Synergies in lobbying? Conceptualising and measuring lobbying coalitions to study interest group strategies, access, and influence

Treating interest groups mainly as independent units of observation overlooks highly frequent coalition activities between actors and risks affecting the results of studies of lobbying and political influence. Yet, conceptualising and measuring lobbying coalitions is inherently difficult. In order to facilitate important future research, this article provides a roadmap of the main conceptual and methodological choices involved in studying lobbying coalitions. It distinguishes three main approaches to identify coalescing actors: a preference similarity approach, a behavioural approach, and an organisational approach. The article presents concrete operationalisations of coalitions from these vantage points and provides empirical evidence that various forms of cooperation activities on specific issues, as well as general cooperation structures, are highly frequent in lobbying in European countries. The article is relevant for scholars of interest groups and political advocacy more broadly by informing the design of new research on lobbying strategies, access, or influence.


Introduction
Central questions of politics revolve around who gets what, when, how (Lasswell 1950), and, one might add, the crucial element of 'with whom', be it on lobbying sides (Klüver 2013;Mahoney and Baumgartner 2015), in multiplex networks of influence (Heaney 2014) or in advocacy coalitions (Sabatier 1988). Interest group research both in the American and the European context seems to have reached consensus that lobbying in coalitions is a prominent influence strategy (e.g. Hanegraaff and Pritoni 2019: 207;Junk 2019a: 9;Mahoney 2007b: 379;Nelson and Yackee 2012: 339;Schlozman and Tierney 1986: 148). This literature has addressed the question of why lobbyists choose to join coalitions despite the costs involved in terms of autonomy, reputation, or even organisational survival (e.g. Beyers and De Bruycker 2018;Hanegraaff and Pritoni 2019;Hojnacki 1997;Holyoke 2009Holyoke , 2014Hula 1999;Mahoney 2007b;Sorurbakhsh 2016). Moreover, it has begun to document beneficial effects of coalition lobbying on the attainment of policy preferences or (perceived) lobbying influence (Box-Steffensmeier et al. 2013;Heaney 2014;Heaney and Lorenz 2013;Junk 2019a, c;Nelson and Yackee 2012;Tallberg et al. 2015).
However, the findings in existing studies can be hard to compare and relate, because researchers work with a multitude of understandings of the concept of a 'coalition'. The verdict on how large a share of lobbying actors works in coalitions in the USA, for instance, can range from 7% (Baumgartner et al. 2009: 10) to 90% (Schlozman and Tierney 1986: 148), depending on whether one looks at formal, issue-specific coalitions or the general practice by organisations to use coalitions as an influence strategy.
In practice, lobbying cooperation takes many, crosscutting forms, be it ad hoc cooperation on specific issues or the institutionalisation of general ties between organisations. Without a clear terminology to communicate and compare different conceptions of coalitions under study, and theories that make explicit why this type of coalition is the most relevant notion to explain the political outcomes of interest, the literature on collective lobbying is likely to remain scattered and potentially puzzling. This is why this article seeks to clarify and systematise coexisting conceptions of coalitions and discuss their usefulness and limitations.
Promoting and easing the future study of lobbying coalitions in this way is of prime importance, because our understanding of how lobbying works and affects policy outcomes is likely to remain limited unless we take cooperation between actors into account. The study of lobbying influence, for instance, often focuses on single actors, irrespective of interactions that take place between advocates working towards the same goal. Such an approach is likely to overlook important synergies between cooperating actors stemming from resource pooling (Hula 1999;Strolovitch 2007) or the ability to manage complexity and interdependence in increasingly crowded policy spaces (cf. Baumgartner and Jones 2010). The flipside of this coin is that where there are such synergies, we might miss parts of the actual effects if we keep looking only at individual units rather than sets of cooperating actors. We might not see the forest for the trees, so to speak.
