Abstract
Despite a great deal of research that examines consequences of transparency policies, there have been few empirical assessments of the relationship between transparency and citizens’ perceptions of public sector performance. This study focuses upon the relationship between computer-mediated transparency and perceptions of public sector performance in particular. We hypothesize that citizens’ increased exposure to computer-mediated transparency will be positively related to their generalized perceptions of public sector performance. We also hypothesize that this positive relationship is mediated by citizens’ satisfaction with public service provision. Results suggest that increased exposure to computer-mediated transparency is positively associated with citizens’ perceptions of public sector performance, but negatively related to citizens’ satisfaction with public service provision. Moreover, the positive relationship between computer-mediated transparency and perceptions of public sector performance increases when controlling for public service satisfaction, indicating the presence of inconsistent mediation.
Keywords
Introduction
Poor perceptions of public sector performance are often attributed to the presence of some form of information asymmetry between citizens and their government (Kelly, 2002). By correcting these information asymmetries, via means such as transparency policies and open government initiatives, some have reasoned that citizens’ perceptions of government performance can be improved (Im, Porumbescu, & Lee, 2013; Yang & Holzer, 2006). This logic has underpinned countless information and communications technology (ICT)–oriented administrative reforms in recent years, which have attempted to enhance the transparency of the public sector and subsequently increase citizens’ access to information regarding various facets of their government’s performance (Worthy, 2010). Yet, while there is a wealth of research that attempts to understand how citizens’ exposure to government information through their use of new technologies impacts their evaluations of specific areas of government (e.g., Grimmelikhuijsen, 2012; H. Park & Blenkinsopp, 2011; Tolbert & Mossberger, 2006; H. Welch, Hinnant, & Moon, 2005), there have been very few empirical assessments of how this computer-mediated transparency reform trend actually relates to citizens’ perceptions of public sector performance in general. To this end, the central research question addressed in this study is, “To what extent is computer-mediated transparency related to citizens’ general perceptions of public sector performance?”
This research question is examined within the context of Seoul, South Korea. Seoul is particularly relevant for the purposes of this research for the reason that it is consistently considered to be a world leader in terms of its adoption and applications of e-government websites (S. Park, Choi, & Bok, 2013).
Information and Perceptions of Public Sector Performance
Perceptions of Public Sector Performance
Perceptions of public sector performance can be divided into three levels: micro, meso, and macro (Bouckaert & Halligan, 2011). Perceptions of performance at the micro-level of the public sector can be thought of as citizens’ perceptions of the performance of individuals responsible for the delivery of public services, such as “street-level bureaucrats,” policemen, or teachers, for example. Perceptions of performance at the meso-level of the public sector refer to perceptions of particular public services, or the institutions responsible for the provision of a particular service, such as the Internal Revenue Service, in the United States. At the micro- and meso-levels of the public sector, perceptions may differ substantially across services and areas of government (Kampen, De Walle, & Bouckaert, 2006). Therefore, perceptions of performance at these two levels can be considered more particularized (Christensen & Lægreid, 2005; Easton, 1975). Perceptions of public sector performance at the third “macro” level reflect a generalized assessment of how the public sector, in its entirety, is performing. Bouckaert et al. (2002) suggest that perceptions of performance at this level of government relate to factors such as unemployment, security, health care and citizens’ overall happiness—factors that stem from the concerted effort of a variety of actors and institutions in the public sector. As such, perceptions of performance at the macro-level are not so much a function of citizens perceiving their government “doing things well,” but instead of citizens perceiving their government as “doing good things” (Bouckaert et al., 2002, p. 48). Put differently, perceptions of performance at this macro-level of the public sector are closely tied to citizens’ perception that the public sector is performing in a way that matches their best interests.
Citizens’ perceptions of the micro-, meso-, and macro-levels of government are not entirely independent of each other. For example, positive perceptions of the performance of an individual associated with the delivery of a specific public service (micro-level) can, in turn, positively affect perceptions of said public service or the institution responsible for the provision of said service (Heintzman & Marson, 2005; Im & Lee, 2012). Elsewhere in the literature, this is sometimes referred to as a halo effect (Chong & Wong, 2005). Perceptions of performance at the meso-level are also strongly related to perceptions of performance at the macro-level of government. For example, several empirical studies have uncovered a strong association between satisfaction with specific public services and levels of trust in government in general (Christensen & Lægreid, 2005; Kampen et al., 2006; Porumbescu & Park, 2014; Van Ryzin, 2007). These findings can be interpreted to suggest that higher levels of satisfaction with public services (meso-level) facilitate citizens’ ability to perceive how the actions of their government align with the public’s best interests (macro-level; Porumbescu, 2013).
