Introduction

What do citizens think of interest groups when lobbying is transparent? Despite the growing interest for lobbying and public opinion research in political science, little is known about citizen’s attitudes towards interest groups, especially after legislation making lobbying visible to the public is passed. Western democracies have in recent years experienced a push to introduce more transparency in lobbying with the promise in mind that lifting the lid over lobbying may help improve trust in the political process and decrease the perception that vested interests have undue influence over the democratic process. The extent to which lobbying transparency achieves this goal remains understudied but also unchallenged.

Using data from a pre-registered survey experiment conducted with 569 participants from a larger experiment involving 1800 citizens from the Republic of Ireland, this study shows that citizens exposed to information about lobbying do not have more positive evaluations of trust and integrity of interest groups compared to citizens assigned to a control group.

I explore the effects of two levels of lobbying transparency (low and high) on trust and perceived corruption when information about the lobbying activity of non-governmental organisations (NGOs) and business organisations (associations and firms) is disclosed on lobbying registers. These are online instruments of government openness established through lobbying regulations/acts, which require interest groups to disclose information about their activity, and state institutions to organise and make this information accessible to the public. In this experiment, I use a short vignette followed by replicas of such a lobbying register of which I manipulate the level of detail of the disclosed information about lobbying actors. I then collect respondents’ evaluations of trust and integrity of two interest group categories, business organisations and NGOs, and compare them to those of a group of citizens assigned to a neutral control condition.

The results of the experiment challenge what is generally assumed by advocates of lobbying transparency and indicate that no obvious beneficial direct effect of transparency in improving citizens’ attitudes towards lobbying exists. On average, lobbying transparency has a null effect on trust and perceptions of corruption in interest groups. Such null effect is robust to the inclusion of potential moderators as well as equivalence tests of smallest meaningful effect (Rainey 2014). While the absence of such an effect may be explained by mechanisms found in the public administration literature on transparency, one key finding stands out: lobbying transparency does not backfire, that is, it does not produce undesired negative effects (De Fine Licht 2011; Worthy 2010).

This study moves beyond existing scholarship of interest groups and public opinion which has predominantly focussed on the congruence between policy preferences and its effect on public policy (De Bruycker and Rasmussen 2021; Rasmussen et al. 2018). Moreover, it speaks to the growing interests for how citizens perceive interest groups’ involvement in policymaking (Bernauer and Gampfer 2013; Bernauer et al. 2016; Beyers and Arras 2021; Rasmussen and Reher 2023; Aizenberg et al. 2023). This study analyses the ability of a commonly accepted instrument of lobbying transparency to support such perceptions (Chari et al. 2020).

Finally, this study contributes to the policy analysis of lobbying regulations (Kanol 2018; Crepaz 2020; Crepaz and Arikan 2023), suggesting that while useful, lobbying transparency may have a weaker direct impact on the public than what assumed and hoped for by governments, advocates and international organisations that promote their introduction. While the results of a single experiment remain, of course, limited in scope and are restricted in terms of external validity, they do widen the debate around the usefulness of transparency in the realm of lobbying.

In the next section, I review the existing literature on attitudes towards interest groups, identify relevant knowledge gaps, and explain how, despite such gaps, transparency advocates assume that lobbying transparency will improve the citizens’ attitudes towards lobbying. Then, building on the existing public administration literature, I present pre-registered expectations on the effects of lobbying transparency on evaluations of trust and integrity of business organisations and NGOs. The following sections explain the experiment design and present the results of the analysis. The final sections discuss the results and suggest avenues for future research.

Public attitudes towards interest groups and lobbying

Interest groups are essential vehicles of economic and societal interests that are typically voiced through engagement in a range of political activities, such as protest, advocacy, political contributions, and lobbying (Lowery et al. 2015; Dür and Mateo 2016; Alemanno 2017). With a focus on the latter two forms of political activities, public opinion studies display a tendency of portraying interest representation and lobbying as ‘shady business’ by default. These works put attitudes towards public institutions, political parties and politicians at the centre and portray interest groups instrumentally as ‘special interests’ to better understand how citizens react to corruption and potential conflicts of interest (Fisher et al. 2010; Hibbing and Theiss-More 2002). In doing so, they assume the respondents’ negative evaluation of interest groups and of their involvement in decision-making, without providing an assessment of public opinion towards interest groups beyond the image of lobbying as a corrupt attempt of policy capture.

Interestingly, studies of non-state actors, like unions and voluntary groups, suggest that attitudes, such as trust, tend instead to be higher for these organisations compared to state institutions and political parties (O’Neill 2009; Frangi et al. 2017). At the same time, however, attitudes documented in these works are untied from the organisations’ political activities, such as lobbying and advocacy, providing no helpful perspective on the latter.

A recent group of studies finally contextualises citizens’ attitudes within the role of interest groups as policy stakeholders. They find that citizens perceive external stakeholder involvement in policymaking as more legitimate and accepted if visible and balanced between representatives of economic and societal interests (Bernauer and Gampfer 2013; Bernauer et al. 2016; Beyers and Arras 2021; Rasmussen and Reher 2023; Aizenberg et al. 2023). This work added much needed clarity to how citizens form perceptions of internal and external legitimacy, procedural and policy acceptance in the context of stakeholder involvement in policymaking.

