Interest group leaders strengthen the political power of the groups by

Abstract

Although research on interest group influence has had a revival in recent decades, little is known about interest group influence on political parties. This study considers how parties’ goals can affect interest groups’ ability to influence these actors. Interest groups are expected to be more likely to influence parties that are ideologically approximate to the interest groups as well as parties that are more willing to make policy compromises; the access that interest groups have to parties mediates these effects. Results from the empirical analysis of 5000 party-interest group observations from Denmark, France, Germany, the Netherlands, Norway and the United Kingdom support these expectations. In addition to contributing to the literature on parties and interest groups, the findings shed light on party position taking and when the link between citizens and parties is likely to be affected by interest groups.

Supplemental data for this article can be accessed online at: https://doi.org/10.1080/01402382.2021.1921496 .

Research on interest group influence has had a revival in recent decades (e.g. Baumgartner et al. 2009; Bunea 2013; Dür et al. 2015; Junk 2019; Klüver 2011), but little is known about interest group influence on political parties and their positions. Given the crucial role that parties have when it comes to legislating and governing (Dalton et al. 2011), interest groups that aim to influence public policy might find them important targets. Examining when interest groups are able to influence parties sheds light on the political impact of groups in the stages before final policy decisions are made.

The party-interest group literature has mainly focussed on their relationships (e.g. Allern and Bale 2017). The few existing studies that systematically examine groups’ influence on parties have focussed on the effects of interest group resources and interest group–party relationships (Clifton 2004; Karol 2009; Victor and Reinhardt 2018). A consideration of party characteristics is, however, missing. Parties have their own agenda and face certain constraints and incentives when it comes to using interest group input. This article considers how the goals parties pursue can affect interest groups’ ability to influence these actors. Influence is here understood as persuading parties to take a position that is (more) in line with the group’s preferences (Dür 2008). Interest groups, in other words, have to perceive that they have been able to affect a given party’s decision.

Parties are assumed to want office, policy and votes (Strøm 1990). Interest groups are on the other hand purer policy maximisers and specialists but do not compete in elections (Hansen 1991). This leaves room for potentially mutually beneficial exchanges between the two actors. In exchange for political influence, interest groups can provide information to parties that can help them achieve their goals. As all parties are assumed to be policy seekers to some extent, I assume that information from ideologically similar groups is more interesting to parties. This type of information likely helps parties in their pursuits without them potentially incurring substantial costs. Interest groups might furthermore be more likely to focus their efforts on such ‘allies’ which can make them more likely to succeed in wielding influence.

The goals that parties pursue might conflict with each other, however, and how parties choose to prioritise their goals can affect their propensity to use interest group input. More specifically, primarily policy-seeking parties that are relatively unwilling to make policy compromises are expected to be less responsive to interest groups. These parties are assumed to place a higher premium on their positions, which can make it harder for the average interest group to wield influence. Primarily office- or vote-seeking parties on the other hand are assumed to have fewer policy constraints on what they consider as useful input and to be interested in catering to a broader group of voters. This means that the pool of interest groups these parties are interested in listening to is likely larger. Interest groups, realising this, might also put more effort into influencing parties that are more willing to compromise on policy.

I test these expectations on 5000 party–interest group observations from Denmark, France, Germany, the Netherlands, Norway and the United Kingdom. The hypotheses are supported. Interest groups that are ideologically closer to a party are generally more likely to be influential. Interest groups are in addition found to be more likely to influence parties that are more willing to compromise on policy. I moreover examine whether these effects are mediated through the access that interest groups have to parties, and find that this largely seems to be the case.

In addition to contributing to the party-interest group literature, the article’s findings add to the interest group influence literature at large. The theoretical focus here has mainly been on issue and interest group characteristics (Hojnacki et al. 2012). A smaller issue scope, less conflict, less salience and interest group positions that are aligned with public opinion tend to be associated with more group influence (Hojnacki et al. 2012; Mahoney 2007; Rasmussen et al. 2018). As these aspects broadly concern how much attention an issue gets in the public eye, they might be particularly important to consider for decision makers who can be held electorally accountable. Research on interest group characteristics supports this. Citizen groups, which tend to promote more diffuse public interests, are generally more likely than business groups to influence EU policy outcomes that involve the directly elected European Parliament (Dür et al. 2015). Business groups are, on the other hand, more likely to influence bureaucratic rule making (Yackee and Yackee 2006). The priorities of elected actors may, in other words, differ from non-elected ones. This article also considers that actors of the same type may differ as well. I unpack how differences in the motivations of elected actors (political parties) can vary and how this can affect groups’ ability to wield influence.