This study is primarily concerned with gaining insight into the relationship between citizens’ exposure to computer-mediated transparency and their perceptions of performance at the meso- and macro-levels of government. We focus upon citizens’ perceptions of performance at the meso- and macro-levels of government for three reasons. First, e-government websites generally focus upon disclosing information pertaining to the performance of a particular institutions or public service (Grimmelikhuijsen & Welch, 2012). As such, computer-mediated transparency is likely to be a factor that is relevant to understanding citizens’ perceptions of public services and institutions at the meso-level of government. Second, much of the discourse surrounding transparency suggests the potential to not only transform citizens’ perceptions of specific institutions and services, but to also foster more positive attitudes toward government in general (Cucciniello & Nasi, 2014). Nevertheless, there has been very little empirical assessment of how enhanced transparency relates to citizens’ perceptions of performance at the macro-level of government (Porumbescu, 2013). As such, by incorporating citizens’ perceptions of performance at the macro-level of government into our assessment, this study takes an initial step toward filling an important dearth in the extant transparency literature. Third, by incorporating perceptions of performance at the meso- and macro-levels of government into our analysis, we can provide a more nuanced understanding of how computer-mediated transparency relates to citizens’ perceptions of the public sector.
Defining Transparency
The topic of transparency has attracted a great deal of attention from academics and practitioners. This attention has, in turn, led to variety in terms of definitions of transparency. Despite the diversity, Grimmelikhuijsen and colleagues (2013, p. 4) observe that most definitions of transparency tend to emphasize “the availability of information about the internal workings or performance of an organization,” which in turn enhances the “inward observability” of an organization by external actors, such as citizens. Following such logic, the extent to which the public sector is transparent, in principle, speaks to citizens’ capacity to access information about the activities of their public sector (Etzioni, 2010).
Grimmelikhuijsen (2011, 2012) sets out three dimensions that can be used to clarify the object of the information provided to citizens regarding activities of their public sector, which are decision-making transparency, policy transparency, and policy outcome transparency. Decision-making transparency refers to the extent citizens are provided with information that explains the steps taken to arrive at a certain decision, as well as the reasons why one course of action was adopted as opposed to an alternative. Policy transparency refers to information outlining what the policy actually is, what issues it intends to address, and the implications of this policy for citizens and others who stand to be affected by the policy. Policy outcome transparency refers to the timely release of data or summaries of data that discuss results of particular policies.
Historically, government publicly transmitted information regarding decision making, public policies, and policy outcomes directly to external stakeholders (e.g., citizens), via face-to-face interactions at venues such as town-hall meetings (Meijer, 2009). However, as the scale of society has grown, indirect or “mediated” forms of transparency have emerged (Hood, 2007). Due to the rapid sophistication and diffusion of ICT over the course of the past decades, one of the most prominent forms of mediated transparency today is “computer-mediated transparency” (Grimmelikhuijsen et al., 2013, p. 575). Meijer (2009, p. 259) defines computer-mediated transparency as the use of computerized technologies to foster greater “inward observability” of government organizations. To this end, e-government websites and portals constitute the “backbone” of computer-mediated transparency (Grimmelikhuijsen et al., 2013). This article focuses in particular upon computer-mediated transparency.
Meijer (2009) outlines three key differences that exist between computer-mediated transparency and direct, face-to-face transparency. 1 The first difference is that computer-mediated transparency is, to a large extent, one directional. Whereas direct, face-to-face exchanges of information provide the opportunity for two-way flows of information between citizens and their government, computer-mediated transparency is largely conducive to one-way flows of information. As Meijer points out, “Citizens can watch councils, but councils have no idea whether citizens are watching them” (p. 259). Second, when compared to direct, face-to-face transparency, computer-mediated transparency presents information in a decontextualized fashion. This is to say that the information being made public by government via channels such as e-government websites take “transparency away from the context that it refers to” (p. 260). In turn, this can present challenges with respect to interpreting information that is accessed. For example, Meijer argues that while performance information pertaining to schools from a certain district may be readily accessible to the public, this information will likely be difficult to accurately interpret without an understanding of the particular school district the information was drawn from. Third, computer-mediated transparency is conducive to the transmission of information that is more structured and better organized when compared with information obtained from face-to-face interactions with government. This is because computers are designed to make large volumes of information readily accessible to a particular audience. To do this effectively and efficiently, computers organize and communicate information in a highly structured format, particularly when compared with the face-to-face exchange of information (cf. Meijer & Thaens, 2013, p. 344).
