Framing and advocacy: a research agenda for interest group studies

ABSTRACT This research agenda contribution starts from the observation that an increasing number of interest group studies have been addressing questions about framing. Although this emerging literature has made great progress towards being able to study interest group framing in large-n designs owing to advances in data-gathering techniques, many analytical and conceptual challenges still lie ahead. One important question that remains is how framing can serve as a political strategy and, more precisely, which frames are most effective. This article gives an overview of the recent work on interest group framing. It highlights some key issues that interest group scholars face when they undertake research on framing. Various studies on interest group framing are contrasted in terms of the types of frames studied, the level of analysis employed and how influence is determined. I conclude by developing an agenda with some concrete recommendations for interest group scholars that deal with questions about framing.


Introduction
Lobbyists and interest groups are framers (Baumgartner 2007: 486). Since Schattsschneider (1960), interest group scholars have been fascinated by the arguments and issue definitions interest groups promote and how this affects their policy influence. Recently, an increasing number of interest groups scholars have been explicitly relying on the concept 'framing' or the idea that interest groups strategically communicate and promote arguments and issue-definitions in order to influence policy decisions in the desired direction Baumgartner and Mahoney 2008;Boräng et al. 2014;Boräng and Naurin 2015;Eising, et al. 2015;Klüver Mahone and Opper 2015a). Some scholars in this literature have argued that studies on framing in interest group politics needs to move towards large-n designs, so inferences can be made across different actors and policy debates (Baumgartner 2007: 437;Baumgartner and Mahoney 2008;Klüver and Mahoney 2015: 225). Moreover, various contributions to a recent special issue in this journal have demonstrated that interest group frames strongly vary depending on the context of policy cases (Eising, et al. 2015;Klüver Braun and Beyers 2015;Klüver, Mahoney and Opper 2015).
Studying interest group framing in large-n research designs has faced important methodological obstacles, such as the identification of frames in a reliable and valid way. To cope with these types of obstacles to large-n research designs, scholars have been developing a number of new methods and empirical techniques (Boräng et al. 2014). For instance, Klüver and Mahoney have proposed a new method based on automated-text analysis to identify and scrutinize interest group frames (Klüver and Mahoney 2015). Although the recent scholarship has strongly advanced in methodological terms, many conceptual and analytical challenges nevertheless still lie ahead for interest group scholars dealing with questions about framing.
Framing is studied in different research disciplines, including communication studies, political science, psychology and sociology. There is no clear consensus across these disciplines on how framing should be studied (Dewulf et al. 2009: 156;Scheufele and Iyengar 2012). The fractured and diffuse nature of framing studies brings about some demanding and complex challenges for interest group scholars who engage in this type of research. Most importantly, researchers draw on different types of frames, analytical frameworks and methodological approaches which impedes cooperation and convergence between the different studies. In the spirit of strengthening the cumulative development and the dialogue of the interest group literature on framing, this article takes a first step towards conceptual and analytical convergence in the field.
This research agenda contribution addresses a number of conceptual and analytical challenges that face interest group scholars in framing research and how to deal with them. Different studies on interest group framing are contrasted in terms of the types of frames studied, the level of analysis employed and how influence or success is determined. The article starts by addressing broader questions in framing studies which are then specified for the interest group literature. I conclude with presenting a research agenda in which concrete recommendations are made for future research on interest group framing.
What's in a frame?
One definition of framing that is very broadly accepted and probably the most extensively cited is that by Entman. Entman (1993: 52) suggests that: to frame is to select some aspects of a perceived reality and make them more salient in a communicating text, in such a way as to promote a particular problem definition, causal interpretation, moral evaluation, and/or treatment recommendation for the item described.
Although other definitions are employed, Entman's definition is widely used by interest group scholars (e.g., Boräng and Naurin 2015: 501;Eising et al. 2015: 521;Klüver, Mahoney and Opper 2015a;Kurzer and Cooper 2013: 730;Voltolini 2015: 3). The definition aligns with the conception of lobbying as a communication process (De Bruycker 2015;Milbrath 1960). Drawing on this conception, framing should be seen as an intentional and strategic process, part of the communication by interest groups. It is assumed that interest groups strategically deliberate about which frames are best suited to promote their policy goals and ideas.
While most interest group studies adopt Entman's definition of framing, there is still much incongruence between these studies in terms of the types of frames they address. These different types of frames each represent different phenomena. Since definitions of framing leave much manoeuvring space about what framing constitutes, it is crucial to be very explicit about which types of frames are of interest (Cacciatore et al. 2016). Two common distinctions are those between issue-specific and generic frames, and emphasis versus equivalence frames.