My findings furthermore give a more complete picture of the political processes that lead to public policy outcomes. Understanding what leads parties to take certain positions is important in order to understand how public policy is made. While parties are known to adjust their agendas in response to interest group mobilisation (Klüver 2020; Otjes and Green-Pedersen 2019), studies that aim to explain party positions have paid little attention to the potential impact of interest groups (Adams 2012). This article hence contributes to the literature on party positions by shedding light on the kinds of parties that interest groups are more likely to influence.

Understanding how parties arrive at their positions is also important given their roles as intermediaries between citizens and the state. Substantive representation, that citizens’ preferences are reflected by policy makers, to a certain degree hinges on parties. Depending on who their constituents are, the information interest groups provide can reflect broader, public or narrower, sectional interests. Interest groups can therefore strengthen or weaken the link between citizens and the parties that take their input into account (Giger and Klüver 2016; Lax and Phillips 2012). Knowing which parties interest groups are more likely to affect is pertinent in order to be able to gauge when the link between voters and parties is likely to be affected.

Political parties are considered to have three main goals: policy, office, and votes. Office seeking entails a wish to gain office benefits, while policy seeking is understood as a desire to influence public policy. Votes are mainly a means to these two ends (Strøm and Müller 1999). Interest groups, defined as non-governmental, formal associations that try to influence policy outcomes but do not compete in elections (Beyers et al. 2008), are in comparison purer policy maximisers. They generally focus on a narrower range of policy issues and tend to be more specialised than parties (Hansen 1991).

Given parties’ role in legislatures and governments, they can be important targets for interest groups. This is supported by the literature on the contemporary ties between interest groups and parties. Parties are found to have ties to a relatively broad range of interest groups (Allern and Bale 2017; Otjes and Rasmussen 2017; Rasmussen and Lindeboom 2013). Groups are also known to lobby party manifestos and government coalition agreements (Däubler 2012; Dolezal et al. 2012; Romeijn 2021). In exchange for political influence, interest groups can provide parties with resources that aid them in achieving their goals.

One important resource is information (Bouwen 2002). Parties have incomplete information about the consequences of different policy alternatives, both in terms of how they might affect certain segments of the population as well as how they might affect their electoral support (Austen-Smith 1993). Interest groups can provide expertise, such as information about the potential impact of a policy proposal, and political information concerning e.g. a proposal’s support among a certain constituency (De Bruycker 2016b). Given interest groups’ higher degree of specialisation, the information they provide can help a party navigate what position to take. Using interest group input might also improve the party’s appeal to the group’s constituency and potentially help the party attract these voters.

When interest groups try to influence parties’ positions and persuade them to move closer to their ideal points, they nevertheless meet another policy seeker. Parties are also interested in steering public policy in their preferred direction. Information that is consistent with a party’s ideological principles might therefore be more interesting to the party. Parties face risks if they take positions that diverge from their ideological principles. Doing this might confuse and repel their core voters, demobilise party activists, and cause internal party divisions (Adams 2012; Adams et al. 2006; Budge 1994). Knowing that a group shares similar principles as the party might also be useful for parties when it comes to trusting their input on specific policy issues (Hall and Deardorff 2006). Using information from ideologically proximate interest groups is therefore assumed to be less costly, while also potentially improving the party’s arguments and appeal to the groups’ constituents. Knowing this, interest groups might also be more prone to target these ‘allies’ which can positively affect their chances of wielding influence (Hojnacki and Kimball 1998; Otjes and Rasmussen 2017).

Both from a party and interest group perspective, the first expectation is therefore,

H1 Interest groups are more likely to influence more ideologically proximate parties.

While parties are assumed to be interested in pursuing office, policy, and votes simultaneously, it can be hard to maximise all at the same time as the goals can conflict. Parties have to decide what they primarily want to pursue. More policy-seeking parties have been found to be less responsive to changes in the median voter’s position (Adams et al. 2006; Bischof and Wagner 2020), and the premium a party places on policy seeking might also affect its inclination to use input from the average interest group.

More specifically, the value parties place on policy purity is assumed to be important. While all parties are likely to want to influence public policy, policy purists are less willing to make policy compromises to gain office benefits or more votes (Pedersen 2012). This type of party is assumed to focus on appealing to and representing the interests of its core constituency and not be as interested in catering to voters outside of this group (Berkhout et al. 2019; Bischof and Wagner 2020). An interest group’s input will hence likely need to be consistent with the interests of the party’s core constituency for the party to find it useful. The pool of interest groups whose input these parties are interested in is therefore likely smaller than that of parties with other priorities.