There are, however, two potential shortcomings of the approaches taken in these groups of studies this paper wishes to address: First, the omission of trust and perceptions of corruption as foundational components of the support for the democratic process alongside legitimacy (Levi and Stoker 2000). Second, the dependence—in these studies—upon specific policy issues and consultations practices to evaluate citizens’ attitudes towards interest groups (e.g. global environmental governance: Bernauer and Gampfer 2013; parliamentary/public hearings and fiscal incentives: Rasmussen and Reher 2023). Lobbying transparency regulations, as meta-laws (laws on the processes of policymaking), are a useful case of study to move beyond the context of specific policy issues and styles of governance, because they seek to shed light on lobbying across all policy areas (Chari et al. 2020).

To my knowledge, this is a rare approach, with Kanol (2018) providing the only other existing study of the effects of lobbying regulations on public attitudes.Footnote 1 This study is, however, limited to the effect of ‘knowing that lobbying is regulated’ on general perceptions of corruption in politics, without revealing how citizens view interest groups. Based on the studies reviewed in this section, my approach on transparency in lobbying allows me to theorise how individuals form evaluations of integrity and trust in interest groups when exposed to information about lobbying, thus filling a theoretical gap. To do so, I additionally build on the literature on transparency in public administration and that on trust and integrity of political actors.

Theories of transparency applied to lobbying

This study defines transparency as ‘any release of information about institutions that is relevant for evaluating those institutions’ (Lindstedt and Naurin 2010, 301). Typically, this takes the form of disclosures by state institutions after freedom of information requests. In more recent times, transparency has shifted towards proactive forms of transparency, whereby public institutions make information available to the public through open data portals, government websites and public registers (Porumbescu et al. 2022).

Irrespective of its form, the literature describes transparency in politics as a necessary and desirable condition for political trust and freedom from corruption (Hood and Heald 2006). In relation to proactive forms of transparency, scholars have shown that the disclosure of information can generate new knowledge among citizens and that this affects their political attitudes and willingness to participate in politics (Grimmelikhuijsen and Meijer 2014; de Fine Licht 2014). Transparency has been shown to also have a symbolic effect, in the sense that it is the ‘act of disclosure’ which provokes attitude changes (Crepaz and Arikan 2021, 2023). By disclosing their activities, political actors ‘signal’ to the consumers of information that there is nothing to hide, that they are being trustworthy and free from corruption (Park and Blenkinsopp 2011; Kanol 2018).

The importance of these mechanisms for trust and perceived corruption have been explored (but sometimes only tangentially) in relation to transparency in local public administration (Grimmelikhuijsen and Meijer 2014), open government (Chalikova and Mabillard 2020), access to information laws (Worthy 2010), political donations (Sances 2013). However, they remain unexplored in the realm of lobbying transparency. Yet, its role is important given that, without knowing which interests have the ear of elected representatives, citizens will find it more difficult to hold politicians accountable for their decisions (Naurin 2007; Rasmussen and Reher 2023).

The disclosure of information about lobbying typically takes the form of public searchable registers. This is a form of transparency in process, which allows external actors to scrutinise the involvement of interest groups in the policymaking process (Hood and Heald 2006; Grimmelikhuijsen and Meijer 2014). This form of transparency is advocated in contemporary democracies in the form of lobbying acts. These acts require interest groups to disclose details of their lobbying activity, their aims, targets and expenditures, which should, along with other aims, allow citizens to evaluate the honesty and integrity of interest groups acting as external stakeholders in policymaking (Chari et al. 2020).

Indications about the theoretical validity of this mechanism are found in studies of citizens’ perceptions of legitimacy and procedural fairness of the policymaking process where interest groups are involved. In a series of survey experiments, Beyers and Arras (2021), Bernauer and Gampfer (2013) and Bernauer et al. (2016) find that participants’ awareness that external stakeholders are involved in policymaking through different forms of consultations increases their perceived legitimacy and acceptance of the policymaking process and its outputs compared to scenarios where they are not, or no information is provided.

This is not only the case because citizens normatively prefer policy that is negotiated with different stakeholders to one that does not see their involvement (Beyers and Arras 2021), but also because ‘seeing’ that interest groups are involved in policymaking gives a sense of control and empowerment to citizens: First, they have rubber-stamped interest representation and can hold decision-makers accountable in case they dislike what they see Naurin (2007). Secondly, being able to monitor disclosed information and having knowledge about the transparency requirements disclosers are subject to reduces the perceptions that their activities are illegal and corrupt (Kanol 2018).

With this feature in mind, I expect that when information about lobbying is disclosed on public registers, citizens will evaluate interest groups involved in lobbying more positively by considering them as more trustworthy and less likely to engage in corruption. This expectation is informed by the mechanisms explained above as well as by the widespread expectation found in transparency literature (in particular the experimental strand of research studying the direct effect of information disclosure on attitudes) that the disclosure of process information is likely to improve citizens’ attitudes towards the actors involved in the process: firstly, because, citizens—by accessing information otherwise not available—learn about the process and are better able to judge it (Grimmelikhuijsen and Meijer 2014; de Fine Licht 2014); secondly, because the disclosure of information by actors involved in the process signals honesty and good intentions to citizens who are therefore more likely to trust those actors and accept the perception that there is ‘nothing to hide’ (Crepaz and Arikan 2021, 2023). Both mechanisms are present in Kanol’s (2018) work on knowledge about the regulation of lobbying, which suggests that individuals that know that lobbying is more heavily regulated (and hence more transparent) display higher levels in trust in government and lower perceptions of corruption, compared to less strict regulation of no regulation. It follows that, if transparency has different degrees of disclosed information, more transparency should be associated with higher levels of trust and lower perceptions of corruption compared to lower and no transparency (Grimmelikhuijsen and Klijn 2015; Kanol 2018).Footnote 2

Hypothesis 1

Transparency in lobbying increases evaluations of trust and reduces perceptions of corruption of interest groups. This effect is expected to be stronger for higher compared to lower levels of transparency.

