familiar.”
4. Results
Summary Statistics and Correlations
Table 1 shows descriptive statistics and bivariate correlations for the variables used in this study. The mean of trust in government is just below the scale midpoint, and about 43% of the sample received the good policy performance treatment and 43% received the public participation treatment. The sample leans liberal, with the mean value of the political orientation scale being about 2.26 on a scale of 3. The mean of propensity to trust is near the scale midpoint.
The good policy performance condition is positively correlated with trust in government at p < .001, which suggests, consistent with prior literature (for example, Mizrahi et al., 2010), that performance is an important factor driving trust. In contrast, the participation condition is not correlated with trust at statistically significant levels (p > .05). Means of all control variables did not differ at statistically significant levels by treatment condition, which suggests that randomization in the distribution of conditions was successful.
Figure 1 shows the median, lower, and upper quartiles of trust in government based on the 4 conditions. The graph is consistent with the study’s expectations: respondents exposed to the good policy performance with participation condition report the highest median level of trust. Those exposed to poor performance with no participation report the lowest. The two other conditions are between these values.
Table 2 shows the regression results. In model 1, trust in government is modeled as a function of policy performance. Model 2 introduces citizen participation. Both factors are statistically significant, and the addition of participation to the model increases the R-squared value from .233 to .281. These statistics are consistent with hypotheses 1 and 2. Model 3 introduces an interaction term to test whether the effect of participation on trust is conditional on policy performance. The coefficient is not statistically significant, which suggest that the two effects are independent of one another. Finally, model 4 introduces the control variables. While none of the controls are statistically significant, the positive, statistically significant relationships between the 2 treatment variables and trust in government persist.
Table 3 shows regressions with the individual treatments specified as dummy variables. In model 5, the reference group is the good policy performance without citizen participation group. The results suggest that involving the public in the policymaking process may have a positive effect on trust when policy outcomes are good (p < .10). However, poor performance with participation is also negatively related to trust compared to the reference group (p < .05) (as is, of course, poor performance without participation). These statistics suggest that participation is not enough to overcome the trust-destroying effect of poor performance.
Model 6 tells a similar story. Here, the reference group is poor performance without participation. Obviously, both of the good performance conditions (with and without participation) lead to higher comparative levels of trust (p < .001). However, participation, again, cannot substitute for poor performance, as there is no statistically significant difference in terms of trust between the with and without participation treatments when performance is poor. I note that the models include control variables.
A final post hoc set of regressions was performed on the individual dimensions of trust (results available from author). The results of these 3 regressions were similar to those in table 2, model 4. However, adjusted R-squared values were not equivalent, with an adjusted R-squared value of .35 for competence, .19 for benevolence, and a low .04 for integrity (participation was not significant in this model). These statistics are not surprising given the conceptual overlap between performance and competence. However, they provide further evidence of the dominant role of performance relative to participation in determining trust in government.
5. Discussion
Although involving the public to the greatest extent possible in the decision-making processes of government is a democratic imperative, participation nevertheless has nontrivial costs (Campbell & Im, 2016; Moynihan, 2003). This study demonstrates that both policy performance and public participation in the policymaking process have a positive impact on trust in government. In the case of the latter, however, the results do not support the proposition that participation can make up for poor performance - a finding that has both theoretical and practical significance. Although the relationship between these constructs has been studied variously in the public administration literature, this study makes a number of distinct contributions. In the space below, I discuss the results of the study, its contributions, as well as the unanswered questions it leaves for future research.
Before turning to these, I will first discuss the limitations of the study, as acknowledging and understanding these can help put the results in context. First, the study uses a relatively small sample of students, and consequently one may question whether the results will generalize universally. Student samples are common in the experimental public administration literature (examples include Campbell, 2020; Esteve et al., 2015; Kaufmann & Feeney, 2014; Tummers et al., 2016, and many others) and findings uncovered using student samples often generalize unproblematically to the wider population (Druckman & Kam, 2011; Mullinix et al., 2015; Trottier & Gordon, 2018). Moreover, the legitimacy of the experimental design derives from its internal validity (that is, the isolation of the causal effect) rather than its claim to generalizability (Aguinis & Bradley, 2014). This being said, more research is necessary to confirm the universal applicability of the results. Second, I also note that the particular scenario described in the experimental vignette represents a unique situation which may have idiosyncratic characteristics relevant to the findings. Some findings based on survey experiments in the public administration literature have failed replication tests when subtle differences are introduced into the fictional situation (for instance, see Riccucci, Van Ryzin, and Li (2016) versus Sievert (2021)), and it may be the case that participation is more or less important based on policy area. To establish the veracity of the effects and potentially detect limit conditions at which they fail to hold, the findings should be replicated while systematically varying the policy domain.