For office- and vote-seeking parties, the policy constraints on what they consider as useful input are assumed to be weaker. These parties might be interested in catering to a broader group of voters compared to policy-seeking parties. For a vote-seeking party, this is a way of maximising votes. Maximising votes is also attractive for office-seeking parties as the party’s size affects its chances of entering office (Martin and Stevenson 2001). In the case of coalition governments, votes are also associated with the allocation of ministerial portfolios, i.e. a party’s share of office benefits (Warwick and Druckman 2006). Information from a more diverse set of interest groups might help parties appeal to more voters. Potentially incurring the costs of compromising on its principles can be worth it to a vote- or office-seeking party to the extent that using the interest group input has a positive impact on its chances of appealing to the group’s constituents.

It is in other words likely easier for the average interest group to influence a party that is more willing to compromise on policy. Interest groups, realising this, might spend more energy on trying to influence such parties. Interest groups furthermore have an incentive to concentrate their efforts on primarily office- or vote-seeking parties to the extent that they are more likely to have powerful positions in parliament or office (De Bruycker 2016a; Otjes and Rasmussen 2017). This may positively affect their chances of being successful in their endeavour.

Both from a party and interest group perspective, the second hypothesis is therefore,

H2 Interest groups are more likely to influence parties that are more willing to compromise on policy.

A mechanism that potentially links ideological proximity and parties’ willingness to compromise on policy to interest group influence is groups’ access to parties. Access, which requires that interest groups seek it and parties grant it, is generally thought of as a prerequisite for influence (Binderkrantz et al. 2017; Truman 1951). Being in regular touch with a party likely gives a group more opportunities to voice its opinions which can increase its likelihood of getting the party to yield. Whether a group seeks access to a party likely depends on both the group’s ideological proximity to the party and the party’s willingness to make policy compromises. These two aspects might also affect the party’s motivation to grant access. I therefore expect that,

H3a Interest groups are more likely to influence more ideologically proximate parties because they are more likely to seek and gain access to these parties.

H3b Interest groups are more likely to influence parties that are more willing to compromise on policy because they are more likely to seek and gain access to these parties.

In order to test these expectations, I use interest group survey data from the Party–Interest Group Relationships in Contemporary Democracies (PAIRDEM) project supplemented with data from other sources as detailed below. The survey was carried out between 2017 and 2018, with the informants generally being those responsible for their group’s governmental affairs or public relations or the director-general (Allern, Hansen et al. 2020). Interest groups and parties from six established democracies – Denmark, France, Germany, the Netherlands, Norway, and the United Kingdom – are covered here.1 These countries represent both pluralist (France, United Kingdom) and corporatist (Denmark, Germany, the Netherlands, Norway) ways of structuring state–interest group relations (Jahn 2016).

Examining interest groups and parties in these countries makes it possible to examine a variety of parties. Generally, the parties that were represented in parliament when the survey was carried out are included. Both established and newer parties of varying sizes are covered. The parties are moreover relatively strong and cohesive, which can make influencing parties a more important strategy for interest groups. In this regard, there is also variation in how closely tied the parties have been and continue to be to specific interest groups (Allern and Bale 2017). Overall, the findings here are likely to apply to a relatively broad range of cohesive parties in established democracies.

Interest groups are defined as ‘any non-party and non-governmental formal association of individuals or organisations that, on the basis of one or more shared concerns, advocates a particular interest/cause in public and usually attempts to influence public policy in its favour in one way or another’ (Allern, Hansen et al. 2020: 2). The PAIRDEM survey includes one random sample drawn from the identified interest group population in each country as well as one purposive sample. The purposive sample includes the ten most important interest groups in each country, as identified by the country expert who carried out the survey, in eight categories.2 These categories allow for including groups that are concerned with economic as well as postmaterialist issues. Given the heterogeneity of the interest group universe, the purposive sample hence ensures that groups that might be more oriented towards issues that concern political parties are covered. About 21% of the groups that responded belong to this sample.

The aggregate response rate is 29%, which is similar to that of other interest group surveys. There are no substantial biases with regards to the types of groups that have responded (Allern, Hansen et al. 2020). In sum, a variety of interest groups are part of the overall sample studied here, including business groups, professional associations, trade unions and public interest groups. Both membership and non-membership organisations are included. Given that the ideological proximity hypothesis requires that interest groups have at least one ideological position, however, the applicability of the results might not extend to groups that pursue less politicised interests.

Interest group influence is understood as a group’s ability to persuade a party to take a position that is (more) in line with the group’s preferences (Dür 2008). I use the following question from the PAIRDEM survey to measure this:

Thinking about when your organisation has input into decisions made by parliamentary party groups about the top three policy areas you are the most active in during the current legislative term, how would you rate the influence of your organisation on [party]? Note: If the present legislative term has just begun, please refer to the preceding period. To be considered ‘very influential’, your organisation’s input must have had a decisive influence on the positions taken by the parliamentary party.