Computer-Mediated Transparency and Citizens’ Perceptions of Performance
An important determinant of citizens’ perceptions of government performance at the macro-level is their use of different information sources (Nye, 1997). This notion is central to many ICT-based transparency initiatives that have been launched in recent years, reasoning that citizens are not well aware of what their government is doing and how the actions of their public sector, in reality, often serve to advance their best interests (Porumbescu & Im, 2015). In the absence of such transparency initiatives, citizens must rely on private sources of media for information about their government, which are often said to engage in sensationalist and negatively biased coverage (Gordon, 2000). Therefore, enhancing transparency through the use of computers can provide citizens with accurate information outlining why the government chose the course of action it did, and how that particular course of action is envisioned to advance citizens’ best interests (Im et al., 2013). Moreover, enhanced public disclosure of information may also signal to citizens that the public sector has no secrets to hide. Taken together, these points suggest that computer-mediated transparency affords the public sector an effective means of communicating ways it is working to advance citizens’ best interests:
Yet, drawing upon the discussion in the “Defining Transparency” section, some may argue that the relationship between computer-mediated transparency and perceptions of performance at the macro-level of government is not direct, but rather mediated by citizens’ satisfaction with public service provision (cf. Christensen & Lægreid, 2005; Easton, 1975; Van de Walle & Bouckaert, 2003), which relates to perceptions of the meso-level of the public sector (Bouckaert et al., 2002, Grimmelikhuijsen, 2012).
Citizen satisfaction is explained as the difference between citizens’ perceptions of public service provision and their previous expectations of public service provision (Im & Lee, 2012; James, 2009; Kampen et al., 2006; Van Ryzin, 2004). Higher levels of satisfaction result from citizens’ perceptions of public services exceeding their expectations. Citizens’ perceptions of public services may stem from actual experiences with a particular public service, but as Van Ryzin (2007, p. 528) notes, citizens’ perceptions of public services often do not the result from direct experience, but are rather derived from factors such as the media. As such, low levels of citizen satisfaction with public services may result from citizens’ inability to accurately perceive how different public services are performing. As Janet Kelly (2003) has written on the disparity between actual and perceived performance, “if citizens’ only understood how well their government was performing they would (or should) be satisfied” (p. 858). Others have suggested that, due to a lack of accurate information pertaining to their public sector, citizens’ expectations of public service provision may be unrealistically high, thereby contributing to low levels of citizen satisfaction with public services (Bouckaert & Van de Walle, 2003; Pollitt & Bouckaert, 2004). In this scenario, citizens overestimate the capacity and/or resources their public sector possesses to provide public services and therefore perceive public services as falling short of their potential.
Computer-mediated transparency initiatives may enhance citizens’ satisfaction with public service provision due to their potential to improve perceptions of public service provision and to create more realistic (or lower) expectations of the public sector’s capacity to deliver public services. Ho and Coates (2004) reason that from greater exposure to government websites providing information about the performance of public services, citizens will have better access to objective information related to public policies and actual performance levels of public services. In turn, this information can be used to counter what some have suggested is a negatively biased portrayal of the public sector by third parties such as the mass media, and increasingly, private online media (Gordon, 2000; Im et al., 2013). Moreover, exposure to objective information regarding public sector budgets and various other constraints the public sector faces in delivering public services allows citizens to become better aware of why the public sector is performing the way it is, thereby engendering more realistic expectations of the public sector’s capacity to deliver services.
Therefore, computer-mediated transparency stands to enhance citizens’ satisfaction by fostering more realistic expectations of public service provision, while improving citizens’ perceptions of public service provision. In addition, drawing upon the interrelated nature of perceptions at the meso- and macro-levels of the public sector discussed in the “Defining Transparency” section, computer-mediated transparency’s effect upon satisfaction will, in turn, likely carry over to perceptions of macro-level performance, via an effect upon satisfaction:
Despite the potential for a positive relationship between transparency and citizens’ perceptions of government performance, a growing vein of research has begun to challenge such arguments. Grimmelikhuijsen (2012), synthesizing much of this work, provides three concise criticisms of the possible link between transparency and improved perceptions of government performance. First, the massive influx of information attributed to transparency policies can result in “information overload” (p. 37), and consequently create confusion among recipients (O’Neil, 2002). Second, in gaining access to information on the inner workings of the public sector, citizens may become disenchanted, viewing their public administration as less competent than they would have otherwise (Grimmelikhuijsen et al., 2013). Third, computer-mediated transparency may lead to citizens placing more blame upon the public sector. As Grimmelikhuijsen (2012) argues, “A fault of the government can always be construed, and if citizens, media and politicians use transparency for their own gain with no restraints, this could result in the politics of scandal” (p. 37).
Data, Method, and Measurements
Data
This study uses data from a 2012 online survey of Seoul citizens. The survey was delivered online, as well as via face-to-face structured interviews. The online survey was distributed by an online survey research firm, Macromill Embrain, to those in age groups younger than 60 years. Respondents for the online survey were recruited from a preexisting panel of 21,419 potential respondents maintained by Macromill Embrain, which, in aggregate, matches the population of Seoul on key parameters of age (for age groups younger than 60 years), location, income, and gender. 2 To ensure representativeness of older segments of the population (age groups older than 60 years of age), who are considered less likely to have Internet access, interviewers visited homes of respondents and conducted face-to-face structured interviews. Surveys were distributed to this segment of the population by a second survey research firm, Research Lab.