The extent to which lobbying transparency achieves such objectives may however depend on the actors (that is the interest group) that are the object of the evaluation. Both political trust and perceptions of corruption are relational concepts, meaning that certain characteristics of political actors may signal more or less distrust or lack of integrity to citizens (Levi and Stoker 2000). For example, a politician that is also a media mogul is likely to provoke different evaluations than a politician without ties to the business world.

In lobbying, a key aspect, both in public opinion and political science scholarship, is bias. The perception that policymakers may be ‘captured’ by special interests and that this will then bias policy outcomes is a constant concern (Schattschneider 1960; Lowery et al. 2015). The ability to hijack policymaking in this sense is generally ascribed to business interests, first because of their generally higher availability of resources to hire lobbyists to represent their interests in policymaking (Dür and Mateo 2016), and secondly, because of their structural power in the economy, which politicians acknowledge and second because of electoral incentives (Culpepper 2015).

In the study of bias, business interest representation is benchmarked against that of so-called cause groups, representing broader societal interests, such as the protection of the environment, or the fight against poverty. The comparison between the two is one possible way of assessing balance and fairness in the participation of interests in policymaking (Lowery et al. 2015). Importantly, it is clear from recent studies that citizens are aware and care about the dangers of this kind of unbalance. For example, the same research that studied perceived legitimacy and procedural fairness of interest group involvement in policymaking found that balanced stakeholder involvement between business and cause groups, such as NGOs, improves citizens’ attitudes towards the process (Bernauer and Gampfer 2013; Beyers and Arras 2021; Rasmussen and Reher 2023). Even when only one of these interest group categories were involved in policymaking, citizens perceived NGOs’ involvement as comparatively more acceptable than that of business alone.

This is explained, first, by the vicinity between citizens’ preferences and the ‘public’ interest represented by cause groups (Giger and Klüver 2016; Flöthe and Rasmussen 2019). This preference and/or ideological proximity improves perceived representation and does then translate in other positive attitudes such as trust and overall perceptions of the integrity of the organisation; Second, citizens perceive cause groups included in policymaking as scrutinisers that make sure that public interests will be respected by governments and other stakeholders as agreed (Bernauer and Gampfer 2013).

With these assumptions in mind, the effects of transparency (instead of being the result of an update of attitudes as a result of information acquisition) derive from the reduction of psychological distance between information users and information disclosers (Grimmelikhuijsen et al. 2021): information disclosure about lobbying of business groups and NGOs, one the one hand, allows citizens to assess whether representatives are in accordance with their interests; and their interests are already more closely aligned with ‘public’ groups (Bernauer and Gampfer 2013). On the other hand, it signals to information consumers, not only, that organised interests have nothing to hide, but also that some, more trustworthy and legitimate organisations, will keep a close watch on potential drifts away from public interest. According to this reasoning, citizens will draw a distinction between business groups and NGOs and form comparatively higher evaluations of trust and lower perceptions of corruption towards NGOs than towards business groups.

Hypothesis 2

The effect of transparency is expected to be stronger for evaluations of trust and perceived corruption towards NGOs compared to business groups.

It needs to be, however, further specified that I do not expect lobbying transparency to have, on average, negative effects on attitudes towards business groups. When observable, the participation of business interests in policymaking is perceived (on average) positively compared to no participation at all, although comparatively less so than for NGOs (Bernauer and Gampfer 2013; Beyers and Arras 2021; Rasmussen and Reher 2023).

The potential for transparency to create diverging (if not polarising) effects, where information disclosure has a different impact on attitudes towards disclosing actors for different citizens and other external scrutinizers, becomes apparent when we consider the influence of conditional or moderating factors, such as the characteristics of individuals who consume the lobbying information. Previous literature explored the effects of education, knowledge and political interest as moderating factors of public administration and government transparency (Lindstedt and Naurin 2010; Grimmelikhuiisen and Meijer 2014; Schmidthuber et al. 2021). In the context of lobbying transparency, I explore the effects of two moderating factors: political interest and political ideology.

First, I argue that higher levels of political interest moderate the beneficial effects of lobbying transparency. While research on citizens’ attitudes towards interest groups is rare (see Frangi et al. 2017; O’Neill 2009), it has been recently established that individuals tend to have a lower understanding of interest groups and their involvement in policymaking compared to that of political parties and members of parliament (Crepaz and Arikan 2023). This is also evident from historical poll data as well as more recent studies that discuss citizens’ attitudes towards lobbying (Benedict 2004; Alemanno 2017). More interested individuals that are aware and knowledgeable of the complexities of the political process, should, however, be more likely to, first, recognise the role of interest groups in policymaking, second, process lobbying information disclosed on lobbying registers and then respond positively to the disclosure and acquisition of new information about lobbying (Grimmelikhuiisen and Meijer 2014; Crepaz and Arikan 2021, 2023).