The interest groups were asked to rate their influence on each of the parties that were currently represented in parliament in the policy areas where they had indicated being the most active earlier in the survey.3 The groups could choose up to three policy areas. A given interest group and party pair can hence occur up to three times in the dataset. Note that it was possible to answer ‘not applicable’ for cases where the interest group did not provide any input. These observations are excluded.

One of the drawbacks of measuring influence in this way is that it relies on the memory and knowledge of the respondents. The respondents might furthermore have incentives to exaggerate or downplay their influence (Dür 2008). The question is moreover one-sided in that only the interest groups are asked to rate their influence. At the same time, the groups might have a better overview of the instances where they successfully wielded influence since they may be dealing with fewer actors and policy areas compared to the parties.

A drawback with the PAIRDEM question in particular is that it is not decision specific. It is therefore not possible to include issue characteristics in the analysis. The survey is not exclusively concerned with influence, however, and asking respondents about influence on specific decisions would have required more questions which could have increased respondent fatigue and lowered response rates. Moreover, compared to previous interest group surveys, the PAIRDEM question has the advantage that respondents are asked about their influence on decisions in specific policy areas during a given legislative term, as opposed to their influence in general (Pedersen 2013). This also makes it possible to include a policy area control variable.

Given the wording of the question, it might moreover be reasonable to assume that the respondents will consider an aggregate of specific instances where the group provided input to a party when answering the question. The specification of what ‘very influential’ means furthermore increases the likelihood that different respondents interpret the question similarly. Overall, the measure likely gives a good first indication of whether the hypotheses receive empirical support.

The influence measure ranges from 0 to 3. 0 equals not at all influential, 1 not very influential, 2 somewhat influential and 3 very influential. The mean is 0.99 (see Online appendix 1 for descriptive statistics). The typical group is in other words likely to rate its influence on a given party as weak. There is, however, a relatively large degree of variation. The standard deviation is 0.97.

Given the categorical and ordered nature of the dependent variable, I run ordered logistic regressions. The unit of analysis is party–interest group pairs. Given the structure of the data, the observations might not be independent. Empty multilevel models indicate that the interest groups and parties account for most of the variance (Online appendix 2). I therefore include random effects at both the interest group and party level. I moreover include country fixed effects to account for potential differences between the countries, also in terms of response rates. Linear mixed effects regressions as well as ordered and multinomial logistic regressions with standard errors clustered by interest groups and parties yield similar results (Online appendices 3–5). Multicollinearity is not found to be a problem.

The first independent variable is the ideological proximity between a party and interest group. To measure this, I construct a variable based on several PAIRDEM questions as well as party positions from the 2014 Chapel Hill Expert Survey (CHES) (Polk et al. 2017). The PAIRDEM respondents were asked to indicate their positions on six different scales that are identical to the ones used in CHES (see Online appendix 6 for the wording of each question). Three of the scales cover economic issues (improving public services versus reducing taxes, redistribution, and state intervention in the economy) and three cover postmaterialist issues (social lifestyle, immigration, and the environment). The questionnaire included a visual representation of where the parties in the group’s country were located on each scale. The PAIRDEM and CHES data in other words allow for a relatively precise comparison of the ideological positions of a given group and party both when it comes to more traditional, left–right issues as well as newer, ‘GAL–TAN’ issues.

I calculate the average proximity between a given party and interest group on the dimensions where the group has indicated its position. The resulting measure hence includes the party–interest group proximity on at least one and up to six dimensions. The variable ranges from 0 to 10, with a mean of 7.3 and a standard deviation of 1.66. Higher values indicate greater proximity. Note that the party positions precede the interest group positions in time, which eliminates the problem that a party’s ideology might be close to a given group because the group has influenced the party.

The second independent variable is the priority parties give to policy purity, understood as a greater unwillingness to make policy compromises. I rely on party behaviour to measure this and consider government participation as well as the policy distance between a party and its most dissimilar government coalition partner. As discussed below, these aspects tap into a party’s willingness to make policy compromises.