For the purposes of this study, a target sample of 1,100 respondents was identified. This number constituted roughly 0.01% of the population of Seoul at the time the study was conducted. The composition of the target sample was selected to reflect the population of Seoul on parameters of age, income, location (administrative district), and gender. To recruit respondents under the age of 60, the online survey firm, Macromill Embrain, emailed its panel of 21,419 Seoul citizens, inviting them to fill out the survey used in this study. In response to the email, 4,163 unique online surveys were returned. After discarding surveys that were incomplete; ensuring the completed surveys matched the characteristics of the target sample in terms of age (for age groups under 60), location, income, and gender; and discarding surveys that were filled out in less than 5% of the allocated time, 898 online surveys remained. 3 To recruit respondents who were over the age of 60, the research firm, Research Lab, arranged to deliver the survey face-to-face via structured interviews to respondents who matched the target sample in terms of age (for age groups above 60), location, income, and gender. The total number of surveys delivered via face-to-face structured interviews to those in the target sample over age 60 was 202. Taken together, the 202 surveys delivered via face-to-face interviews and the 898 online surveys combined for a total of 1,100 usable surveys that matched the target sample on all parameters. Finally, all of the completed surveys were then examined for response patterns. 4 Those exhibiting response patterns were discarded as their presence suggested that respondents did not read the items before selecting responses. This reduced the total number of surveys used to 907. No systematic bias was found among the surveys that were discarded, and the remaining 907 surveys did not deviate significantly from the parameters of the target sample.
Context of Seoul
Seoul is an interesting case to assess relationships between computer-mediated transparency and perceptions of public sector performance. First, South Korean citizens, whose experiences with democracy are relatively short, and citizens of Seoul in particular, have been argued to hold increasingly critical attitudes toward their government (C. M. Park & Shin, 2004). Kim (2010) has argued that this trend suggests the emergence of “critical citizens” in South Korea. Critical citizens are a common feature of developed democracies and generally support the basic tenets of democracy, yet at the same time are less deferential to the authority of government and critical of the way their political system and public sector institutions operate (Norris, 1999). However, the emergence of critical citizens in South Korea (and Seoul), which up until 1987 was led by an oppressive military regime, does not necessarily imply that central, regional, or local governments in South Korea are faced with increasingly negative citizens, but rather that citizens’ criteria for assessing the performance of their government are shifting (Inglehart, 1997; Norris, 2011). To this end, research by Kim (2010, p. 807), attempting to elucidate the relationship between the rise of critical citizens in South Korea and the implications for trust in government, found that an important factor positively influencing citizens’ attitudes toward their government was citizens’ perception that the government has transparent and open practices.
Models
The hypotheses examined in this study can be illustrated using the path diagrams found in Figures 1a and 1b. Figure 1a illustrates the total effect of computer-mediated transparency on perceptions of public sector performance (Path c). In keeping with Hypothesis 1, this figure suggests that greater exposure to computer-mediated transparency is positively associated with perceptions of public sector performance. However, previous literature indicates (i.e., Bouckaert et al., 2002; Van de Walle, 2004) that the relationship between computer-mediated transparency and perceptions of public sector performance is better understood through the incorporation of a third mediating variable, citizen satisfaction with public services. Through the inclusion of this mediating variable, we are able to obtain a more nuanced understanding of how computer-mediated transparency relates to citizens’ perceptions of public performance by disaggregating the total effect (Path c) into two distinct constitutive effects—direct effects (Path c’) and indirect effects (Paths a and b). These effects are illustrated in Figure 1b. First, as Figure 1b depicts, at least a portion of computer-mediated transparency’s effect upon perceptions of public sector performance will occur indirectly—computer-mediated transparency will improve satisfaction with public services (Path a), and, in turn, levels of satisfaction with public services will improve perceptions of public sector performance (Path b). This indirect relationship is addressed by Hypotheses 2 and 3. The balance of computer-mediated transparency’s total effect upon perceptions of public sector performance is likely to occur via a direct relationship between these two variables, as illustrated by Path c’ in Figure 1b.
Path Diagram of Relationships.
To examine the relationships explained above, this study makes use of Baron and Kenny’s (1986) three-equation approach to modeling mediating effects. For mediation to be present, Baron and Kenny outline four criteria: (a) the independent variable must possess a significant relationship with the dependent variable in the absence of the mediator, (b) the independent variable must possess a significant relationship with the mediating variable, (c) the mediating variable possesses a significant relationship with the dependent variable, and (d) the strength of the relationship between the independent variable and dependent variable shrinks, when controlling for the mediator.