Hypothesis 3

Political interest is expected to moderate the effect of transparency on the evaluations of trust and perceived corruption of interest groups.

Different information users may however have priors based on different factors, one of which being political ideology, that influences the formation of attitudes towards business interests and NGOs. This expectation is based on the well-established observation that (centre) rightist political parties tend to have closer relationships with organisations representing business interests on multiple dimensions, ranging from ideological proximity to more formal ties and resource dependencies (Thomas 2001; Allern and Verge 2017). (Centre) Leftist parties, especially, what is called the New Left, emerging from (e.g. environmental or human rights movement) activism and social movements, are described instead as having ties to various cause groups, including NGOs (Allern and Verge 2017; Barberà et al. 2019). The assumption underlying the expectation of a conditional effect of transparency based on political ideology is that citizens recognise (in)congruences in the three-way relationship between their own political ideology, the stance of political parties and that of external actors they are linked to. I argue that this is a reasonable assumption given the high-level congruence between elected representatives, voters and mobilising interest groups documented in recent studies (see De Bruycker and Rasmussen 2021; Rasmussen et al. 2018).

I therefore expect leftist individuals to improve their trust and decrease their perceived corruption evaluations of NGOs, but not necessarily of business groups, when exposed to lobbying transparency. Conversely, rightist individuals are more likely to improve their trust and decrease their perceived corruption evaluations of business groups, but not necessarily of NGOs, when exposed to lobbying transparency.

Hypothesis 4

Political ideology is expected to moderate the effect of transparency on the evaluations of trust and perceived corruption of interest groups. Leftist citizens are expected to improve evaluations of NGOs (but not necessarily of business groups) when exposed to lobbying transparency. Rightist citizens are expected to improve evaluations of business groups (but not necessarily of NGOs) when exposed to lobbying transparency.

These diverging effects of transparency on expected attitudes moderated by political ideology can be explained by the mechanism of motivated reasoning, whereby, in processing information, individuals are motivated to seek out and carefully consider information with the goal of reaching a correct and accurate conclusion. In the particular case of lobbying transparency, confirmation bias, which refers to the process by which individuals search for or interpret information in a way that reinforces their prior beliefs (Taber and Lodge 2006), can explain why individuals react to transparency by selectively updating attitudes towards the interest group they feel closest to. At the same time, cognitive dissonance, by which individuals ignore or dismiss information that is discordant with their priors, explains why transparency is expected to have no beneficial effect on attitudes towards interest groups citizens feel most distant from (Redlawsk 2002). In such circumstances, transparency’s effect is symbolic and follows a different psychological mechanism than theorised under the knowledge-based mechanism. It is not the information itself, but rather the meaning and priors attached to it that determine the user’s reaction (Grimmelikhuijsen et al. 2021). This process, which potentially undermines or complicates the expected positive effects of information disclosure about lobbying, has been already observed in relation to other transparency policies, which led to the generally accepted conclusion that the effects of transparency policy are more complex than often assumed (Grimmelikhuijsen and Kljin 2015; Crepaz and Arikan 2021; Porumbescu et al. 2022).Footnote 3

Research design

The experiment

Between May and June 2020, the research team conducted a survey experiment as part of the project Assessing the Effects of Transparency Laws on Different Political Actors (Irish Research Council funding: GOIPD/2018/52) with a panel of 1800 Irish citizens recruited through the polling company Ireland Thinks and representative in terms of age, gender and level of education. The survey experiment was aimed at studying the effects of transparency in donations to political parties, asset disclosures of Members of Parliament, lobbying activities of interest groups in Ireland.Footnote 4 The experiment consisted of one control condition and eight experimental conditions. While I here present the results of the experiment for the two treatment groups concerning lobbying transparency, the results of the other treatments are presented elsewhere (Crepaz and Arikan 2021, 2023). The experiment was pre-registered.Footnote 5

For this study, the respondents were randomly assigned to one of the two vignettes concerning lobbying registers in the low (LT) and high transparency (HT) conditions plus a control condition. The latter contained a short vignette of no political relevance concerning a cultural event; no information about lobbying was shared or mentioned (See Online Appendix 1). This establishes a common and neutral baseline against which all other experimental groups can be compared. The downside of this approach is that it makes it impossible to capture the effect of baseline information and separate it from the treatment effects. Without information about the ‘general activity of lobbying’ in the control group condition, it is, for example, theoretically possible that the simple awareness that lobbying is happening could have counteracted the positive effects of transparency. This would have, however, required the inclusion of multiple control groups inflating the already high costs associated with running a 2 × 4 factorial design experiment. With this in mind, the concluding section discusses possible improvements to be considered in future studies.

The treatments represented vignettes including fictional replicas of information found on the Irish Lobbying Register (lobbying.ie) which has been in place since 2016.Footnote 6 Because citizens are unlikely to directly access this type of information, the vignettes were written in the form of a newspaper article that reported about the most recent report published by the enforcement authority of the register (SIPO). The vignettes, shown below, were designed after the implementation of a pilot with 371 students enrolled in politics modules at Trinity College Dublin.

This part of the survey concerns the memory of everyday events. Specifically, we are interested in how well people pay attention to information about Irish lobby groups found online (lobby groups are organisations that seek to influence public policy through lobbying). Next you will be asked to read a short newspaper article containing detailed information about political events in Ireland. Please pay careful attention to what you read. You will be asked to answer a short set of questions after.