For these reasons, this article equips researchers who intend to include some form of cooperation in their analyses of political advocacy with a roadmap of the conceptual and methodological options for designing this research. It contrasts three main approaches used in existing studies to define a 'lobbying coalition', namely (1) a preference similarity approach, focussing on positional 'camps' or 'sides' on an issue, (2) a behavioural approach, based on observing cooperation activities by actors on specific policy issues, and (3) an organisational approach, identifying ties between organisations as general organisational characteristics. The article provides a conceptual discussion of these issue-specific and general approaches and suggests methods of data collection, as well as serving empirical evidence to gauge the prevalence of different forms of lobbying coalitions in selected European counties. The focus throughout the article herein lies on coalitions between 'political advocates', defined as interest groups, as well as firms and individual expert actors outside the political system who try to affect policymaking. 1 In addition, however, some of the conceptual and methodological tools in the article may be transferable to coalitions in social movements and minority activism (Gillion 2013;Staggenborg 1986;Van Dyke and McCammon 2010), as well as to broad conceptions of 'advocacy coalitions' (Sabatier 1988), which include exchanges between actors outside and inside the political system.

Approaches to studying lobbying coalitions
In this article, I argue that a more detailed classification of a lobbying coalition is helpful, if not necessary, to advance our understanding of the effects of collective lobbying in policy processes. My aim is to systematise the main differences between existing understandings and aid the selection of an approach and methods of data collection. To do so, the article contrasts what I call a preference similarity approach, meaning a mere positional definition of coalitions, with two approaches that require some closer form of interaction, either in terms of cooperation activities on a specific policy issue (behavioural approach) or organisational structures that generally institutionalise collaboration of an advocate with others (organisational approach).
I argue that these three approaches can extent to a variety of countries and political systems, as they highlight different focal points or necessary conditions for recognising a coalition in the dense and crosscutting webs of ties between political advocates that try to affect policymaking (cf. Heaney 2014; Heaney and Lorenz 2013). I expect the three features of positional commonalities, issue-specific cooperation activities, and organisational structures of cooperation to be present in collective lobbying across countries and contexts, although the frequency and effects of specific forms of cooperation may differ depending on demands by gatekeepers in the political system.
Notably, the approaches were drawn up against the background of existing work, especially in the United States (US) and European Union (EU) context. The empirical material I present from five Western European countries then illustrates that these concepts are applicable in corporatist and pluralist interest group systems and yield distinctive patterns of 'coalition lobbying', depending on which of them is applied. In the following, I outline these approaches in detail and discuss conceptual and methodological choices involved.

Issue-specific cooperation: the preference similarity and behavioural approaches
Both the preference similarity and behavioural approaches to coalitions are, in essence, issue specific. An advantage of these issue-centred approaches to coalitions is that they arguably allow testing their effects most immediately by relating the specific preference constellation and/or cooperation activities on an issue to respective political outcomes (holding organisational characteristics constant). The main difference between these approaches, however, lies in how inclusively the term 'coalition' is used for actors lobbying for the same policy preference on an issue.

Conceptual distinctions: shared goals and active cooperation on the issue
The preference similarity approach here applies the broadest definition of an issue coalition in the form of a positional 'camp' or 'side' of all advocates actively promoting the same policy position on an issue (Baumgartner et al. 2009;Klüver 2011Klüver , 2013Lorenz 2019;Mahoney and Baumgartner 2015). Both in the US and the EU context, these sets of like-minded actors are sometimes called 'coalitions' (e.g. Klüver 2011; Lorenz 2019).
Such lobbying camps exist on all issues where advocates have mobilised different positions on an issue. Yet, how many such camps one distinguishes on an issue will vary depending on how many distinct positions are advocated, as well as how detailed these positions are coded by the researcher. Measures characterising this positional 'coalition', such as (relative) camp size and camp resources (Baumgartner et al. 2009: 222;Klüver 2013;Mahoney and Baumgartner 2015) or the diversity of social and economic interests supporting or opposing a bill (Lorenz 2019), will be affected by the level of detail in this positional coding.
While some common denominator of shared preferences or common goals will arguably be a necessary condition for a lobbying coalition, some will question whether they are a sufficient condition for speaking about a 'coalition'. Actors on the same side 'may be working hand in hand, or they may not even know the other is working on the issue' (Mahoney and Baumgartner 2015: 205). A mere focus on preference similarity ignores this question of whether there was any active cooperation between the like-minded advocates. The behavioural approach to coalitions therefore adds a second necessary condition to speak of a coalition, namely that there needs to be some 'degree of coordinated activity ' (cf. Sabatier 1988: 139) in the camp of like-minded advocates.