Measurement of Main Variables 5
The dependent variable: Perceptions of public sector performance at the macro-level
Perceptions of public sector performance at the macro-level consist of four dimensions related to security, health, wealth, and overall happiness (Bouckaert et al., 2002; Porumbescu, 2013; Van de Walle, 2004).
Citizens’ perceptions of performance at the macro-level of government were measured using an average of six items. Each of the six items addressed one of the four dimensions of macro public sector performance and asked respondents to rate performance of the Seoul Metropolitan public sector on a 5-point Likert-type scale, where 1 corresponded to extremely poor and 5 corresponded to excellent. Mean values of the six items were then computed and used as an aggregated measure of perceptions of macro-level performance. The Cronbach’s alpha for these six items was .902. The items that were used to measure the perceptions of macro-performance can be found in the appendix.
The independent variable: Exposure to computer-mediated transparency
We measure exposure to computer-mediated transparency according to an average of citizens’ frequency of use of two particular online government sources. The first is the city of Seoul web portal, which provides information pertaining to the delivery of specific public services, as well as policy proposals for new services and goods. The second is a weblog (blog) webpage that is operated by the Seoul Metropolitan government to share information about different aspects of public services and events in the city with citizens. These sites do not directly deliver specific public services, and are generally not used by citizens to directly interact with government. Respondents were asked, “To what extent do you make use of the online government sources listed to obtain information about services provided by the Seoul Metropolitan government.” Respondents were asked to rate their use on a scale of 0 (do not use at all) to 6 (several times a day). Responses were averaged to form an aggregate measure of exposure to computer-mediated transparency. This measure is similar to those used by Tolbert and Mossberger (2006) and Welch et al. (2005) to measure e-government use. These items can be found in the appendix.
The mediating variable: Citizens’ satisfaction with public service provision
An expectation disconfirmation approach was taken to measure citizens’ satisfaction with different public services provided by the Seoul Metropolitan government (cf. van Ryzin, 2007). Survey respondents were asked, “Compared to two years ago, to what extent have the following public services fallen short of or exceeded your expectations.” Respondents were then able to rate 18 specific public services that ranged from public transportation to city cleanliness, on a scale of 1 (significantly below their expectations and very dissatisfied) to 7 (significantly above their expectations and very satisfied). 6 The measure of satisfaction used in this study is an average of assessments of all 19 specific services.
Control variables
To glean insight into the unique effects of transparency on satisfaction with public service provision and perceptions of public sector performance, several control variables are introduced. In addition to standard demographic control variables, of income, age, gender, and marital status, further control variables were selected according to a review of research that has examined factors affecting citizens’ perceptions of public sector performance (see Table 1). Political factors, such as ideological leanings or attitudes toward elected officials have been argued to shape preferences for public services and policies, thereby affecting evaluations of public service provision and perceptions of the public sector (Christensen & Lægreid, 2005; Porumbescu, 2013). To control for this influence, citizens’ satisfaction with the mayor of the Seoul Metropolitan government was included into our analysis. Satisfaction with the mayor of Seoul was measured using a 5-point Likert-type scale (1 = strongly dissatisfied, 5 = extremely satisfied). A dummy variable assessing whether citizens belonged to a progressive political party was also included as a control variable. To operationalize “member of a progressive party,” respondents were asked which political party they belonged to. Those belonging to parties commonly viewed as progressive were aggregated into a single “progressive” group. At the time the survey was conducted, the mayor of Seoul was a member of a progressive political party. Social trust was included to account for a basic generalized predisposition toward government that a great deal of research suggests forms the basis for citizens’ evaluations of their government (Kampen et al., 2006; Van de Walle, 2004). To account for this disposition, a modified version of the Rosenberg scale was used, which consists of five items that are measured using a 5-point Likert-type scale (1 = strongly disagree, 5 = strongly agree). Perceptions of the public sector may be framed by attitudes toward the citizens of Seoul in general (Im et al., 2013). To account for this, we include a measure of citizens’ “relationship with the city,” which asks participants to respond to a single item, “Generally, Seoul is a pleasant city to live in.” Responses were measured using a 3-point Likert-type scale (1 = strongly disagree, 3 = strongly agree). An additional factor, trust in government, is commonly found to be highly associated with perceptions of public sector performance and satisfaction with public service provision (Bouckaert & Van de Walle, 2003; Christensen & Lægreid, 2005; Kampen et al., 2006; Morgeson & Petrescu, 2011). To account for this association, we include a 13-item measure of perceived public sector trustworthiness that has been validated in previous research (Grimmelikhuijsen, 2011, Grimmelikhuijsen et al., 2013; Porumbescu, 2013). Responses are assessed with a 5-point Likert-type scale. The location citizens’ live in also affects levels of satisfaction with public service provision, as wealthier districts may receive higher quality public services than less wealthy areas (Jilke & Van de Walle, 2013). To account for this, a single item measuring citizens’ satisfaction with their administrative district is included. This item is measured according to a 5-point Likert-type scale (1 = not satisfied, 5 = highly satisfied). Two items are included to control for the potential confounding influence of non-computer-mediated transparency sources of information. The first item we use is an average measure of citizens’ use of traditional, offline media. Here, citizens’ frequency of use of radio, television, and newspaper is assessed using a 7-point Likert-type scale (1 = never use, 6 = use multiple times a day). The second item measures how frequently citizens call government to request information. A 7-point Likert-type scale is used. Finally, because surveys were administered to respondents over the age of 60 face-to-face, a dummy variable was included to control for effects that this method of delivery had when compared with the online survey administered to those under the age of 60.
