Citizen requests and the price of public information: An experimental test

3.1Costs and willingness to submit FOI requests

Public choice economics would lead us to believe that individuals are less likely to pay for a service that does not serve their self-interest (Cullis et al., 2009). Indeed, public administration research shows that this is true across a range of different service areas including education (e.g., Glaser et al., 2004), police and fire services (e.g., Donahue & Miller, 2006), and public parks (Glaser & Hildreth, 1996). In general, individuals are proportionately less willing to pay for a service as the cost increases.

The straightforward economic logic becomes slightly more complicated when it comes to a ‘higher level’ value of information when citizens consider whether to submit a public information request that could potentially benefit not only themselves, but society as well. Prior research shows that even citizen actions that are connected to higher level positive social impacts such as public policy participation are still dependent on having the right financial incentives in place (Weathers & Hemmeter, 2011). Thus, we expect financial incentives to also apply to ‘higher level’ public service use such as requesting public information.

3.2Costs and satisfaction with FOI procedures

Prior findings in literature on the relationship between costs of a service and resulting satisfaction with the service are quite mixed. Some research suggests that the introduction of individual user costs to public services is associated with greater service quality and therefore give greater satisfaction to citizens (Warner & Hefetz, 2008). On the other hand, higher costs have been found to be associated with lower levels of satisfaction (Dahlström et al., 2018; Rubery et al., 2013). Citizens approach public services with specific cost-quality expectations in mind and high cost services can lead them to judge the services poorly (Van Ryzin & Immerwahr, 2007).

Another consideration when it comes to FOI costs is that whether satisfaction is raised by a service cost or not may depend on the kind of services where the costs are being applied, and, in particular, whether it is a type of service where costs and user benefits can be clearly justified (Collins & Kim, 2009). Sometimes the real costs of public programs are difficult to quantify given the broad and far-reaching nature of their impacts (Willems et al., 2016). It is therefore easy for citizens to misinterpret the implications of price and the relationship with outcomes and outputs (Grosso et al., 2017). We reason that adding high costs to a public service such as delivering information seems particularly likely to raise dissatisfaction levels for two main reasons. First, for a strongly rights-based service such as freedom of information, a cost barrier could evoke a particularly negative response in terms of citizen satisfaction because being asked to pay for something undermines the sense of ‘having a right’ to it. Second, it is difficult for citizens to see what is entailed administratively in retrieving public information. Implementing a right does come with some operating costs, and may thus lead to misunderstanding of the purpose of fees.

3.3Costs and perceived fairness of FOI procedures

Smith (1996) argued that introducing market mechanisms with higher individual user costs for citizens through NPM reforms also introduces specific norms and expectations regarding fairness in public services. In general, this could have a positive impact on perception of fairness among citizens because costs are borne by the direct users of services rather than all taxpayers. Some research findings do support this perspective of cost. For example, Jang and Eger (2019) found that applying pay-for-performance reforms to services can increase the perceived procedural fairness of those services. On the other hand, introduction of higher costs may sometimes lead to lower perception of fairness. Indeed, contrary to Smith’s view, perceived equitability of services for citizens might go down when user fees are introduced to services because ability to pay can override the need to deliver a service equitably.

Similarly to the case of citizen satisfaction, the effect of cost may lie partially in citizen expectations and the service area. Concerns about cost and equitability are frequently linked in literature on public service delivery especially when it relates to an important public good. In this light, Hemelt and Stange (2016) found that when marginal price rises are introduced in public goods areas such as education, they negatively affect poorest students most. While individual costs and fees in public services can be a fairer way to distribute costs, the right to public information, like other public goods such as justice or education, is typically viewed as a right for everyone regardless of their economic situation. We expect this rights-based characteristic of FOI to influence individuals so that high cost information requests procedures are viewed as less fair than procedures that are not as costly.

3.4Moderating variables: PSM and political ideology

Decisions about the cost of FOI requests also circle around the question of individual normative values and attitudes. Evidence shows that how well FOI systems work is linked not just to the quality of the legal stipulations but also to the role of informal and symbolic mechanisms involved in the communication of public agencies with the public (Spáč et al., 2018; Worthy et al., 2017). According to Jones and Cullis (2003), when it comes to evaluation of public services, there is often an interactional relationship between instrumental considerations of cost and normative responses to public policies. Whether cost considerations influence attitudes partially depends on how a policy triggers normative and motivational attitudes of service users.

Freedom of information requests often have strongly normative, public accountability seeking goals, but they can also be motivated by purely individual or micro-political goals. Berliner, Bagozzi, and Palmer-Rubin (2018) offered big data insights into what proportion of FOI requests were of either kind of motivation in the context of the Mexican federal government. Just over half of the nearly 1 million requests analysed were about medium or high accountability relevant topics such as land and environment, budgets and spending, public utilities, education and welfare, finance, and behaviour of politicians. This suggests that pro-social motivations often drive FOI requests, but a large number of requests are for more mundane matters such as, for example, finding out how to apply for a public program or making a business-related request on patent ownership.