Types of cooperation activities in the behavioural approach
As Fig. 1 summarises, I distinguish three important forms of cooperation activities between advocates in a lobbying camp on a policy issue, namely: (1) information exchange, (2) strategy coordination, and (3) concerted action in signalling coalitions.
Information exchange is likely to be a common-yet not costless 2 -form of interaction between like-minded advocates who seek to further common policy goals on an issue. This may span exchanging technical and political information, as well as strategic information on the lobbying tactics used. The literature on coalition formation highlights gathering information and aggregating political intelligence as one of the main benefits of coalitions (Heaney 2006;Heaney and Lorenz 2013;Hula 1999;Phinney 2017). Put differently, one can see it as an activity that defines a form of active cooperation on an issue. This is important to account for, because the success or failure of a lobbying camp vis-à-vis its opponents is likely to depend on whether it remains an uncoordinated 'side', or whether like-minded actors share relevant information with each other, perhaps especially so on complex and/or salient issues. Still, the boundaries of such an informational coalition can best be conceptualised as blurry (see dotted triangle in Fig. 1), given that information exchange is a matter of degree (in terms of frequency, quantity, or quality), so there is no clearly demarcated set of actors in the 'informational coalition'. 3 Instead, one can ask for each advocate how extensively s/ he exchanged information on an issue with other advocates and potentially aggregate this information to account for the level of information exchange within the camp.
A closer form of active cooperation between like-minded advocates is the coordination of strategies on an issue. This form of cooperation demarcates a distinct set of cooperating actors, namely including all those actors aligning their lobbying strategies with one another, for instance by dividing labour or concentrating efforts on agreed target audiences. More efficient goal attainment due to strategy coordination and resource pooling is named as a second main benefit of coalitions in the literature (e.g. Hula 1999;Mahoney 2007b;Strolovitch 2007: 175-205). Again, from a behavioural perspective on coalitions this can be seen as a defining activity to identify a strategy coalition. Where these tactics are explicitly coordinated between partners (rather than individually reactive), we can speak of strategy coalitions, composed of the set of actors collaborating actively on strategy use.
Finally, Fig. 1 includes cohesive signalling coalitions (cf. Hula 1999; Junk 2019c; Mahoney 2007b; Nelson and Yackee 2012; Phinney 2017) which explicitly join forces on the issue. These coalitions entail the most visible form of active cooperation on an issue, as coalition members implement strategies concertedly, such as by jointly approaching decision-makers or the public on a specific issue. An example is the initiative against shale gas extraction in the United Kingdom (UK) by a broad alliance of environmental groups, which delivered a petition to Prime Minister Cameron before the first House of Commons vote on shale gas legislation. Such a concerted lobbying effort may span both outside lobbying, such as joint press releases or public events, and inside lobbying, for instance through joint consultation responses. In the example of the anti-shale gas initiative, there was a combination of both outside and inside lobbying as a 'signalling coalition'.
Decomposing active cooperation into these distinct activities of information exchange, strategy coordination and joint signalling has the advantage of allowing more fine-grained analyses of their effects. Moreover, other than seeing them as distinct activities, one can see them as part of a spectrum of increasing cooperation (Hula 1999). Information exchange (of increasing degree) may here constitute a lower level of cooperation and necessary condition for the other types of cooperation. In addition, the tighter forms of cooperation increase the level of cooperation when additionally coordinating strategies (with increasing numbers of partners) and lobbying concertedly in a signalling coalition (of increasing size) on an issue (cf. Junk 2019a). Figure 2 illustrates that active cooperation can be conceptualised on a scale depending on the intensity of all three cooperation activities.

