A theory of organizational choice

Abstract

This article argues that niche party formation is only one of several substitutable strategies for niche activists seeking policy influence. Other organizational mechanisms are argued to be superior to political parties under certain institutional conditions. I introduce a formal model of how activists achieve policy influence by choosing to support the formation of whichever type of organization will optimally send a signal of electoral threat to mainstream politicians. Institutions determine the different access costs for the various organizational mechanisms that can be used to send this signal, and also determine the associated response costs to mainstream politicians of making concessions to different organizational mechanisms. Therefore under particular institutional circumstances that affect these two cost parameters, interest group entry may surpass party entry as the optimal strategy for activists. The predictions of the model are assessed via a nested logit model using cross-national survey data on how activists have allocated their individual support to different organizational mechanisms.

Keywords

electoral system environmental interest groups niche parties party entry

Introduction

Niche party entry is a rare phenomenon but an important one, associated with sometimes dramatic shifts in policy-making and political competition (Adams et al., 2006; Ezrow, 2010; Hug, 2001; Meguid, 2008; Somer-Topcu and Adams, 2009; Spoon, 2009). Many authors have argued that such parties emerge and succeed because of the actions of mainstream parties. However, in many cases niche parties prove ineffectual. In this article, I argue that niche activists have important alternatives to new party entry. Party entry and interest group entry are both mechanisms that activists can use to gain policy influence, but their choice to support one or the other will depend on their own resources and particularly on the institutions of the state. That is, niche organizations depend on policy-motivated activist support, and so new party entry should only occur when it is the option that maximizes activists’ policy influence. Where institutions make policy-making particularly responsive to political competition, mainstream politicians will be more susceptible to niche pressure, and will yield to even the less costly and less credible interest group mechanism. I argue that proportional representation electoral rules (PR) force policy-making to be more responsive to political competition, whereas corporatism and centralization lower this responsiveness of policy-making to political competition. I introduce a formal model of this process, showing that conditional on their resources, niche activists will support the less costly but less credible option of interest group entry when institutions induce high responsiveness, and will support the more costly but more credible option of party entry when institutions induce low responsiveness. I then evaluate this implication empirically, using data on individual participation in different organizations. I conclude that niche party entry is only one potential outcome in a larger game where activists can predict and respond to mainstream party strategies.

To begin with an illustrative anecdote; in May 1990 the rural province of Niedersachsen in Germany held regional elections.¹ As post-election bargaining commenced, it became increasingly clear that the ruling CDU/FDP coalition would be replaced by an SPD/Green coalition. But negotiations over cabinet allocations did not proceed smoothly. The Greens were profoundly sceptical of the SPD’s environmental credentials, and were especially critical of their ambiguous position on waste incineration – a policy vehemently opposed by the Greens. The Greens thus demanded control of the environmental ministry in order to be sure of obtaining their desired policy in their most important issue area. The SPD responded by appointing Monika Griefahn, leader of Greenpeace in the Federal Republic of Germany and a political independent, to be Environment Minister. The SPD were prepared to give control of the environmental ministry to an environmentalist, but it would be less politically costly for the SPD if that environmentalist came from an interest group rather than from a competing party (Lees, 2000).

The 1990 Griefahn appointment represented progress for the environmental movement in general, but the episode highlights a puzzle about the relative importance of interest groups and parties within this movement. Did Greenpeace have an influence only because of the electoral threat signalled by the Green Party? Or was the SPD really responding to a strong Greenpeace organization rather than the electoral success of the Greens? This article argues for the former interpretation, because if interest groups are a useful organizational mechanism capable of sending a signal of electoral threat, then activists should have no need to turn to the more costly option of new party formation. In many countries, activists do pursue only the interest group option, but in Germany the high degree of corporatism, even alongside PR and federalism, meant that activists with high public support could maximize their policy influence through a party mechanism.

The article proceeds as follows. The first section discusses the premises of the argument and identifies two substitutable organizational mechanisms for policy influence. A second section details the main causal mechanisms by which institutions influence organizational choice: access costs and response costs. The third section introduces a formal signalling model of this process, and draws out the implications that strong signals are unnecessary in particular countries. The fourth section presents a test of these observable implications; a nested logit estimation where individual-level characteristics determine the choice between participation versus non-participation, while system-level characteristics determine the form of participation. The fifth and final section discusses extensions and applications.