Figures released today by the Standards in Public Office Commission (SIPO) in its annual report of lobbying activity in Ireland show that almost 9,800 returns of lobbying activities have been submitted for the period of 2018. 1,800 registered organisations declared to have lobbied Irish Public Bodies at least once in 2018.

Of the 1,800 registered organisations, 40% are organisations representing business and professional interests, 40% are non-governmental organisations (NGOs like Greenpeace and Concern). All information is open and can be accessed on the Website of SIPO and lobbying.ie. Below is an abstract of the declared activities.

This text was then followed by a table containing a sample of disclosed information (See Online Appendix–1. Treatment Description). The low transparency (LT) condition displayed information about the lobbying organisation (name), the number of lobbyists employed by the organisation, the name of the person responsible for the lobbying activity, the number of meetings interest groups had with policymakers and summary information about the total number of active lobbyists and number of meetings. These are typical features of what the lobbying regulation literature defines as ‘low robustness’, whereby transparency registers provide limited information; not beyond a list of lobbying organisations and of their activities (Chari et al. 2020).

The high transparency (HT) condition also displayed the target of the activity (e.g. the government department and elected official) and the expenditures related to the lobbying activity (in €). These features were purposely selected for the high transparency condition, as the literature associates them to features of more robust, hence more transparent, regulations (Chari et al. 2020). In addition, whether to disclose lobbying expenditure or not is a controversial debate in regulatory reform and has divided activists, who want full transparency in lobbying expenditures, and regulators, who have claimed that the use of money in lobbying may be mis-represented in the media and public debate.Footnote 7

The reporting of information about lobbying activities is rare but not unprecedented in the Irish media. Broadsheets occasionally report stories about lobbying, with information directly taken from the lobbying register (Crepaz and Kneafsey 2021). These stories often involve suspected cases of corruption, undue influence or questions about conflict of interest.Footnote 8 The experiment vignettes refrain from providing such negative angle favouring a neutral approach to avoid provoking the so called ‘Bismack/sausage-making effect’ where disclosure of negative information backfires (de Fine Licht et al. 2014). For example, the number of meetings with designated public officials disclosed on the mock lobbying register are intentionally similar, to avoid portraying bias and unbalance in external stakeholder involvement. A similar approach was taken for the number of employed lobbyists per organisation and the lobbying expenditures displayed in the treatments (see Online Appendix–1. Treatment Description). This neutrality was achieved with only 4.8% of the respondents indicating that they found the content to be biased in favour of one of the actors. In addition, as part of the experiment’s manipulation check, I collected open-ended responses from a post-treatment question, which contain details on the respondent’s perception of transparency in lobbying. These are used as proxy evidence in the discussion of the results.

In addition to the above design choices, I intentionally avoided including cues explicitly stating that some information ‘had not been disclosed’ under the low transparency treatment. Such cues would have evoked benchmarking (Charbonneau and Van Ryzin 2015), cynical responses and attitudes provoked by ‘openwashing’ rather than by transparency itself (Heimstädt 2017), making it therefore more difficult to disentangle whether the observed effects were produced by the disclosure of information or by selective disclosure.

I opted for a treatment design that relatively resembles how individuals would consume and visualise information on transparency registers in real life, increasing the experiment’s mundane realism and potential for external validity of the findings (Druckman and Green 2021). The downside of this approach that, with so much information available in the treatment—from the names of lobbying organisations, the subject matter of their activity, their targets etc.—it becomes impossible to pinpoint which element potentially causes an effect (if any). Future research could investigate this by isolating the effects of specific disclosure details improving the causal-inference potential of this experiment. For external validity reasons, it is however important to stress that users of transparency portals, such as citizens, never consume isolated pieces of information but always in rather a complex combination.

Finally, Ireland’s approach to lobbying transparency aligns with that taken by the EU and other European countries with lobbying laws in place (Chari et al. 2020). Furthermore, it is no outlier in western Europe when it comes to media bias, lobbying or trust in politics and levels of perceived corruption (Dür and Mateo 2010; Crepaz and Chari 2022). The choice to focus on Ireland therefore does not necessarily prevent one from applying the approach followed here to other polities in western Europe.

The experiment used a between-group design whereby attitudes towards interest groups are compared between LT and HT conditions and control group. These are expected to differ in the expected directions under H1–H4. Additionally, where a beneficial effect was hypothesised, this was expected to be stronger for the HT compared to LT according to the incremental effects of transparency (Grimmelikhuijsen and Klijn 2015; Kanol 2018).

Variable operationalisation

To measure trust in interest groups, I rely on a battery of post-treatment questions which ask respondents to rate perceived honesty, competence and legitimacy of business groups and NGOs on a 5-point scale. I follow the approach taken by Levi and Stoker (2000) and Grimmelikhuijsen and Meijer (2014) which describe honesty and competence as key components of political trustFootnote 9 and to this I add legitimacy,Footnote 10 defined as ‘the extent to which members of a political system believe that the authorities and structures are adequate to meet the members’ own expectations as to how the political system ought to behave’ (Fraser 1974: 118), of which trust is recognised as a ‘political component’ (Weatherford 1992).Footnote 11 The scores of these three survey items are combined where each component carries equal weight and standardise it so that it varies from 0 to 1 (αtrust in business groups = 0.51; αtrust in NGOs = 0.76).Footnote 12