Methods of data collection and empirical evidence
Unfortunately, data on these forms of active cooperation on specific issues are far from easily available. Some studies collect information regarding issue-specific cooperation in interviews Mahoney and Baumgartner 2004), or (telephone) surveys (Haider-Markel 2006;Nelson and Yackee 2012). Given that the levels of information exchange and strategy coordination with others are private information, it can probably only be gathered by asking actors themselves to report on these activities.
In contrast, active signalling coalitions should also leave traces, such as joint media statements, concerted consultation responses, or other concerted campaigns. If the research interest lies in this cohesive form of issue coalitions, then coding such primary sources is an alternative to using interviewing or surveying. Box-Steffensmeier et al. (2013), for instance, study coalitions by using the co-signing of amicus curiae briefs to the US Supreme Court. Similarly, Phinney (2017) studies formal and informal coalitions by assessing source material on US congressional hearings and newspaper articles, thus capturing 'signalling coalitions' in inside and outside lobbying. To gather such information across large numbers of issues and/or long periods of observation in future, automated or computer-aided text analysis is a promising tool. Dwidar (2019), for example, analyses co-signature patterns on public comments by a sample of interest groups on proposed US federal agency rules between 2005 and 2015 in this way.
Of course, both observational and surveying methods come with their advantages and downsides. Mobilising interview partners or implementing surveys with satisfactory response rates (cf. Marchetti 2015) is labour intensive and comes with many uncertainties, such as whether the resulting number of observations will be sufficient. Moreover, results will face potential non-response bias, and other biases, such as misrepresentation or misremembrance of cooperation activities. Using primary sources, such as consultation responses, press releases, or newspaper articles, is likely to be less biased in these respects. However, it comes with the danger of underestimating the actual frequency of signalling coalitions, given that these can occur in many different venues, some of which may be overlooked in the coding. Before embarking on either form of data collection, it is useful to have a yardstick as to how frequent these different cooperation activities are in practice. I present here an overview of their use based on data from an online survey sent to 1410 advocates, including interest associations, companies, and experts, 4 in five European countries (Denmark, Sweden, the Netherlands, Germany, and the UK) as part of the GovLis project. 5 The survey gathered detailed information on cooperation activities on 50 specific policy issues, 6 giving explanations and examples of what is entailed by each cooperation activity. 7 The surveys in the five countries have an overall completion rate of 33.9%, as 478 respondents completed the survey to the end. 8 Table 1 lists the operationalisations of the three forms of active cooperation covered in the survey and summarises the gathered responses. It shows the relative frequency of these forms cooperation: 74% of respondents reported that they exchanged information on the issue with other advocates 'very often', 'often', or 'sometimes'. In addition, 50% responded to have cooperated strategies with other actors. Moreover, 36% answered to have been in a formal coalition on the issue, for instance through a declared coalition, joint press releases, joint position papers, or jointly approaching policymakers.
To compare, both relying on interview data, Hojnacki (1997) in the US context and Beyers and De Bruycker (2018) on EU lobbying, document similar frequencies of active, issue-specific coalitions. Hojnacki (1997: 74) finds that 69% of active organisations in her sample were members of active coalitions, whereas 55% of the actors studied by Beyers and De Bruycker (2018: 696) were in a coalition. In contrast, Phinney (2017: 71) finds formal 'signalling' coalitions for only ca. 14% of all organised actors, when looking at concerted testimonies to US congressional hearings on welfare reform. Yet, she writes herself that this number 'understates the extent of collaborative activity among organised interests' (Phinney 2017: 72), because the testimonies provided additional evidence of more informal collaboration, some of which may also fall under 'strategy cooperation' in the categories I suggest.