Most of the parties included in this study have to join a coalition or form a single-party minority government if they want to access office benefits. These two options generally involve policy compromises (Martin and Vanberg 2014; Strøm 1990). Discussions of party goals in the existing literature furthermore support using government participation as a proxy. Being in office for the Danish, Dutch, German and Norwegian parties has typically meant being part of a coalition and this has involved policy compromises (Allern 2010; Allern and Karlsen 2014; Faas 2010, 2015; Kosiara-Pedersen 2012, 2016, 2020; Pulzer 2003; van Holsteyn 2007, 2018). The parties that have not been in office have generally deliberately chosen to stay out, partly because of the policy compromises that would be required. A different logic applies to the parties in the United Kingdom and France. The parties that have been in office here, however, are generally argued to pursue office and/or votes rather than policy purity (Samuels 2002; Quinn 2010). The parties that have not been in office are considered to be more concerned with policy (Laver and Hunt 1992; McAngus 2016). There is hence also broadly support for using government participation as a proxy in these cases. Online appendix 7 includes a more thorough discussion of each party.

I take government participation between 2000 and 2016 into account in order to include several elections and potential changes in government. This increases the chances that those who have wanted to govern have had a chance to do so. I consider the number of days a party has been in office during this time period to distinguish between parties that have governed for longer periods of time and those that e.g. have had a short spell in office and ended up withdrawing because they found the costs of governing to be too high.

In addition to government participation and similar to Warwick’s (2006) policy horizons measure, I consider the policy distance between a given party and its most dissimilar coalition partner. Coalition partners with fairly similar ideological positions might have to compromise less than more dissimilar partners. This also accounts for the possibility that those who govern alone might not have to compromise as much as those who govern with other parties. The policy distance therefore gets at the parties’ ‘limits of policy compromise’ (Warwick 2006: 41). A party that has governed with a more ideologically dissimilar coalition partner is thought to be more willing to make policy compromises.

The policy distance between a party and its most dissimilar coalition partner between 2000 and 2016 is calculated using the general left–right positions from CHES between 1999 and 2014 (Bakker et al. 2015; Polk et al. 2017). The position of a given party’s most dissimilar partner is subtracted from the party’s position. If a party has participated in several coalition governments, the average distance is used. Parties that govern alone are given a score of 0.

I normalise the days a party has been in office and the policy distance between the party and its most dissimilar coalition partner to range between 0 and 1 and add the two variables together. The resulting measure ranges from 0 to 2, with a mean of 0.69 and a standard deviation of 0.54. A score of 0 equals parties that have not been in government and are hence considered to be less willing to compromise on policy (e.g. the Danish People’s Party). Higher values indicate a greater willingness to compromise. The party has been in office for a longer period of time and potentially governed with parties that are more ideologically dissimilar (e.g. the German Social Democratic Party).

Note that a party’s willingness to compromise on policy is distinct from how moderate its positions are. Moderate (extreme) parties might have to accept a more extreme (moderate) position than they would have preferred when they are in office because of e.g. other parties’ preferences or public opinion. Parties that want to preserve a ‘pure’ policy profile might hence be deterred from entering office.4

There are some potential concerns with this measure. First, it does not account for pariah parties that might be office seeking but unable to take part in a coalition because it its ostracised by established parties (Downs 2001). Two parties in this sample, Front National and the German Left, are considered to be ostracised (Akkerman and Rooduijn 2015; Bräuninger et al. 2019; van Spanje and de Graaf 2018). The results are, however, similar when I control for pariah party status (Online appendix 8).

Second, the results might be driven by the parties that were in government during the term that the influence question concerns or by the parties that have more parliamentary seats. These two aspects tap into the power that parties possess. Interest groups might target government parties or larger parties more (De Bruycker 2016a; Otjes and Rasmussen 2017), which can affect their ability to wield influence. Government parties, however, may be harder to influence since making promises that turn out to be empty can be more consequential for these parties (van Spanje 2010). The results are again similar when I control for these two aspects (Online appendix 8).5

As an additional robustness check I code a variable based on the aforementioned literature review (Online appendix 7) that distinguishes between parties that are not willing, somewhat willing, and willing to compromise on policy. This variable is strongly correlated with the willingness to compromise measure (r = 0.73, p-value < 0.01) and the results are robust (Online appendix 9).

In order to examine whether the effects of ideological proximity and parties’ willingness to make policy compromises are mediated through interest groups’ access to parties, I use Imai et al.’s (2011) causal mediation framework. I include country fixed effects and cluster the standard errors on interest groups and parties. Access is measured using a question from the PAIRDEM survey where the respondents were asked whether they ‘usually talk to [party] when trying to give input into the decision-making process on a major issue within the top three policy areas you are most active.’ The response alternatives were yes/no.

While the dichotomous nature of the variable makes it somewhat crude, the measure is party- and policy area specific which importantly makes it possible to know whether interest groups tend to talk to a given party in a given policy area, akin to the influence question. Talking to a party requires that the group is seeking contact and that the party is willing to grant this. This makes it a useful first test of a possible access mechanism. Around 28% of the observations have this kind of contact.