To measure perceived corruption defined as the perception of “illegal, or unauthorised, profiteering by [officials] who exploit their positions for personal gain” (Blackburn et al. 2004: 5), I rely on three survey items, which asked respondents to rate, on a 5-point scale, the extent to which they believe that business groups and NGOs in turn “engage in illegal lobbying”, “disclose all information about their activities” and “favour private interests”. The first question captures the perception that interest groups act according to existing lobbying regulations. The second relates more specifically to their willingness to be transparent. The final one taps the perception that these actors represent special interests at the cost of public interests. Taken together these capture the perception whether lobbying takes place outside the boundaries of the law regulating transparency (shown in the treatment) for private grain. Like for trust, I combined these three items into an index (αperceived corruption-business groups = 0.58; αperceived corruption-NGOs = 0.62).Footnote 13 These two indices are the dependent variables in the analysis (for NGOs and business groups separately and then combined into an aggregate index considering both interest group types).

To measure the moderating factors of political interest and political ideology, I rely on pre-treatment items which ask respondents to rate how often they engage in debates about politics with friends and family (political interest: standardised 0–1; 0 = never, almost never; 0.25 = rarely; 0.5 = occasionally; 0.75 = most of the times; 1 = almost always) and their self-placement on a left–right scale of political ideology following standard approaches (political ideology: standardised 0–1 whereby values close to 0 indicate left, 0.5 indicates centre and 1 indicates right). The exact wording for these items is shown in the Online Appendix 2 ‘Description of Variables Used in the Study’ and Table A1 in Appendix 3 ‘Summary Statistics and Balance Test’ summarises key statistics for the variables of interest.

A first balance test shown in Table 1 suggests that random allocation to the experimental conditions was successful with very similar averages for covariates between treatment and control groups. Additional balance test including demographics as covariates (such as gender, age, income, education) are shown in the Appendix (Table A2 in Appendix 3 – Summary Statistics and Balance Test) and further demonstrate that randomisation was successful.

Table 1 Summary statistics (average and standard deviation) per experimental condition (balance test)
Full size table

I test the study’s hypotheses using regression analysis which allows for interpretation of substantive effects as well as interaction effects.

Results

I first test H1 and H2 by regression analysis in which I predict levels of trust and perceived corruption for interest groups at the aggregate level and then separately for business groups and NGOs with experimental treatments. Results are shown in Table 2.

Table 2 Results of OLS regression with robust SEs in parenthesis
Full size table

The left-hand side of Table 2 (Models 1–3) shows the results of the models where trust in interest groups (aggregate, business groups and NGOs) is the dependent variable. The right-hand side of Table 2 (Models 4–6) show the results of the models where perceptions of corruption of interest groups (aggregate, business groups and NGOs) are the DV.

All results suggest the absence of a significant relationship between lobbying transparency and attitude changes towards interest groups.Footnote 14 This rejects H1 which predicted that transparency would improve trust in interest groups, and that higher transparency would do so even more. These results also reject H2, since transparency does not significantly improve trust in NGOs to a larger and significant extent than in business groups.

While these results suggest that treatment and control group are not statistically different in terms of levels of trust and perceived corruption of interest groups, it does not imply that the two populations are similar. To further investigate the plausibility of a ‘nil’ hypothesis (rather than ‘null’), I performed an equivalence test which helps improving the plausibility of a falsified hypothesis by assessing the absence of a meaningful effect (Hartman and Hidalgo 2018). The results of TOST tests, which use t-tests to test the hypothesis of equality between two means (Lakens et al. 2018), suggest that the control and treatment groups (LT and HT) are statistically equivalent for smallest effect sizes of interest (using the same scale of the DV) of 0.10 (on 0–1 scale) for trust in NGOs, and at 0.05 for trust in business groups and at the aggregate level (levels of 0.04 and 0.03 are inconclusive while only at 0.02 equivalence is generally rejected). This suggests that, with this data, if lobbying transparency had an effect on trust in interest groups, then it would be as small as 0.02 with high degree of certainty. For perceived corruption, control and treatment groups (LT and HT) are statistically equivalent for smallest effect sizes of interest (using the same scale of the DV) of 0.10 for corruption of NGOs, at 0.05 for corruption of business groups and at the aggregate level (levels of 0.04 and 0.03 are generally inconclusive for both LT and HT while only at 0.02 equivalence is rejected). Whether or not these effects are small enough to be considered negligible or meaningful is open to debate (Rainey 2014). Nevertheless, these findings may be a first indication that lobbying transparency does not impact citizen’s attitudes towards interest groups as transparency research would have predicted.

To provide further robustness to these null findings, the results of the interaction effects which test the moderator role of political interest and political ideology are presented. As Fig. 1 shows, none of the hypotheses (H3 and H4) are confirmed. For ease of comparison only the results of the effect of LT on attitudes towards business groups are shown, but the same results are found for the effect of HT also when the DVs are attitudes towards NGOs and at the aggregate level (See Table A3 in Online Appendix–4. Analysis of Interaction Effects). The effect of the treatment (LT) is flat and statistically indifferent from the control group (dashed line) at all levels of political interest (high to low) and political ideology (left to right).