Additionally to the total shares of cooperating actors, Table 1 provides an impression of the variation in these shares between countries by listing the country with the minimum and maximum share of cooperating actors for each operationalisation. This shows that all cooperation activities are relatively frequent (≥ 28%) in all countries and that there is most cooperation in the Dutch responses on all three measures. 5 For more information and other project publications, please visit: www. govlis. eu. 6 Issues were selected as a stratified random sample (varying media salience, policy type, and the level of public support for policy change) from the universe of national policy issues on which public opinion surveys were conducted between 2005 and 2010 (cf. Rasmussen et al. 2018). Appendix A in the supplementary information (SI) gives an overview of the 50 issues. 7 Appendix B provides the wording of the questions on each coalition activity. 8 Appendix C in the SI shows how response rates vary by country ( It is not conclusive, however, whether this variation is due to actual country differences in coalition patterns or due to differences in the sampled issues. Finally, Table 1 gives mean values of the level of cooperation according to the three measures. In case of information exchange, the mean ranking on a 1-5 scale is 3.2, so exchanging information with other advocates more than 'sometimes' on the issue. The mean number of partners with which strategies was cooperated is 1.3 in the total sample. For those actors who did cooperate strategies, there is an average strategy coalition size of 3.1 partners. In case of formal signalling coalitions, the mean number of partners across all actors is 1.1, and the mean coalition size is 3.5 partners. 9 Overall, Table 1 indicates clearly that all three cooperation activities are frequent on the issues in the sample. Moreover, the three measures of cooperation activities vary and seem to capture distinct behaviour. There are, however, relatively high correlations between the cooperation activities (0.54 < r >0.62 10 ): This means, firstly, that constructing an index, such as by summing the levels of cooperation in the different operationalisations, may be preferable to including them as separate variables in the analysis. Secondly, one might argue that given they are relatively highly related, it might, for some purposes, be sufficient to include only one of the 9 Few existing studies have reported such data on average coalition size, but anecdotal evidence may suggest that an average size of 3.1 or 3.5 partners seems relatively low. This could, in part, be owed to the survey methodology, where respondents needed to enter the names of coalition partners. Observational data collection may be expected to yield 'larger' coalitions. In future work, a closer analysis and comparison of coalition size (across countries and sources) would be fruitful. 10 See Appendix Table C.3 in the SI. Notably, however, this should be chosen based on the theoretical assumptions in the respective research design about whether cooperation in terms of informational resources, cooperated strategies, or formal coalitions will be most relevant for the relationships at hand. Moreover, it is relevant to ask whether the issue-specific approach to cooperation activities is the most suiting one for the research design in the first place. The next section outlines an alternative approach that sees cooperation as a general organisational characteristic.

General cooperation structures: the organisational approach
A focus on single issues may overlook that lobbying is a repeated game. Organisations with similar preferences may, in fact, build up institutionalised ties that foster cooperation on many issues, for instance looser partnerships networks of affiliated advocates, or separate organisational structures, which bring together member organisations in so-called umbrella organisations.

Conceptual distinctions: the presence and institutionalisation of general ties
From this perspective, cooperation between political advocates is seen as an organisational characteristic that measures the general embeddedness of the actor in cooperation structures with other actors. As Fig. 3 illustrates, these can be conceptualised on a scale with varying degrees of embeddedness ranging from unconnected organisations to the existence of separate formal organisational structures employing their own staff in an umbrella organisation.
In contrast to isolated organisations, actors embedded in partner networks will benefit from regular exchange with others in the network, which might enhance their exchange position vis-à-vis political gatekeepers and thereby affect political access or influence. Heaney (2006), for example, shows that interest groups with brokerage positions in the communication networks on US health policy are more often cited as influential actors on policy. The ties between organisations in partnerships can here vary in strength and nature, including, for instance, more or less frequent and more or less direct communication, events or projects Heaney 2014). As  show in the European context, the types of such ties may have differential effects on different political audiences, such as elected and non-elected officials.
Alternatively to analysing the connectedness of actors with network analytical tools (see also : Heaney 2014;Heaney and Lorenz 2013), one can look at the level of embeddedness in networks as an actor-level variable in models of lobbying access or influence. Tallberg et al. (2015), for instance, include a categorical variable to measure whether non-governmental organisations (NGOs) lobbying International Organisations (IOs) have a low or high frequency of interactions with other (business and/ or non-business) actors. In this way, the authors provide evidence that involvement in transnational networks, especially with for-profit organisations, positively affects NGO influence in IOs, such as the United Nations.