Issue-specific and generic frames
Issue-specific frames are tied to the specific nature of the issue or conflict under scrutiny and emerge by looking from the bottom-up. Generic frames are not tied to a specific policy debate or issue, but can be identified across issues (de Vreese 2005: 54-6). The study of  on the framing of the death penalty debate in the United States (US) is a prominent example of a study in the interest group literature that relies on issue-specific frames. They show that the morality frame long dominated the death-penalty debate in the US, but that it has been replaced by the innocence frame and that this has coincided with important changes in public opinion and policy. Other notable studies on interest group framing that rely on issuespecific frames are Kurzer and Cooper's (2013) study on food information labelling, Candel et al.'s (2014) study on the Common Agricultural Policy (CAP), Voltolini's (2015) study on Israel-EU relations and Agustín's (2008) study on gender equality. These studies all define frames via a bottom-up approach, looking at one particular policy debate or field and the role of organized interests therein. For case studies, these issue-specific frames serve as descriptive devices to unveil what is at stake in an issue and how this evolves through time and space. One of the limitations of issue-specific frames is that they are less suitable for studying interest group framing across different policy issues and areas. For instance, the morality and innocence frames from the study by Baumgartner and his colleagues (2008) are not applicable to a policy debate about consumer safety, biotechnology or gender equality.
Instead, generic frames are applicable across different policy areas and settings. An example of the use of generic frames in the interest group literature is the study of Boräng and Naurin (2015) who distinguish between self-, other-, public-and ideal-regarding frames (see also Naurin 2007). Another example is the study of Klüver Mahoney and Opper (2015), who distinguish between economic and public frames. The use of generic frames by interest group scholars has been increasing and this is a logical result of the explicit ambition to engage in large-n studies, which tackle framing questions across different policy issues and areas (Baumgartner and Mahoney 2008;Eising et al. 2015;Klüver and Mahoney 2015).

Equivalence and emphasis frames
Another important distinction is between equivalence and emphasis frames. Emphasis framing refers to emphasizing one aspect of an issue over others (Druckman 2004: 672). All the aforementioned examples of generic and issue-specific frames are also emphasis frames. One problem when studying emphasis frames as part of a lobbying strategy is that it is very difficult to untangle the actual framing effect from other factors. Emphasis frames are often endogenous to the underlying interests and positions represented. For example Candel et al. (2014) show that in the EU policy debate on the CAP reform, producers operated from a 'productionist frame', nature organizations from an environmental frame and local authorities from a regionalist frame. Frames are, thus, often a reflection of interests rather than the strategic presentation of a message. This is one of the reasons why Scheufele and Iyengar (2012) plead for a stronger empirical focus on equivalence frames, as for these types of frames it is more straightforward to distinguish between the framing effect and other factors that might be at play, such as the underlying interests represented or the types of information communicated.
The idea of equivalence framing was developed in psychology research and involves presenting similar information in a different way (Druckman 2004;Kahneman and Tversky 1979). All equivalence frames are also generic frames, as they are applicable across different policy issues. Equivalence framing also involves emphasizing certain aspects of an issue in a communicative message, but the scope of emphasis is much narrower than with emphasis frames. One can distinguish between three types of equivalence frames: 1. opportunities versus risks; 2. gains (benefits) versus losses (costs); 3. positive consequences versus negative consequences.
Equivalence frames are relevant for interest group scholars as they match with the argumentation patterns interest groups use. For instance, Baumgartner et al. (2009: 132) analysed a range of frames and arguments that were used by interest groups across a sampled set of 98 policy issues. They found that interest groups in the US regularly adopt arguments related to the costs or benefits of a specific policy issue. Studies on equivalence framing typically do not focus on public policy-making and political élites, but rather on shifts in public opinion or the media discourse. Although potentially relevant, little application of the idea of equivalence framing can be found in interest group and policy studies.
What still remains unexplored is whether some equivalence frames that are used by interest groups are more influential than others. The effects of equivalence frames are grounded in prospect theory. Namely, frames emphasizing costs, risks or losses are expected to render stronger effects on attitude and behaviour because people are more apt to take risks to avoid a loss than to achieve a gain (Tversky and Kahneman 1981). The focus on equivalence frames thus comes with a clear theoretical rationale interest group scholars can draw from when developing expectations and apply to lobbying processes.