Motivations may also influence not just requesting behaviour but the perceptions that individuals have of the worth of the FOI requesting procedure as a whole. A similar relationship has been observed in research on other public service areas. For example, research by Overton and O’Mahony (2017) found that individual pro-social values moderated whether the introduction of individual cost schemes in public services will be perceived as fair or not. Pedersen, Stritch, and Taggart’s (2017) research found similarly that the involvement of a private provider in an internal administrative process was more likely to be perceived as fairer by staff who had higher PSM. In this case, PSM gives individuals a reason to look beyond the monetary costs to the bigger public service goal.

It appears that PSM gives individuals trust towards the public sector that allows a certain extra degree of acceptance when self-interested private motives may be at work. Of course, not all FOI requests are driven by public interest motives, but public interest does play a big role in motivating many individuals to submit requests (Berliner et al., 2018). Thus, we expect PSM to influence request behaviour by positively moderating (i.e., mitigating) the effect of cost on willingness to submit a request. Based on prior findings on the moderating role of pro-social motivation and PSM, we also expect cost information and PSM to interact in the same way in their effect on perceived fairness and satisfaction.

When it comes to non-cost related factors that influence requesting behaviour, political orientation could play an important role. As the design of FOI laws has been shaped by political considerations of how information and costs of information should be distributed in society, we also expect that a similar moderating effect could be found in individual political orientation and the effect of cost on satisfaction and fairness. Political orientation including political preferences for levels of public spending and favourability towards public values such as transparency plays a significant role in whether individuals will be satisfied with the service they have paid for (Rabovsky, 2014). Given that citizen political orientation influences which types of policies are evaluated highly or poorly (Hoff et al., 2015), it appears likely that political ideology may play a role in the evaluation of the fairness of FOI procedures too as free-market liberals tend to be more tolerant of paying for improved public service performance (Dahlström & Lapuente, 2010; Rabovsky, 2014). Mediano and Carrasco (2019) have even shown that the political orientation of judges who review rejected FOI requests plays a role in adjudicating final decisions.

Liberals and conservatives tend to approach decisions about pricing of public services differently. In general, liberals support social equity and rights-based approaches to funding of public services delivery (Jacobs, 2014; Rudolph & Evans, 2005; Skitka & Tetlock, 1993). In contrast, conservative political ideology is generally associated with support for market-oriented public service delivery (Boyer & Van Slyke, 2019; Mohr et al., 2010; Schoute et al., 2018). While, in line with hypotheses 1–3 above, paying more for a service is generally likely to elicit a negative response from liberals and conservatives alike, the normative or rights-based character of FOI services is likely to be more important for liberals, and a cost barrier is more likely to be interpreted unfavourably. We therefore expect that the negative effect of cost on perception of fairness, and satisfaction to affect liberals more strongly than for conservatives.

4.1Experimental design

We conducted an online survey experiment to test our hypotheses. Experimental studies are a suitable tool for measuring perceptions of citizens in salient areas of public policy and public service provision (e.g., Bell, 2021; Favero & Kim, 2021), and are especially useful for studying the effects of small information variation or framing effects (Jones, 2014; Valant & Newark, 2020). By experimentally varying amounts of an independent variable in a treatment group where other variables are otherwise held constant, researchers can ascertain what leads to differences in an outcome of theoretical interest in the policy area (James et al., 2017). In our experiment, participants were randomly allocated to one of three treatment groups, each with different information on the fee associated with the procedure to submit an information request. Aside from the cost information, all participants received the same vignette describing a fictional case involving a situation where they had gone to a well-known restaurant chain and subsequently fallen sick with suspected food poisoning. A FOI request for information about the restaurant’s health and safety record was presented in the vignette as a way to take action that could prevent other people from getting sick in the future. In the US context this could be directed at local, state or federal government levels. The federal level was the most plausible choice to make as the Centers for Disease Control and Prevention (CDC) in the Department of Health and Human Services monitors foodborne illness outbreaks across the country and shares some of its data publicly, though not at the level of individual restaurant franchises. This choice also avoids the complication of having to specify a local agency for each survey participant. This specific focus of the vignette on a public health scenario was selected in order to draw upon a public service area where citizen attitudes to transparency are important to study because of the implications for citizen life and well-being (De Fine Licht, 2014; Ingrams et al., 2020). The full text of the vignettes can be found in Appendix A.

We manipulated cost using three realistic cost levels based on the rates given by the US federal agencies. A lower bound cost is frequently set as a guideline for agencies to determine whether administrative costs are negligible enough or not to make the request free for the requester. In the United States, where the study’s participants live, this lower bound ranges from $5 (Department of Education) to $50 (Department of the Interior) but most agencies set this amount at $15 or $25. We thus set the ‘low cost’ amount at $15, which is low but still substantial enough to give an average citizen requester pause for thought.

We set the high cost at $150, which is exactly ten times the low-cost amount. The high cost amount represents a cost level that is both substantial and realistic. Specifically, based on the clerical hour rates given by the Department of State FOI Office of $21 plus 15 cents per page, the high cost seems like a plausible amount for a more substantial request for information that took about 5 hours of clerical time to process plus page printing costs.22 Of course, much higher costs are possible, but at very high costs of thousands of dollars the empirical basis for understanding differences in reactions of requesters between no cost, and low and high costs starts to become harder to interpret. Put differently, a high amount should still be within reasonable limits of affordability for an average citizen requester.