Correlation Table. a
Note. N = 907, for all variables included in analyses.
An exploratory factor analysis was conducted using promax rotation to assess the divergent validity of the items used to measure trust in government and perceptions of public sector performance. The items used to measure perceptions of performance loaded onto a single distinct factor: Factor loadings ranged from a low of 0.70 to a high of 0.89. Items used to measure trust in government loaded onto two factors distinct from that of perceptions of performance, which is in line with research on the topic of trust in government that suggests that citizens’ perceptions of this construct are multi-faceted (Grimmelikhuijsen & Meijer, 2013, Porumbescu, 2013).
p < .1. **p < .05. ***p < .01.
Findings and Discussion
Descriptive statistics are presented in Table 2. The mean for citizens’ perceptions of public sector performance at the macro-level was 2.81. This suggests that citizens are slightly positive in their perception that the public sector’s performance is contributing toward a better society. However, the mean value of 4.14 for citizens’ satisfaction with public service provision is very high. This suggests that many of the survey respondents were very satisfied with the public services provided by the Seoul Metropolitan government. It is also interesting to note that the standard deviation of citizens’ satisfaction is much higher than citizens’ perceptions of government performance at the macro-level, implying that there are more individual differences in terms of citizen’s satisfaction when compared with their perceptions of public sector performance at the macro-level. Finally, the mean value of 2.96 for citizens’ frequency of exposure to computer-mediated transparency suggests that Seoul citizens, on average, access city websites for information about their government fairly often.
Regression Results.
Note. All estimates were produced using OLS. OLS = ordinary least squares.
p < .1. **p < .05. ***p < .01.
Table 2 shows results of the three regression models that are used for the mediation analysis. 7 All variables included for analysis are standardized. In all three models, F values are highly significant (p < .001), and the adjusted r2 values range from .105 to .676. These values suggest that the models used in this study are significant and explain variation in the data well.
The first model, Equation 1, suggests that citizens’ increased exposure to computer-mediated transparency is positively associated with perceptions of government performance at the macro-level (Path c: total effects); the mediating variable—satisfaction with public service provision—is not present in this regression. This result provides support for the argument that increased use of information afforded to citizens via computer-mediated transparency is positively related to their perception that the public sector is generally doing the right things. However, it is interesting to note the low level of significance. This will be discussed at greater length in the following section. Referring to the criteria for mediation, the first criterion, a significant relationship (p < .1) between the independent variable and dependent variable, is satisfied.
The second model, Equation 2, shows that citizens’ satisfaction with public service provision is negatively associated with exposure to computer-mediated transparency (Path a). This suggests that while citizens’ perceptions of public sector performance at the macro-level may be positively associated with transparency, their satisfaction with public sector service provision, which pertains to the meso-level of the public sector, is negatively associated with transparency. The significant relationship (p < .1) found in Equation 2 satisfies the second criterion for mediation.
Models 1 and 2 offer evidence to suggest that the consequences of transparency may vary according to the level of government they are assessed, with transparency having a positive relationship with citizens’ perceptions of public sector performance at the macro-level of government, and a negative relationship with citizens’ attitudes toward public service provision at the meso-level of the public sector. This is discussed at greater length in the “Discussion and Conclusion” section.
Model 3 reveals two findings of interest. First, Model 3 shows that citizens’ satisfaction with public service provision is positively related to their perceptions of government performance at the macro-level of the public sector (p < .001); in Figure 1b, this is Path b. To this end, the greater the satisfaction with public service provision, the more positive citizens’ perceptions of public sector performance at the macro-level were found to be. These results are in line with previous research (i.e., Bouckaert & Van de Walle, 2003; Van Ryzin, 2004). However, the indirect relationship between computer-mediated transparency and citizens’ perceptions of government macro-performance (Coefficient from Path a × Coefficient from Path b) was found to be trivial (β = −0.004, SE = .004, p < 0.257) 8 .