Fig. 1
figure 1

Marginal effects of low transparency in lobbying on trust and perceived corruption of business groups by levels of political interest and political ideology. Note Based on models 2, 4, 6 and 8 in Table A3. CI 95%

Full size image

These null findings provide further support for the null effect of transparency on attitudes towards interest groups, at least in this context. However, it remains unclear why transparency in lobbying did not have the expected direct effect on attitudes in the same way transparency in government and local public administration was found to have on citizens’ evaluations in experimental political science research. I reflect about the potential reasons for this null finding next. To provide a first possible explanation, I present proxy evidence drawn from of open-ended manipulation-check questions asked to respondents after the lobbying transparency treatment.

Possible explanations to null results

As manipulation check and to collect additional information about the impressions respondents formed around the disclosed information, the survey asked each respondent to answer an open-ended question immediately after the treatment. The question was ‘what was the article about, and did you find it interesting?’ and was answered by all 377 respondents that were assigned to the lobbying transparency treatment. Very few left the answer blank or simply wrote “yes”, “no” or “did not have time to read it [referring to the vignette]” (24). The vast majority of respondents engaged with the content of the vignette reporting more or less elaborate details about what they read and noticed/remembered.

Here are some observed trends. Firstly, only 1 out of 377 respondents has mentioned SIPO (the enforcement agency) or anything related to the register possibly suggesting that the information about the register itself and about SIPO in the vignette has not influenced attitudes in an obvious way.

Secondly, among the respondents assigned to the HT condition, many (that is 42 respondents—21 percent of the respondents in HT) commented on the lobbying expenditures specifically. Some of the answers expressed cynicism towards money in lobbying: ‘Why is so much time and money spent lobbying?’ or ‘I would believe they [interest groups] spend much more’ or ‘nothing but money’ or ‘lobbyists trying to lawfully influence or bribe politicians’ or ‘[it shows] lobbying expenditures […] I think lobbying should be illegal’.

Thirdly, some answers point towards what transparency scholars have called the knowledge-based mechanism (Grimmelikhuijsen and Meijer 2014), by which information disclosure provides new information to information seekers. This is most obvious in the following answers: ‘I didn’t know this information was available’; ‘I didn’t know much about lobbying’; ‘it’s quite informative’; and it made them realise ‘how little an average a person knows’ about these complex political processes.

Finally, in line with the effects of transparency explained by motivated reasoning, two negative reactions to lobbying transparency were found: ‘I am a bit cynical about lobby groups really’ and ‘this is why I do not trust lobby groups’.

The absence of a moderating effect of political ideology was also surprising. The word lobbying—as already discussed—carries negative connotations and is, in the view of the public, typically associated to the political activities of business actors but not to those of NGOs. As a handful of respondents declared in [what I interpreted as] surprising tone: ‘also charities employ them [refers to lobbyists]’; ‘I am amazed that charities spent […] on lobbying’; ‘Interesting that there's almost an even split of business and NGO lobbying’. However, only a handful of respondents commented on group categories specifically, which is not enough to infer anything significant.

While interesting for illustrative purposes, none of these instances are helpful to understand why, on average, no direct impact of lobbying transparency on attitudes towards interest groups was found. Perhaps more relevant for the null finding in the statistical analysis, is the fact that 102 respondents (27% of respondents pooling LT and HT together) rated the disclosed information as uninteresting. Moreover, answers show that respondents struggled to understand or were unfamiliar with the information about lobbying. Some confused lobbying expenditures with ‘funding received from the state’; one respondent confused lobby groups with political parties; some declared that they found the information ‘complicated’ and that they were not aware that they did not know ‘that lobbying was happening in Ireland’. As a result, low interest and/or low understanding of the lobbying may have reduced the respondents’ motivation to engage with the disclosed information and update their evaluation of interest groups accordingly.

This could explain why no moderating effect for levels of political interest was found. In past research, levels of political interest measured trough the frequency of engagement in discussions about politics, reading of newspapers etc., were found to moderate the effect of government openness on trust in government and other institutions (Schmidthuber et al. 2021). This measure of political interest however may not be a good proxy for interest in a ‘more niche’ process such as lobbying.

There are, however, other possible explanations to the null findings that are worth considering, some of which relate to the use of experiments to test transparency effects. Some have argued that the treatments employed in single experiments like this may not be powerful enough to generate substantive meaningful effects (Rainey 2014). This is a fair criticism that questions the expectation that transparency will have straightforward direct effects on citizens’ attitudes, against the argument that transparency is learned, socialised to, through repeated interaction and exposure (Naurin 2007; Lindstedt and Naurin 2010). Experimental research in this area has however demonstrated that information disclosure can produce immediate effects on citizens’ attitudes (Porumbescu et al. 2022). This research however focuses on institutions and political organisations citizens are more familiar with. This could add strength to the observation above about how ‘niche’ lobbying may be for many citizens. The following concluding section discusses the implications of these considerations for future research on this topic.

Discussion

This paper experimentally investigated the effects of lobbying transparency on citizen’s attitudes towards interest groups, more specifically, levels of trust and perceived corruption of business groups and NGOs. The experiment manipulated information generally found on lobbying registers to test the effects of (different levels of) transparency on attitudes. The experiment involved 569 participants divided between two treatment groups (low and high transparency) and a control group and were part of a larger panel of participants involved in a pre-registered study on the effects of political transparency.

Building on the transparency literature as well as the literature on interest groups, this study expected that transparency in lobbying would improve trust and reduce perceptions of corruption, and that such effect would be stronger when evaluations are directed at NGOs relative to business groups. Additionally, the effect of transparency was expected to be moderated by levels of political interest and political ideology.