After reading the vignette and seeing how much it would cost to submit the information request, participants were asked questions on the three dependent variables of interest: (1) how likely they would be to submit the request, (2) how satisfied they were with the request procedure, and (3) how fair they perceived the procedure to be (information on our measurement items and approach is in Appendix B).

We used the Qualtrics platform to implement the survey. TurkPrime, which is run with participants from MTurk, was used to recruit survey participants (James et al., 2017; Stritch et al., 2017). We took advantage of TurkPrime’s features to only select experienced and reliable participants. Specifically, participants were required to have at least a 99% approval rating from prior researchers, and to have completed at least 1,000 MTurk tasks for prior research. MTurk participants are paid for their work. We followed recent scholarly precedents to determine compensation rates and response validity checks, which we report below, to address data quality concerns (Kaufmann et al., 2019; Keiser & Miller, 2019; Vogel & Willems, 2020). However, it should be noted that MTurk participants do not represent a true random sample as they tend to be disproportionately male and have low levels of current organizational experience (Chmielewski & Kucker, 2020; Stritch & 2017). The mean length of time needed to complete the survey was just over five minutes, and the participants were paid 80 cents.

4.2Data

We collected 1002 usable responses from the survey. Reliability analysis showed that our principal component factors for PSM (α= 0.90) and satisfaction (α= 0.97) had good Cronbach alpha scores. This suggests that they can safely be scaled. The results of the attention check indicate moreover that respondents paid sufficient attention to the survey. Of all respondents, 8 failed the check and were removed from the sample to ensure the quality of the data. Similarly, the results of the manipulation check suggest that the manipulation was successful. When asked to identify the correct manipulation, 69 of the remaining respondents chose the wrong manipulation corresponding with the manipulation they were subjected to and were also removed from the dataset. This brings the final sample to 925. The questions for the attention and manipulation checks are in Appendix C.

We carried out a test to assess the demographic balance in the experimental groups. An Analysis of Variance (ANOVA) suggests that there are no significant differences between the experimental groups on demographic variables like gender (F(2, 922) = 1.97, p= 0.14), age (F(2, 922) = 0.87, p= 0.42), and income (F(2, 922) = 0.43, p= 0.65) (see Table 1).33

Figure 1 shows mean responses for willingness to request, satisfaction, and perception of fairness in the three cost treatment groups. For all variables, average scores are highest for the no cost group, followed by low cost, and, finally high cost. This is in line with our theoretical expectations.

A series of ANOVA’s confirms these descriptives. For willingness, the results suggest that respondents are less likely to put in a FOI request as the cost of doing so increases (F(2, 922) = 211.66, p< 0.001). Similarly, the results suggest that respondents are less satisfied with a process that imposes more costs (F(2, 922) = 360.09, p< 0.001), and consider such a process less fair (F(2, 922) = 341.37, p< 001). Post hoc tests confirm that this is true for both the difference between no cost and low cost, and the difference between low cost and high cost (see Table 2).44 Effect sizes are large for both satisfaction (η𝑝𝑎𝑟𝑡𝑖𝑎𝑙2= 0.44), perceived fairness (η𝑝𝑎𝑟𝑡𝑖𝑎𝑙2= 0.43), and willingness to submit (η𝑝𝑎𝑟𝑡𝑖𝑎𝑙2= 0.32).

Figure 1.

Mean scores, with 95% CI.

Figure 2.

Predicted values for willingness to request and PSM.

Figure 3.

Predicted values for satisfaction and political orientation.

Figure 4.

Predicted values for perceived fairness and political orientation.

The results provide mixed evidence for our fourth hypothesis (see Table 3). For willingness to request, there is a significant interaction effect with PSM (F(109, 746) = 1.32, p< 0.05). Further analysis suggests that the negative effect of cost is less pronounced for respondents with high PSM, particularly with regard to the low cost scenario (see Fig. 2). This is in line with our hypothesis. The effect size is large (η𝑝𝑎𝑟𝑡𝑖𝑎𝑙2= 0.16). Interaction effects for satisfaction (F(109, 746) = 0.96, p= 0.59) and perceived fairness (F(109, 746) = 1.10, p= 0.23) are not significant.

The results for the interaction effect of cost and political orientation (shown in Table 4) also show mixed support for our theory. Contrary to our results for the previous hypothesis, political orientation does not interact significantly with the effect of cost on willingness to request (F(12, 904) = 1.12, p= 0.34). However, in line with our hypothesis political orientation does interact significantly with the effect of cost on satisfaction (F(12, 904) = 2.27, p< 0.01) and perceived fairness (F(12, 904) = 2.12, p< 0.05). Further analysis suggests that the negative effect of cost is more pronounced for respondents that have a liberal political orientation (see Figs. 3 and 4). This is in line with our hypothesis. The size of both effects is small (η𝑝𝑎𝑟𝑡𝑖𝑎𝑙2= 0.03).