I control for a party’s degree of intra-party democracy – how involved members and activists are in decision making – since this can affect both how parties prioritise their goals and interest group influence. First, party leaders have (the prospect of) access to office benefits and are assumed to be motivated to pursue office. These benefits are inaccessible to the average member and activist who instead are assumed to be motivated by policy. When these actors are given more of a say, the party might therefore prioritise policy (May 1973; Strøm 1990). This might make the party less prone to use interest group input. Some have, however, argued that members and activists are not exclusively motivated by policy (Kitschelt 1994; Norris 1995), which means that this might not constitute such a hurdle for interest groups. Second, the number of actors involved in decision making is higher in internally democratic parties (Bäck 2008; Schumacher et al. 2013). This can make it harder for interest groups to wield influence since their input has to be palatable to more actors. At the same time, members and activists constitute additional entry points. If a group has difficulties in accessing the leadership, it might find sympathetic ears among these actors. Overall this means that intra-party democracy might affect interest group influence on parties but whether it mainly facilitates or hampers it is unclear.

In order to measure intra-party democracy, I use expert survey data from Rohrschneider and Whitefield (2012) that concern the role of the party leadership and members in determining party policy.6 Higher values indicate more membership involvement. This constitutes the most up-to-date data with the best coverage of the parties examined here. Norwegian parties are, however, not covered. For these, I use similar data from Laver and Hunt’s (1992) expert survey. The results are also robust to the exclusion of the Norwegian parties.

To what extent parties have niche profiles might similarly both affect parties’ goal priorities and interest group influence. Niche parties focus on a small number of non-economic issues that other parties generally ignore (Bischof 2017). These parties might prioritise policy which can decrease the likelihood of interest group influence. I use Bischof’s (2017) procedure to calculate party nicheness. Using MARPOR data from the two latest elections preceding the PAIRDEM survey (Volkens 2020), the variable measures the emphasis a party puts on five non-economic issues compared to the other parties in the same system. The more a party is alone in emphasising few of these issues, the more niche the party is.

I furthermore control for variables that take the interest group side of the coin into account. The first is the importance interest groups ascribe to political parties when they participate in public policy processes. Interest groups that view parties as more important may put more effort into pursuing parties (Rasmussen and Lindeboom 2013), which might positively affect their chances of wielding influence. I use two questions from the PAIRDEM survey to measure this, namely how important groups find central party organisations and parliamentary party groups.7 The variables range from 0 to 3, where 0 equals not important at all and 3 very important. I use the average importance ascribed to the two faces of parties.

The second interest group control relates to the resources they have, namely whether a group employs people that deal with monitoring and commenting on public policy for at least half of their working time. This is again a question from the PAIRDEM survey. I distinguish between groups that employ zero such employees and groups that employ at least one.8 This taps into the capacity interest groups have to influence decision makers, including parties, as well as their ability to provide information that these decision makers might find useful (Mahoney 2007). Being resourceful in this way might positively affect a group’s chances of influencing a party. The results are robust if a group’s annual budget is used to measure interest group resources instead.

I furthermore control for interest group type. This to some extent captures the type of information that different groups supply. While all groups are likely to provide information on public preferences, citizen groups are found to do so more frequently than others (Flöthe 2020). They furthermore tend to be more aligned with public opinion compared to business groups (Flöthe and Rasmussen 2019). This type of information might be particularly pertinent to parties and may increase citizen groups’ likelihood of being influential. Following Rasmussen et al. (2018), I distinguish between special interest groups (business, occupational, labour, and institutional groups – coded as 0) and citizen groups that represent more diffuse interests (public interest and identity groups – coded as 1).

Lastly, I include a control for the policy area a group tries to wield influence in. I distinguish between (re-)distributive (coded as 0) and regulatory (coded as 1) policy areas. The degree of conflict and salience that are generally associated with these areas can differ which might affect groups’ ability to wield influence (Dür and de Bièvre 2007). The classification is based on Broscheid and Coen (2007). The results are similar when using a variable that also distinguishes between redistributive and distributive policy areas.

shows the regression results. The first hypothesis is supported. Interest groups are generally more likely to influence parties that they are ideologically closer to (Model 1). This also holds when the control variables and parties’ willingness to compromise on policy are included (Model 4). Figure 1 shows the predicted probabilities. Based on Model 4, the predicted probability of being somewhat influential is 0.09 when ideological proximity is relatively low (5.63). This increases to 0.28 for dyads that are close to the mean proximity value, and to 0.54 when proximity is very high (10). There is a similar albeit weaker increase for being very influential: from 0.01 when ideological proximity is lower to 0.09 when it is very high. This indicates that parties might find information from ideologically proximate groups more useful in terms of improving arguments and appealing to the groups’ constituents while also being associated with fewer potential costs. Interest groups might moreover pursue such allies to a greater extent.