Two faces of framing
Interest group studies on framing not only differ in terms of the types of frames they address, but also in terms of the units of analysis they employ. Some studies are actor-centred, and focus on the frames different interest groups voice. Other studies are more policy-centred, and look at the different frames prevalent in a policy debate. Thus, we might distinguish between two 'faces' of framing: macro-level and micro-level framing (Baumgarter and Mahoney 2008).
First, macro-level framing is seen as the collective definition process of what is at stake in a policy debate (Daviter 2011: 2). Framing is then the process of how a policy debate is defined and understood (Dudley and Richardson 1999: 226). From this perspective frames are seen as the building blocks that construct a policy debate, rather than an emphasis in communication as in Entman's definition. A frame is then a perspective from which a policy problem can be made sense of and acted upon (Rein and Schön 1991: 265). Frames are identified by looking at the overall dominating aspects of a policy debate that are emphasized by institutions, advocacy coalitions and the news media (Baumgartner and Mahoney 2008). From this point of view framing is both a bottom-up process, where different sides of a policy debate promote their own frames, and a top-down process, which structures conflict and mobilization patterns. The definition of what is at stake in policy debates, or the dominant frames, are shaped by, but also structure the formation and mobilization of interests, which may restructure political constituencies in the process (Daviter 2011: 46-7). In this way, frames are both seen as tools of political actors to obtain a certain outcome and as societal forces structuring policy and how interests mobilize. All the studies on issue-specific frames mentioned earlier are also studies that look at framing by interest groups at the macro level of policy debates (Agustín 2008;Daviter 2011;Kurzer and Cooper 2013;Voltolini 2015).
Second, interest groups can also be taken as the central unit of analysis, namely in micro-level framing studies. With this approach, frames are seen as instruments of change or as strategic tools that interest groups rely on to obtain their political and policy goals. The studies on generic frames mentioned earlier all approach framing from the level of interest groups (e.g., Eising et al. 2015;Klüver, Mahoney and Opper 2015). These studies have been successful in producing generalizable findings about which frames are applied by whom and under which conditions; yet, little is known about which frames are successful advocacy tools and which frames are more influential than others.
As pointed out by Baumgartner and Mahoney, both micro-and macro-level perspectives to framing are limited since they only capture one aspect of the framing process: Studying the process of framing only at the individual level has little chance of elucidating collective-level changes in framing. At the same time, researchers focusing only on aggregate-level framing will be unable to understand the forces that led to the collective frame without recognizing the micro-level forces that are at play. (Baumgartner and Mahoney 2008: 436) What is still missing is an analysis that links the framing efforts of interest groups to the macro-level of policy debates. Although the macro-and micro-level framing perspective is contrasted in the former discussion, there are potential pathways to link them (see also Baumgartner and Mahoney 2008 445). One pathway for doing so is studying interest group framing in the media. While recently there has been a surge of studies on the presence of interest groups in the news media (Bernhagen and Trani 2012;Binderkrantz et al. forthcoming;De Bruycker and Beyers 2015;Tresch and Fischer 2014), none of these studies focus on interest group framing. Nonetheless, studies on interest group framing in the media have the potential to look at the frames individual interest groups communicate in the news and at the same time determine which frames come to dominate the overall media debate; and thus link micro-and macro-level framing.

Framing for influence
The most compelling question for interest groups scholars who engage in framing research is probably how frames relate to 'the holy grail' of interest group studies: influence (Leech 2010: 534). In interest group studies, frames can serve either as means to define influence or as external factors that can affect influence. In other words, frames can be understood as either endogenous or exogenous to influence. In this section I clarify both perspectives and how they are linked to the former discussion.