These limitations noted, the first contribution of the study is the establishment of a causal link between government performance and trust in government. Although others have linked trust and performance, and there is undeniably an intuitive relationship between the two constructs, nevertheless, specifying the relationship between performance and trust in a concrete empirical format such that a causal relationship between them is laid bare presents challenges. Importantly, in a real policy and public service context, it is probable that trust and performance influence one another in a virtuous cycle (Van de Walle & Bouckaert, 2003). Consequently, efforts to establish the performance-trust relationship using correlational data - even by leveraging the strengths of panel data - cannot remove all doubt about the validity of the findings. In contrast, the experimental method, which I use here, is particularly suited to address this question, as it isolates performance at the level of the vignette rather than relying on subjective perceptions, and thereby provides strong evidence of a causal effect. Moreover, the vignette itself was carefully constructed to describe performance in concrete terms as well as in a manner the general public is likely to understand. Performance in the public sector is complex due to the multitude of (sometimes competing) values by which it can legitimately be measured (Campbell, 2021; Jørgensen & Bozeman, 2007). Citizens, however, may more readily conceptualize performance in simple private sector terms (Campbell, 2020), and by describing good performance as on schedule and within budget project completion, the type of performance described in the vignette is likely a valid reflection of these. This being said, the complex nature of performance in the public sector should not be overlooked. Future research, therefore, may further probe the relationship between performance and trust by specifying performance in different ways.
In addition to convincingly establishing a causal relationship between performance and trust, this study likewise examined the relationship between participation and trust. Like performance, the relationship between participation and trust has been studied in previous literature (Wang & Van Wart, 2007). Again, however, the experimental method facilitates estimating the effect of participation causally. In sum, this study provides tentative evidence for a causal link between public participation and trust in government. At the bivariate level, no statistically significant correlation was found between participation and trust. After controlling for performance, however, the independent effect was positive and statistically significant (and maintained this significance across a range of models), albeit with an R-squared value increase of about .04 over the base performance-only model. A number of remarks can be made about the manner in which participation was specified in the study.
First, participation was described in a robust way such that those who were involved in the policymaking process had a non-trivial impact on the shape of the final decisions. Consequently, it is unlikely that the weak effect of participation on trust is due to participation itself being implemented in a superficial manner. On the other hand, from the point of view of the respondent, participation was third person, that is, the respondent themselves did not directly participate in the decision-making process. This is not an illegitimate approach, particularly given that only a handful of citizens generally participate directly in open decision-making forums such as the one described here. In other words, in real life, most citizens would be more likely to read about participation second hand in precisely the way it was described in the vignette. This methodological approach is thus likely to have obtained a significant level of experimental realism. Although more difficult from a practical perspective (as well as potentially problematic from an ethical one), it would be possible to test the relationship between participation and trust using a field experiment where participants had (or were denied) a real opportunity to shape policy. Given that correlational research is suggestive of a link (S. Kim & Lee, 2019), it is not improbable that such an approach would produce a stronger effect. Additionally, I also note that the vignette describes a situation in which local government officials are the focal group. I chose to focus on local government because this level of government is generally “closer” to citizens in that most public services are delivered at the local level and public participation initiatives are likewise often implemented by local government officials. While it is not immediately clear how the use of local government as the venue for participation may have affected the results, future research may nevertheless look into this question.
Second, although participation does appear to provide a marginal boost to trust in government, its impact was not sufficient to overcome poor performance. This finding is consistent with some existing research that has found that performance is more important than participation for trust (Mizrahi et al., 2010). More research is necessary, however, should this finding be replicated and prove to be robust, it implies that, purely from the perspective of trust, participation may not be worth the trouble. Stated pessimistically, while participation has speculative benefits, it has inevitable costs (Irvin & Stansbury, 2004; Moynihan, 2003). The finding of the study, therefore, suggests that participation should be adopted cautiously (and specifically when there is a high probability of policy success) when facilitating trust in government is the goal. Of course, trust in government is not the only consideration for policymakers who have the discretion to involve the public in the decision-making process. And, decisions about the appropriateness of various modes of policymaking cannot be decided by a cost-benefit analysis alone. A research program systematically testing the relationship between public participation and a range of citizen-outcomes of value would therefore be a contribution to the literature.
Finally, I note that the context of this study, South Korea, may not be irrelevant to the results. As is well known, South Korea underwent a process of rapid economic development over the latter half of the 20th century (Im et al., 2013). However, less remarked upon but equally if not more impressive is the democratic transformation of the country’s politics over the past 30 years. Whereas policymaking throughout the rapid development period was a highly centralized affair, increasingly, the government has sought (and citizens have demanded) greater citizen involvement in the public decision-making process (Im et al., 2014; S. Kim, 2010). Local governments, however, generally rely on the central government for budgetary resources, and thus, though formal local autonomy was established with local elections in 1995, central government retains significant power to shape the policy goals and development plans of local actors (Cho et al., 2010; Im & Campbell, 2020). In South Korea, therefore, administrative decentralization is an ongoing process, which may inform views about the desirability and efficacy of participation in local policymaking.
Funding
This work was supported by an Incheon National University Research Grant in 2021.