From Model 3, the second finding of note is that, when controlling for citizens’ satisfaction with public service provision, the magnitude of the direct relationship between the frequency of exposure to computer-mediated transparency and perceptions of government performance at the macro-level actually increases slightly. 9 This relationship is referred to as a suppression effect. A suppressor variable is understood as “a variable which increases the predictive validity of another variable by its inclusion in a regression equation,” where “predictive validity is assessed by the magnitude of the regression coefficient” (MacKinnon, Krull, & Lockwood, 2000, p. 174, citing Tzelgov & Henik, 1991; Shrout & Bolger, 2002). As such, when a suppression effect is present, the magnitude of the independent variable’s coefficient in Path c will be smaller than that from Path c’. In keeping with this explanation, in Model 1, the magnitude of the coefficient for Path c (total effect— Equation 1) was 0.34, whereas for Path c’ (direct effect—Equation 3), which controls for satisfaction with public service provision, the magnitude of the relationship actually increases to 0.39. Furthermore, to test whether the coefficient from Path c’ (Equation 3) was significantly greater than the coefficient from Path c (Equation 1), a Chow test was conducted. The Chow test revealed that the size of the coefficient from Path c’ was significantly greater than the coefficient from Path c (p < .05).
This strong evidence of a suppression effect implies that while information afforded to citizens via transparency policies may negatively influence citizens’ satisfaction with public service provision, the willingness of the government to make public information that is (potentially) self-critical, or is at least perceived as unbiased, also signals to citizens that their government is functioning in a way that ultimately promotes the best interests of citizens and the society they live in.
Discussion and Conclusion
While a great deal of research has examined antecedents and consequences of transparency policies (i.e., Bannister & Connolly, 2011; Etzioni, 2010; Grimmelikhuijsen, 2012), there have been few empirical attempts to assess how transparency influences citizens’ perceptions of public sector performance. This void in the literature is odd when considering that much of the research related to transparency suggests that citizens’ poor perceptions of public sector performance may be due to a lack of information (Cook, Jacobs, & Kim, 2010; Kelly & Swindell, 2002) and has subsequently implied that transparency policies possess the potential to positively influence citizens’ perceptions of the public sector’s actions (Porumbescu, 2013; Welch et al., 2005). The findings of this study provide initial empirical evidence that serves to clarify transparency’s relationship with perceptions of public sector performance. These findings are discussed in greater detail in the following paragraphs.
At an initial level, implications of this research suggest that the utility of transparency as a means of improving relationships between citizens and their government will vary according to different levels of government (i.e., meso- or macro-levels). To some extent, the idea that the effects of transparency policy will vary is not new. For example, research by Cook and colleagues (2010) indicated that individual-level attributes such as cognitive capacity, motivation to acquire new information, as well as socio-economic traits could all differentiate the consequences of transparency initiatives. Similar sentiments are also expressed by Etzioni (2010). With respect to computer-mediated transparency, a rich vein of literature has argued that “digital divides” must be accounted for when assessing the utility of computer-mediated transparency initiatives as a means of improving citizens’ perceptions of government and the public sector, as the use of this information will vary according to factors such as age, income, or education (Reddick, 2005). This study adds an additional dynamic to such discussions by suggesting the need to further explore the ways in which the effects of transparency will vary, not only according to features of the user, but also according to the areas and levels of government. Many have argued that attempts to bolster transparency are far from living up to their transformative potential (Worthy, 2010). The findings of this study reveal the need to consider that transparency policies may be living up to their potential in some respects, whereas in others they may be falling far short of their mark.
A second point to be considered at greater length is that the inclusion of citizens’ satisfaction into Path c’ serves to increase the magnitude of transparency’s positive relationship with perceptions of public sector performance (suppressor effect). This is interesting in that Path a reveals that transparency is negatively associated with citizens’ satisfaction with public service provision.
On the one hand, these results suggest that, in making more information available, transparency policies may lead citizens to view the way their government carries out various functions, such as the delivery of public services, more critically. On the other hand, by making such information available to the public, transparency policies may also signal to citizens that their public sector, while imperfect, is making earnest attempts to perform in a way that advances the best interests of citizens. As such, while computer-mediated transparency policies may lead citizens to doubt their public sector is “doing things well” (i.e., negatively influence perceptions of performance at the meso-level), the information afforded to citizens by such policies may enhance their perception that the public sector is, nevertheless, “doing the right things” (positively influence perceptions at the macro-level).