Group resources and the substitutability of different strategies for policy influence

The model of organizational substitution presented here is based on the premises that activists are policy-motivated, and that mainstream politicians set policy, and they are office-motivated. Given the psychological profile of political activists, it seems reasonable to assume that they do not choose between ‘action’ and ‘inaction’ when it comes to attempting to influence the politicians who make policy. Instead, they will almost always find a way to pursue their policy ambitions. If winning office was activists’ only goal then political parties would be their only organizational option (Aldrich, 1995; Schlesinger, 1984; Snyder and Ting, 2002), but the assumption of policy-motivated activists raises the possibility that there are many ways of achieving policy influence. There are several important points to make about these premises. First, they privilege the role of the party in government above the bureaucracy or any other actor, and so the argument relies on a simplified model of how legislation is enacted. More importantly, they imply that niche groups are successful to the extent that they influence the politicians who control policy, i.e. to the extent that their ideas are co-opted by bigger actors (Downs, 1957; Lawson and Merkl, 1988; Meguid, 2008; Przeworski and Sprague, 1986; Schwartz, 2006). Niche interest groups and niche parties simply want major parties to acknowledge their issue. The rest of this section explains why interest groups and political parties are substitutable organizational mechanisms for influencing mainstream politicians.

Mainstream political parties in mature democracies campaign on a small set of issue areas and give scant priority to new issues which offer only uncertain electoral benefits (Adams et al., 2006; Budge and Farlie, 1983; Cox and McCubbins, 2005; Lipset and Rokkan, 1967; Michels, 1912; Przeworski and Sprague, 1986; Snyder and Ting, 2002). Mainstream parties remain strategically silent about new issues, hoping instead to concentrate voter attention on wherever their own reputation is most positive. Niche parties are fundamentally different, reflecting their members’ preferences. By Meguid’s (2008) definition, niche parties are those that introduce a new and non-economic issue, and that focus only on this issue. Ezrow (2010) modifies this definition with the claim that most new issues are in fact still economic and so any distinctly non-centrist political party can be considered niche. This is the definition taken here.

Both authors emphasize that niche parties are more accountable to their policy-motivated core constituency, and so are qualitatively different from the mainstream parties that attempt to appeal to a broader range of voters (Adams et al., 2006; Denemark and Bowler, 2002; Downs, 1957; Ezrow, 2008; Hug, 2001; Richardson and Rootes, 1995; Van der Heijden, 1999). I argue that one implication of this goal-based distinction is that while mainstream parties are fundamentally different from interest groups, niche parties are fundamentally similar. Schwartz (2010) adds that party and non-party organizations can indeed be comparable even if they do not see each other as pursuing identical goals.

However, using the political goals of individual members to explain the entry of political organizations has been criticized since Olson’s (1965) argument that group formation and maintenance are primarily the result of selective incentives. He claims that this perspective is a useful alternative to traditional group theory where political groups are viewed as the result of common interests galvanized by some political disturbance. In terms of explaining participation in groups then, Olson’s (1965) framework brings selective incentives to the front of the stage, usurping the role of collective goals posited by pluralists.

Scholars following Olson (1965) have acknowledged his insight but also moved beyond it, developing organizational theories that are founded on selective incentives but continue to recognize the importance of public goods (Chong, 1991; Cyert and March, 1992; Etzioni, 1964; Frohlich et al., 1971; Keeley, 1988; Salisbury, 1969; Wilson, 1973). Many organizations have formal institutions that empower particular members as leaders or central agents who can dictate public goods goals and then distribute the necessary selective incentives (Alchian and Demsetz, 1972; Frohlich et al., 1971; Salisbury, 1969). Since political organizations – especially niche organizations – rely on voluntary support and often lack the capacity to offer strong selective incentives (Schlesinger, 1984, Schwartz, 2010, Wilson, 1973), their leaders are strongly pressured to optimize organizational choice to prevent the group from failing to achieve some goal held in high priority by many of its members. Unless a group’s goals are reasonably fully achieved, participation will decline or cease. Indeed, goal failures are one of the few things that can cause organizational change at the highest level (Cyert and March, 1992; Hirschmann, 1970; Keeley, 1988; Wilson, 1973). Participation from members is vital in determining whether a political organization can function in any meaningful way, and I argue that both niche parties and niche interest groups must attract such participation by optimally providing policy influence.

Schlesinger’s (1984) typology of political organizations includes interest groups, mainstream parties and niche parties, differentiated by three dichotomous variables:

Team production is either of private or collective goods.

Compensation for members is either direct or indirect.

Goals are reached either through markets or outside them.

Interest groups are similar to niche parties on all of these variables, but mainstream parties are different to both niche parties and to interest groups particularly on the final variable – the role of the organization in the marketplace. Mainstream parties help politicians maximize votes in the electoral market, whereas niche parties and niche interest groups are much less responsive to the median voter and instead respond to the policy wishes of their members (Ezrow, 2010; Grossman and Helpman, 2001).