Macro-level influence
From a macro-level perspective, framing effects are observed by examining the features of a policy debate and how these evolve through time as new frames emerge to dominate the policy debate. By framing the different sides to a debate, interest groups and other stakeholders can affect the collective definition process of an issue which may translate into policy change. Most of the time one single interest group will have little chance of reframing an entire policy debate. Baumgartner et al. (2009) have conductedto my knowledgethe only large-n study on interest group framing at the macro-level across a large and diverse set of cases. They indeed find that most policy debates in Washington cannot be reframed and that dominant policy frames are rather stable (Baumgartner et al. 2009: 186). This, however, does not mean that an interest group cannot use framing as a valuable lobbying strategy to achieve its policy objectives. An interest group may not succeed in reframing an entire policy debate at the macro-level, but it can still achieve other policy goals, such as convincing policy élites or increasing the public visibility and media attention for their cause.

Interest group-level influence
While interest group influence is thus mostly not observable at the macrolevel of policy debates, it might be when looking at the extent to which a single interest group or advocacy coalition achieved its policy goals through framing strategies. For instance, a lobbyist may realize some of its policy objectives behind the scenes, without reframing a policy debate substantively. Moreover, the lion's share of policy cases are not politicized and lack a vibrant public debate. Most policy cases are characterized by low public salience and only see a few interest groups mobilizing (De Bruycker and Beyers 2015;Van Aelst et al. 2015). For many policy cases it will thus be problematic to identify a full-fledged policy debate, whilst influence might have been exerted through framing strategies. Rather than looking at the characteristics of policy debates, influence can be observed at the level of interest groups through methods such as preference attainment, attributed influence or process tracing (e.g., Dür 2008). One important advantage of these measures is that framing can be seen as exogenous to influence, as an independent factor which explains influence. This avoids a situation where frames are both conceived of as dependent and independent variables.

Individual-level influence
A limitation of exiting measures of interest group influence is that they are strongly focused on policy outcomes. An alternative way to assess the influence of framing strategies is to consider the political behaviour and attitudes of individual citizens and policy élites (such as civil servants, politicians and policy experts) as outcome variables. Here, the mechanisms that trigger change are cognitive: when individuals are exposed to a certain frame by an interest group, this frame could become more salient in their minds. This increases the chance that individuals who have been exposed to the frame will consider its aspects when making a decision. Getting a frame prominent in the minds of citizens or policy élites then constitute an intermediate steps towards affecting policy outcomes or getting a frame to dominate an entire policy debate. Framing effects on public opinion have been extensively studied by political behaviour scholars (Druckman 2001;Nelson et al. 1997), but the same causal mechanisms may equally be applicable to policy élites. For example, an interest organization that strongly opposes a particular policy proposal may emphasize the negative consequences and risks of the proposal when contacting a policy-maker. By doing so, these negative consequences and risks may grow more salient in the mind of the policy-maker, which increases the chance that she will consider them when making policy decisions.
One important advantage of looking at framing effects at the level of individuals is that it enables controlling for characteristics of the frame receivers. Numerous studies on framing effects in political psychology have already demonstrated that the effects of framing are strongly moderated by the predispositions and characteristics of the receiver (e.g., McLeod and Shah 2014;Nelson et al. 1997). Some empirical work on interest group framing has also found that the frames interest groups voice vary according to the type of policy-maker that is targeted. For instance, members of parliament are generally approached with different frames than civil servants (Baumgartner and Mahoney 2008;De Bruycker 2015). Different citizens and policy élites are thus likely to be receptive to different types of frames, depending on their personal predispositions, political ideology and institutional affiliations.