A third major point to consider is the magnitude of the coefficients in this study, as well as their levels of significance. While this study’s findings revealed a significant relationship between citizens’ exposure to computer-mediated transparency, their levels of satisfaction with public service provision, and perceptions of public sector performance at the macro-level, the magnitude and significance of these relationships are not very high. This suggests that the relationship between citizens’ exposure to computer-mediated transparency and their perceptions of public sector performance at the macro-level is rather weak, particularly when compared to other variables included in this study such as trust in government or satisfaction with the Seoul Metropolitan government’s mayor. This weak relationship may cast doubt upon arguments suggesting that citizens’ negative perceptions of public sector performance stem from information asymmetries between citizens and their public sector. While previous research has suggested that one method of improving citizens’ perceptions of public sector performance relates to reducing information asymmetries (Goodsell, 2004; Kelly, 2002; Yang & Holzer, 2006), the findings of this study, while demonstrating support for this argument, also suggest that there are several other factors that play a larger role in determining citizens’ perceptions of public sector performance.
In discussing the empirical results of this study, some limitations must be addressed. First, the methods employed by this study do not account for the possibility of simultaneity in relationships between citizens’ frequency exposure to computer-mediated transparency, satisfaction with public service performance, and perceptions of macro-level public sector performance. This is because the data set lacked suitable instrumental variables. 10 Second, results of this study are derived from a cross-sectional data set. Taken together, these factors limit causal claims from an empirical perspective. Nevertheless, as outlined in “The Computer-Mediated Transparency and Citizens’ Perceptions of Performance” section, a strong degree of support exists conceptually for using transparency as an independent variable, and perceptions of public sector performance as the dependent variable. A third limitation is the potential for common method bias. Common method bias is essentially a measurement error that is due to the way data were collected. This form of error can cast doubt upon estimates of the relationships predicted in the analysis (Podsakoff & Organ, 1986). One means of testing for common method bias is by conducting Harman’s single factor test. This test employs an exploratory factor analysis using unrotated principal components analysis, and suggests that common method bias is likely if a single factor accounts for most of the variance in the variables examined. To test for common method bias, an exploratory factor analysis including items used to measure perceptions of public sector performance (macro), satisfaction with public service provision, and exposure to computer-mediated transparency were all included. The unrotated principal components analysis illustrated that no single factor accounted for a majority of the variation among the items examined. Moreover, when rotated, the analysis demonstrated that all items loaded onto their appropriate constructs, while cross-loadings of items were all below 0.2. Therefore, the results of Harman’s single factor test suggest that common method bias is not a significant concern, and therefore supports the discriminant validity of the measures. A final limitation of this study is that while our measure of respondents’ frequency of exposure to computer-mediated transparency consists of channels that are commonly used by governments today to enhance transparency (Grimmelikhuijsen & Welch, 2012; Meijer, 2009), we must also acknowledge that not all of the information respondents were exposed to when using the Seoul Metropolitan government’s service web portal and official blog necessarily served to enhance the transparency of the Seoul Metropolitan government. For example, some of the information that respondents accessed could have portrayed government activities in a positively biased light, or lacked details that were necessary to help users better understand what was going on in their government. As such, because we do not know for certain what content respondents were actually exposed to when using the Seoul Metropolitan government’s service web portal and official blog, additional caution should be exercised when attempting to make causal assertions regarding the relationship between computer-mediated transparency, satisfaction with public services, and perceptions of public sector performance.
Despite the empirical limitations, this article contributes to the extant body of transparency literature by demonstrating that the effects of transparency likely vary according to levels of government. This finding affords more nuanced insights into the relationship between transparency and citizens’ perceptions of their public sector. Such a contribution is topical given the degree to which governments around the world are now looking to various forms of computer-mediated transparency to improve citizens’ perceptions of their public sector. Furthermore, from the findings outlined above, various avenues for future research can also be identified, namely, these relate to a need for greater use of content analysis in studies that attempt to uncover relationships between various forms of transparency and the consequences of citizens’ access to this information. Such empirical analysis is necessary to substantiate the findings that stem from experimental and regression-based analyses. Moreover, as recent research by Grimmelikhuijsen and colleagues (2013) suggests, the effects of transparency are also likely to vary across contexts. With those findings in mind, it is also important to glean insight into how factors such as culture may serve to moderate relationships identified in this study. Such information will afford policy makers additional insight into how to make more effective use of transparency as a means of improving relationships between citizens and their government.
Footnotes
Appendix
Responses are measured according to a 5-point Likert-type scale, where a score of 1 indicates the respondent strongly disagrees and a score of 5 indicates the respondent strongly agrees.
Declaration of Conflicting Interests
The author(s) declared no potential conflicts of interest with respect to the research, authorship, and/or publication of this article.
Funding
The author(s) disclosed receipt of the following financial support for the research, authorship, and/or publication of this article: This work was supported by National Research Foundation of Korea Grant funded by the Korean Government (NRF-2014S1A3A2044898).