The results of the study do not confirm these expectations. Levels of trust and perceived corruption of NGOs and business groups of respondents exposed to lobbying transparency (both at low and high levels) were not statistically different to the attitudes of participants assigned to the neutral baseline of the control group. Further tests of equivalence, analysis of separate items forming the DVs of the study, and potential moderating factors, add robustness to this null finding.

In search of an explanation for such null finding, the analysis of open-ended questions answered by participants in the treatment groups revealed that low interest and low understanding of the disclosed information about lobbying could be a possible reason for why no direct effect of transparency was found. If these illustrations are accepted as plausible, then what previously found in experimental research on transparency may not apply to lobbying transparency.

If transparency in local government, political donations, public administration, and conflict of interest regulation was found to have direct beneficial effects on citizens’ attitudes in past research, this experiment offers an example of a null result for transparency in lobbying. Regulators, international organisations and advocates of lobbying transparency may therefore be careful about promoting lobbying registers as a straightforward tool to improve citizen’s trust in politics and perceived procedural fairness in policymaking. When reformers say that lobbying regulation will ‘help promote a more positive perception’ of the role of lobbyist (Mahon Tribunal 2012: 2529) they need to be aware that such goal is more difficult to achieve than generally assumed. It may, for example, have to be mediated by some other agent (such as the media) that reports the disclosed information (Lindstedt and Naurin 2010; Crepaz and Kneafsey 2021). The role of such agent, however, may adopt framing of the information that is different from the neutral one used here to test the effects of transparency.

With this idea in mind, one could conclude, as far as direct effects of transparency are concerned, that, if they exist, they are likely to be circumstantial, related to specific disclosed information (de Fine Licht et al. 2014) and to the way it is framed and circulated by transparency users (Lindstedt and Naurin 2010). To investigate these using survey experiments, it is likely that treatments will have to be narrowed down to specific instruments and magnified to observe potential effects (Rainey 2014). More specifically, scholars may wish to experimentally manipulate access data on lobbying registers, showing—for example—disproportional access advantages of business groups vis a vis other actors, or manipulate the use of financial resources they make. Manipulating lobbying targets could also be prolific with lobbying the apex of government leading to different perceptions than legislative or grassroots lobbying. Also, manipulation in the use of the word ‘lobbying’ could be explored with the expectation that, combined with certain disclosed information, it would provoke more negative attitudes than, for example, ‘advocacy’. Finally, scholars may consider designing cues for ‘missing’ or ‘incomplete’ information disclosure to test the impact of low transparency on political attitudes such as cynicism and distrust. To maximise causal inference of these potential studies, these features would have to be considered separately, an aspect that has not been done here for reasons related to external validity and mundane realism that has, however, also its shortcomings.

With all these considerations in mind, the fact that transparency is no panacea to distrust and perceived corruption should not be surprising (see already Hood and Heald 2006). To be clear, I am not denying the usefulness of lobbying regulation. Transparency in lobbying was found to ‘civilise the behaviour’ of elite actors (Naurin 2007), support competition in lobbying (Crepaz 2020) and journalists in their profession (Crepaz and Kneafsey 2021). Moreover, individuals aware that lobbying is (heavily) regulated are found to have higher trust in government institutions and lower perceptions in the widespread of corruption (Kanol 2018). However, when benefits of improved trust and reduced perception of corruption are assumed to be directed at interest groups themselves, this experiment finds no evidence of them. It is nevertheless important to stress that, despite the absence of such beneficial effect, transparency in lobbying did not appear to backfire (not even when lobbying expenditures were disclosed). That is, when exposed to information about lobbying, citizens—that are assumed to view lobbying as representation of ‘special interests’—did not react adversely, like some pessimists would have expected.

The results of this study may of course not translate directly into real world scenarios. Frist, citizens do not necessarily experience lobbying transparency exactly in the forms presented in this experiment. Secondly, information disclosure here is once-off ignoring the potential long-term socialising effects of transparency in politics (Naurin 2007; Lindstedt and Naurin 2010). Nevertheless, this study provides initial evidence that lobbying transparency may be more complicated to understand than what transparency research would assume.

Equally, one may question whether the results of this experiment apply to contexts outside of Ireland. I believe they do, especially when in many European countries lobbying registers are becoming more common and lobbying practises are becoming more similar through the professionalisation of the industry (Woll 2012; Chari et al. 2020). While lobbying registers exist also across the ocean, the fact that lobbying in, e.g. North America, may be experienced differently by citizens could limit the external validity of these results. Replicating this study across US states or Canadian provinces where lobbying transparency varies substantially may be a suitable hard test for these findings.

There are, nevertheless, (at least) two tentative lessons that could be drawn from these first results. The first relates to policy and concerns the design of lobbying regulations themselves. If lobbying registers want to be a service for citizens, then more needs to be done to design them as interactive tools for political engagement and learning, rather than mere lists of policy stakeholders. Perhaps, following the example of open data portals, interactive tools and easy-to-read summary statistics could be a cheap but helpful tool to better engage the public.

Secondly, more could be done from the perspective of journalists and civil society actors to provide a picture of interest groups that does not portray them as ‘special interests’. This image of lobbying may be one of the reasons why citizens do not engage with lobbying. The work of some NGOs who help promoting the picture of lobbying as legitimate and healthy form of political participation is beneficial in light of these results.