Download CSVDisplay Table

The second hypothesis is also supported. Figure 2 shows the relationship between parties’ willingness to compromise on policy and the share of groups that have indicated being somewhat and very able to influence each party. Groups are generally more likely to perceive that they have influenced parties that are more willing to compromise on policy. This is supported by the regression results (Model 2), also after controlling for e.g. intra-party democracy and party nicheness (Model 3).

Figure 3 shows the predicted probabilities of being influential given parties’ willingness to compromise on policy, based on Model 3. The predicted probability of being somewhat able to influence the most compromise-averse parties is around 0.20, whereas it is 0.31 for those that are closer to the middle with a willingness score of 1.05, and 0.38 for those that are the most willing to compromise on policy. There is also a slight positive increase in the predicted probability of being very influential as parties are more willing to make policy compromises.

These findings imply that parties that are more willing to compromise might find that the benefits of using interest group input – such as appealing to the groups’ constituents – exceed the costs of e.g. slightly veering from the party’s policy goals. Realising this, interest groups might also spend more energy on trying to influence these parties. The pool of interest groups whose input these parties are likely to be interested in is furthermore found to be larger. 48% of the interest groups are ideologically close (defined as scoring at least 8 on the ideological proximity measure) to the parties that are more willing to compromise (defined as scoring at least 1 on the willingness measure). In comparison, only 29% of the interest groups are ideologically close to the parties that are relatively unwilling to compromise (defined as scoring less than 0.5 on the willingness measure). That is, a majority of groups are unlikely to be able to provide information that is of interest to more compromise-averse parties. In general, the costs of using group input for these parties may therefore exceed the benefits.

shows the results for the access mechanism.9 The effect of ideological proximity on influence seems to be mediated to some extent through whether or not a group usually talks to a party. H3a hence receives support. Ideological proximity is here held constant at 7, i.e. relatively close, and the same controls from Model 4 () are included. The total effect of ideological proximity on being somewhat influential is estimated to be 0.21, and of this 0.09 (43%) is estimated to be mediated through access. Similar results are found for being very influential. The results are stronger when it comes to the effect of parties’ willingness to compromise on policy, where I control for the same variables as in Model 3 () and hold willingness constant at 1.7, i.e. a relatively high willingness to compromise. The total effect of being somewhat influential is estimated to be 0.10, and of this 0.08 (80%) is estimated to be mediated through access. The results are again similar for being very influential. H3b is thus supported.

Download CSVDisplay Table

Access hence seems to be a relatively important mechanism that connects ideological proximity and parties’ willingness to compromise on policy to interest group influence. Parties that are more willing to make policy compromises might be more willing to grant access to interest groups, and interest groups might to a greater extent seek access to these parties. Having access can increase the opportunities groups have to voice their opinions which in turn might increase their chances of influencing the party. Similarly, interest groups might generally be more likely to seek and gain access to ideologically similar parties and therefore have more success in wielding influence.

As for the control variables, intra-party democracy is not found to be significantly associated with interest group influence. Interest groups are, on the other hand, generally less likely to influence parties with a more niche profile. As for the interest group controls, groups that attribute more importance to parties are generally found to be more likely to wield influence, as are groups that employ people who work with policy issues. The findings lend some support to the idea that citizen groups are more likely to influence parties than special interest groups but the result is only significant in one model. Interest groups are furthermore generally more likely to influence party decisions in mainly (re-)distributive policy areas compared to regulatory policy areas. Lastly, there are no consistent, significant differences between the countries (Online appendix 10). This is in line with other recent interest group influence studies (Binderkrantz and Rasmussen 2015; Rasmussen et al. 2018).

The few studies that have investigated interest groups’ ability to influence parties have focussed on the effects of interest group resources and party interest group connections (Clifton 2004; Karol 2009; Victor and Reinhardt 2018). Differences between parties in terms of their motivations to use interest group input have not been considered. This article takes this aspect into account. I theorise that interest groups are more likely to wield influence on ideologically proximate parties as well as parties that are more willing to make policy compromises. Examining party–interest group observations in six established democracies that vary with regards to state–interest group relations and where parties are generally strong and cohesive, I find support for these expectations. I moreover find that these effects seem to be mediated through the access that an interest group has to a party to a relatively large extent.