Formal 'umbrella associations' such as such as the European Environmental Bureau (EEB) that joins roughly 150 civil society organisations with an office and over 50 staff members in Brussels differ from looser partnership networks by introducing a more hierarchical ordering of inter-organisational ties. The umbrella, which can be seen as a permeant 'coalition' of its member groups, can channel member views and information provision and may even affect the positions groups take on government proposals, as Bunea (2015) shows in the EU context for consultations by the European Commission. While a larger literature has addressed the existence and effects of such umbrella associations at EU level (e.g. Bouwen 2004;Bunea 2013;Eising 2007;Kohler-Koch and Quittkat 2013), such organisations also exist at national level and measurably affect access to decision-makers, for instance, in the Dutch, German, and UK context (Fraussen et al. 2015;Junk 2019b).
From the perspective of policymakers, umbrella organisations can make the consultation process more efficient and increase the number of organisational voices that can (indirectly) be heard, as well as ease subsequent implementation and legitimacy problems (cf. Kröger 2014). As illustrated in Fig. 3, institutionalised umbrella organisations can vary in their characteristics, such as the number of member organisations or ties, and their territorial scope (cf. Bouwen 2004;Bunea 2013;Junk 2019b), and such factors may affect the umbrella's effectiveness and its appeal in exchanges with political decision-makers.

Methods of data collection and empirical evidence
To measure such cooperation structures empirically, interviews Heaney 2014) and surveys (Tallberg et al. 2015) are again common forms of data collection. When network analytical approaches are used, it is of special concern that data collection can cover the full network, so low response rates to surveys can pose a special problem. Alternative ways of data collection include the coding of website content (Bunea 2013;Fraussen et al. 2015;Junk 2019b), where groups typically name formalised partner organisations or existing umbrella structures. In addition, it is an option to use social media networks as a way to gauge general cooperation, for instance, by drawing on patterns in twitter followership, retweets or hashtag use between organisations as an indication for ties (cf. Xu et al. 2015). Importantly, these different ways of measuring ties will not be equivalent, for instance because twitter followship entails much lower commitment than entering a named partnership. Table 2 provides empirical evidence on the prevalence of two forms of network embeddedness coded based on website data on a sample of 329 interest groups active on twelve diverse issues in the UK and Germany. 11 These measures of cooperation capture generally whether the active actor was embedded in (national or international) partner networks and whether the group was itself an umbrella organisation, meaning containing several (> 1) member organisations at national or subnational level. 12 Additionally, the coding captured the number of member groups. This information was gathered on the organisations' websites, supplemented with Wikipedia entries, or in telephone or email contact, where the information was missing.
This information is still likely to underestimate looser forms of general cooperation that actors do not publically report on their websites. Still, it gives strong indication that a large share of 66% of active actors on the sampled issues either have national or subnational member groups themselves (i.e. are umbrella organisations), or report to be embedded in partner networks, such as being a member of an umbrella association.
The average number of member groups in an umbrella organisation in the sample of active actors is 36 members, that is, more than ten times the average issue-specific signalling coalition size reported in the previous section. This may tentatively indicate that ad hoc coalitions tend to be much smaller than highly institutionalised 'coalitions' in the form of umbrella organisations. Such size differences are likely to affect dynamics within the coalition, such as coordination problems, member voice and efficiency.
In any case, according to Table 2 general cooperation structures are very frequent and connect a large share of the active actors on the sampled issues to each other and/or to actors that are not active on the issues themselves. Considering these results, one can hypothesise that some member groups delegate active lobbying work to their umbrellas, or divide labour with their partner organisations (cf. also : Eising 2007;Junk 2019b). Such patterns will be consequential for studies of biases in mobilisation and lobbying access, but tend to be overlooked when treating umbrella organisations simply as interest groups in their own right, rather than 'coalitions' that channel member voices.

Theory-driven selection of the approach and dimensions of variation
It needs to be re-iterated that the empirical phenomena behind these approaches are not mutually exclusive. Advocates will share preferences, cooperate actively on a specific issue, and cultivate more general and permanent ties-all at the same time. In this sense, advocates come with a 'coalition portfolio' (Heaney and Lorenz 2013) of crosscutting ties of varying degrees and qualities on several issues at a specific point in time. Still, due to the costs of data collection, most studies will need to focus on specific excerpts of the full 'portfolio'.