Discussion: towards a research agenda for interest group studies
This research agenda contribution started from the observation that an increasing number of interest group studies have been addressing framing questions. Although this literature has made a great deal of progress in its ability to study interest group framing in large-n designs through advances in data-gathering techniques, many analytical and conceptual challenges still lie ahead. One important question that remains is how framing can serve as a political strategy and more precisely which frames are most effective. A more systematic understanding of framing strategies and its effects could aid in answering questions related to interest group influence and the nature of interest representation more broadly: can interest groups win their political battles by voicing the right selection of frames?
Questions regarding the political influence interest groups can have by using particular frames are not new (Milbrath 1960;Schattschneider 1960), but they have only been sparsely addressed empirically across a large and diverse set of policy cases. The contemporary interest group literature possesses the data and the methodological tools to do so. At this point, not methodological and empirical limits, but rather the heterogeneity and diffuse nature of the framing literature is the biggest challenge for interest group scholars that are dealing with questions about framing. In this article, different studies on interest group framing have been contrasted in terms of the types of frames used, the level of analysis employed and how influence or success is understood. Drawing from the insights developed in this paper, I now specify four recommendations for future studies that tackle questions related to interest group framing: First, when studying framing, it is crucial that the types of frames studied, the level of analysis and the research purposes are compatible and explicitly spelled out. Although this is not an absolute rule, studies pursuing external validity align better with a focus on generic frames, studied at the micro level of individual interest groups. This research strategy is ideal for large-n designs that seek to produce generalizable results across different policy cases. One caveat is that these rather crude measures of framing disregard a lot of detailed qualitative information about policy cases, which may be detrimental to the ecologic validity of the results. Issue-specific frames, by contrast, are typically studied at the macro level of policy debates and are ideally (but not exclusively) suited for understanding influence and strategies for one specific policy case. This research strategy provides in-depth and qualitative knowledge about how policy is understood and unfolds, although it is often limited in producing findings generalizable across a wide array of policy cases.
Second, when studying interest group framing and influence across a large set of cases, it is recommended to focus on equivalence frames rather than emphasis frames. Emphasis frames are often a reflection of the interests or constituencies represented (Scheufele and Iyengar 2012). For instance, farmers voice an 'agricultural' frame and environmental groups an 'environmental' frame, not because of strategic purposes, but because of the interests they represent (Candel et al. 2014). To capture strategic framing and to disentangle it from the underlying interests represented, equivalence frames are more appropriate, since they can be used by any interest group, regardless of the particular interests represented. For example, both farmers and environmental groups may (strategically) emphasize the risks or opportunities of different policy alternatives. Moreover, we have well-informed expectations about the effects of equivalence frames, grounded in prospect theory. Importantly, this approach has the cost of losing detail, but it allows us to draw generalizable conclusions about which frames are successful and why.
Third, when interested in the impact of framing as a political strategy, it is recommended to look beyond the macro level of policy debates. An interest group may often fail to reframe an entire policy debate at the macro level, but it might still be successful in achieving some other objectives, such as convincing segments of public opinion or some policy élites; or getting some attention in the media. Moreover, there is often not really a sizable policy debate to influence, since for many policy processes only a few stakeholders mobilize and little controversy and debate materializes (De Bruycker and Beyers 2015;Van Aelst et al. 2015). As argued earlier, an alternative pathway to assessing the influence of a certain frame is to consider the political behaviour and attitudes of individual citizens and/or policy élites as outcome variables. One important advantage of doing so is that this enables controlling for the moderating effects situated at the level of the frame receiver, such as personal predispositions, political ideology and institutional affiliations. A focus on individual-level framing effects implies the use of (survey) experiments in which subjects are exposed to certain interest group frames.
Finally, one broad and ambitious research agenda for studies on interest group framing is to link the individual and macro-level perspectives (see also Baumgartner and Mahoney 2008: 45). How do individual-level framing effects spill over to the macro level of policy debates? Future studies could consider interest group framing to have effects at the level of public opinion and policy élites, as an intermediate step, which in turn may shape how policy debates are defined at the aggregate level. As suggested earlier, another way to move forward is to study the frames that individual interest groups voice in the news and link this to the frames that come to dominate media debates. This leads interest group studies back to classic Schattschneiderian questions: can a single interest group affect the size and scope of a conflict, and under which conditions can societal interests have an impact on how policy debates are defined and understood? Tackling these research questions empirically is feasible with large collaborative research projects that can rely on a rich body of evidence. Considering the recently completed collaborative research projects on interest group politics (such as INTEREURO and INTERARENA), research efforts that link the framing strategies of interest groups with policy élites and ultimately with the scope, size and nature of policy debates might already be feasible in the course of the coming years. 784 I. DE BRUYCKER