These results are also of relevance to the general literature on interest group influence. While previous research has indicated that different political decision makers might have different incentives to listen to interest groups, this article shows that considering differences between actors of the same type can be pertinent to understand when interest groups are able to wield influence as well. The characteristics of the actors that groups try to affect can be important explanatory factors in addition to issue and group characteristics. The results moreover shed further light on how parties arrive at their positions, where the impact of interest groups has been neglected. Taking interest group influence into account can be especially important when studying the positions of parties that are more willing to compromise on policy.

While parties and interest groups are independent actors, the findings here imply that they are able to engage in mutually beneficial exchanges. This may particularly be the case for more compromise-willing parties that might be able to rely on a larger interest group ‘apparatus’. More compromise-averse parties, on the other hand, seemingly have a more limited pool of relevant interest groups that they can draw on information from. This can potentially reinforce these parties’ tendency to eschew appeals to the median voter. The positions of more compromise-averse parties might to a lesser extent reflect information provided by interest groups regarding their potential political and technical consequences, unless this information is explicitly relevant to their core constituents.

The results furthermore provide some evidence that citizen groups, which are more likely to provide information on public preferences and tend to be more aligned with public opinion (Flöthe 2020; Flöthe and Rasmussen 2019), are more likely to wield influence. This can imply that the interest group input that tends to be reflected in party decisions is in line with what the public wants. The voter–party link in other words might on average be somewhat enhanced rather than distorted by interest groups.

Future research could try to incorporate the interest group influence and party position literatures further. To what extent are parties’ decisions to use interest group input for example affected by the actions of other parties? The decisions of other parties might induce or deter a party from listening to interest groups in specific instances. Another aspect that calls for attention is whether certain parties are more likely to use interest group input on certain issues. Does e.g. issue ownership affect parties’ willingness to use interest group input?

Examining interest group influence on specific issues might also be useful to shed light on when ideological proximity is more and less important. If an issue is salient to the public and a clear public opinion majority exists, interest groups that are aligned with the public and provide substantive arguments that can be used to promote this position might be influential even if they do not share the same ideological principles as a given party. The degree of conflict on an issue might furthermore be of interest. If all the interest groups active on an issue take similar stances, parties might be more prone to listen even if the groups are ideologically dissimilar.

The access mechanism could moreover be probed further, for instance by considering the ties that parties and interest groups maintain. Which ties, in terms of institutionalisation (Allern, Otjes et al. 2020), matter for influence? Lastly, investigating interest group influence on parties using other influence measures is relevant in order to establish the robustness of these results. Using for example party manifesto data or media appearances to track party position changes and combining this with PAIRDEM data would allow for a more issue-specific influence measure and could thus be one way forward.

Table 1. The effect of ideological proximity and willingness to compromise on policy on interest group influence.

Table 2. Estimated indirect and total effects of ideological proximity and willingness to compromise on policy on interest group influence with mediator access.

I am very grateful for feedback from Hanna Bäck, Jan Teorell, Elin Haugsgjerd Allern, Petrus Olander, Simon Davidsson, Johannes Lindvall, as well as the anonymous reviewers of West European Politics. An earlier version of this paper was presented at the 2020 digital ECPR conference, and I thank David Marshall, Joost Berkhout and Patrick Statsch for very helpful comments.

No potential conflict of interest was reported by the author.

Notes

1 The United States is part of the PAIRDEM survey but not included here because of missing data on intra-party democracy.

2 The categories are employers’/business/trade/industry/manufacturing groups; agricultural/farm/fisheries/forestry groups; trade unions and labour groups; occupational and professional groups; environmental/nature conservation/climate/animal welfare/wildlife groups; pro- and anti-immigration groups; humanitarian/development/foreign aid groups; religious groups.

3 The policy areas are agriculture/fisheries/forestry; economic affairs; finance and competition; taxation; customs and trade; energy; transport and infrastructure; research and technology; employment and labour market; healthcare and social affairs; consumers and food safety; regional policy; childcare; education and youth; immigration; justice; police and home affairs; environment; equal opportunities; culture; sport; information society; public administration; defence policy; development and humanitarian aid; foreign affairs.

4 Being willing to compromise on policy might lead to more moderate (extreme) positions, however. Controlling for this would hence introduce post-treatment bias.

5 I do not include these variables in the main model as they might introduce post-treatment bias.

6 Data from Giger and Schumacher (2015).

7 The wording of the question is ‘How important are the following types of actors when your organisation participates in public policy processes.’ Two of the actors that groups are asked to rate are central party organisations and parliamentary party groups.

8 The survey question distinguishes between 0, 1–5, 6–10, 11–25, 26–50, and more than 50 employees. The models become unstable if using this variable, however, as few of the respondents employ more than 1–5 policy employees.

9 There are 62 fewer observations in this analysis compared to because of missing values on the access variable. The results in are however similar if the analyses are run with this reduced sample.