This selection of an approach should go hand in hand with the theoretical expectations on how coalitions enhance lobbying access or success in the specific context of study. What is it about coalition lobbying that is expected to drive higher efficiency, or to affect certain target audiences, such as legislators, bureaucrats or the public? Do coalitions help (especially weaker) actors by pooling resources (cf. Junk 2019a; Phinney 2017; Strolovitch 2007), do they signal political weight and broad support (cf. Junk 2019c; Mahoney 2007b; Nelson and Yackee 2012), and/or do they serve to manage complexity and facilitate stakeholder input in increasingly crowded decision-making spaces (cf. Baumgartner and Jones 2010: 177f; Kohler-Koch and Quittkat 2013)? The more targeted such expectations and the respective operationalisations of coalitions, the higher the potential to add new knowledge about the effects of different kinds of collective lobbying-in specific circumstances.
In their conference paper more than a decade ago, Mahoney and Baumgartner (2004: 11) already stress that many 'hypotheses stemming from the literature [on lobbying coalitions and their effects] make perfect sense for certain groups in certain circumstances, but may not hold across the board'. However, until now the interest group literature has only begun to explore such hypotheses on the conditional effects of coalitions. Based on existing studies, there is reason to expect variation in the usefulness of coalition action relating to several factors, namely (1) the type and composition of the coalition, (2) the characteristics of the advocate, (3) the characteristics of the issue, and (4) the political venue the coalition tries to impact (see, for instance, Beyers and De Bruycker 2018;Hanegraaff and Pritoni 2019;Hojnacki 1997;Holyoke 2004;Junk 2019a, b, c;Mahoney 2007a,

Conclusion
In order to facilitate important future research on lobbying coalitions and their effects, this article has discussed some of the important choices involved in conceptualising and empirically measuring coalition behaviour. It distinguished between three approaches to coalitions, namely a preference similarity approach, a behavioural approach, and an organisational approach. The latter two approaches are arguably most fruitful when we seek to understand synergies in lobbying, because they go beyond treating lobbying by like-minded groups as additive, but tap into potential multiplicative effects of lobbying together with others, for instance, by actively pooling resources or signalling consensus to decision-makers as an issuespecific coalition or as an umbrella association. The article addressed conceptual and methodological choices involved when working with these approaches and provided empirical evidence as to how prominent such forms of cooperation are in European countries. Which of these approaches and operationalisations is 'the right one' depends on the research question and theoretical framework at hand, for instance its focus on certain resources that are expected to be valuable in exchange with target audiences of the lobbying effort. Moreover, different approaches may fit more naturally with different theoretical stances. From a traditional elitist approach (Hunter 1953;Mills 1956), we might expect the same powerful and well-connected actors to gain preferential access and exert policy influence across issues and venues. Based on these expectations, general cooperation structures may be of prime concern and can shed new light on elitist biases in favour of resourceful networks of actors. In contrast, a pluralist perspective would stress that power is more diverse in its sources and forms and will importantly vary across policy areas or issues (Dahl 1961;Truman 1958), as well as between arenas of policymaking. From such a stance, issue-specific cooperation behaviour and its effects in different contexts might be of special interest.
This article provided a conceptual toolbox for researchers wanting to pursue such future projects. I believe this is crucial, because the perspective of 'with whom' adds nuance to the questions of 'who gets what, when, how' (Lasswell 1950) in interest group research and political science more generally. Rather than just being a particular version of the 'who' and 'how', a perspective on cooperation activities and patterns can challenge the underlying assumptions with which we approach questions of power and agency. Instead of seeing the 'who'-lobbying actors in this case-as individual and independent actors, the 'who' becomes a collective or network of actors, whose strategies, resources and ultimate success are interdependent. The empirical evidence in this article suggests that both issue-specific cooperation activities and general cooperation structures are very frequent, so the reality we are studying is one of highly connected actors, rather than independent units. This methodological challenge can give rise to important new research projects assessing how cohesive sets of groups succeed in influencing policy. This article has provided the necessary conceptual tools to theorise, operationalise and study such cooperation.