With regard to the first variable, niche parties, mainstream parties and interest groups all produce non-excludable and non-rival public goods and so face similar collective action problems and arise in similar circumstances. Aldrich (1995) argues that parties begin as elite coalitions in legislatures; politicians who are like-minded on one issue will become increasingly willing to cooperate on other issues. An alternative view comes from Cox and McCubbins (2005), who argue that parties are instead a reaction to a collective action problem outside legislatures. A policy record that establishes a label or brand name for the party can aid re-election and thus qualifies as a public good for that party, but politicians can defect in order to create personal reputations and parties are a method of preventing that. Both these views emphasize that the party is an organizational method for translating a set of relatively congruent preferences into collective outcomes (Aldrich, 1995; Clark and Wilson, 1961; Cox and McCubbins, 2005; Penn, 2009; Snyder and Ting, 2002). This is fundamentally equivalent to interest groups, which are also defined around goals involving policy achievements, and also require collective organization often in the form of delegation to a central agent (Adams, 2007; Alchian and Demsetz, 1972; Chong, 1991; Frohlich et al., 1971; McCarthy and Zald, 1977; Nownes and Neeley, 1996; Olson, 1965; Van der Heijden, 1999; Walker, 1983).

As well as the provision of collective goods, interest groups and parties are similar on the dimension of compensation – rewarding their members only indirectly. The inability to sell a product for financial gain means that independent group resources are necessary if both options are to be viable for a group. Alongside a central agent punishing defection, selective benefits in the form of social prestige can accrue to contributors, and both parties and interest groups often rely on this (Chong, 1991; Clark and Wilson, 1961; Cyert and March, 1992; Frohlich et al., 1971; Salisbury, 1969; Wilson, 1973). For any given issue the presence of a sub-constituency with at least some potential members, usually well-educated and active citizens, is a prerequisite for party or interest group entry: individual participation by these activists is always crucial. It is worth noting that party formation is often more logistically costly than interest group formation, but the parallel appearance of new interest groups and parties with identical goals suggests that these minimal resource thresholds are low and are crossed at similar times (Hug, 2001; Müller-Rommel, 1989; O’Neill, 1997).

Schlesinger’s (1984) third variable illustrates a key difference between interest groups and parties in that the latter are beholden to the electoral market. By definition parties compete in elections (Downs, 1957), and if victorious can enter the government and thus set policy. Interest groups may contribute to election campaigns or endorse candidates, but by definition do not enter elections themselves (Burstein and Linton, 2002; Grossman and Helpman, 2001). Schlesinger does not differentiate niche parties and mainstream parties on this dimension, but Ezrow (2010) implies that niche parties are closer to interest groups than to mainstream parties, because any success that they enjoy in the electoral market comes from catering only to a niche sub-constituency. Thus electoral success is important in signalling to mainstream parties that an electorally significant sub-constituency is motivated by the salience of some niche issue (Imig and Tarrow, 2001; McAdam et al., 1996; Tarrow, 1989), but such success is down to activists’ (and some sub-set of voters’) policy-motivated, rather than office-motivated, behaviour.

There is considerable debate as to the precise mechanism by which interest groups exert influence, but the sending of signals to legislators is a key part of these theories. Interest groups are presumed to have better information about what policies their supporters are in favour of, and what policies will actually work (Austen-Smith, 1993; Becker, 1983; Collins, 2007; Ehrlich, 2011; Grossman and Helpman, 2001; Hall and Deardorff, 2006; Kollman, 1997; Lohmann, 1998; McIntosh, 1983; Naoi and Krauss, 2009). They assist politicians by providing them with this information, allowing busy legislators to make decisions which are better for the interest group and thus gain voters for the legislators re-election. It is this resource – effectively ‘information on public opinion’ – which is most associated with environmental interest groups (O’Neill, 1997). However, campaign contributions, public demonstrations or expert testimony can all signal different types of information, and all can be used in conjunction with this organizational mechanism to demonstrate the presence of resources that could be useful for a politician seeking election.

The biggest difference between niche parties and niche interest groups is therefore in the question of the resources that they use to achieve their goals. They must be able to offer something to mainstream politicians in exchange for policy: this constitutes an access cost. Beyond activists’ participation, the support of public opinion is often the most important resource as it reflects the overall size of the issue-cleavage (Cox, 1997; Lipset and Rokkan, 1967; O’Neill, 1997; Ordeshook and Shvetsova, 1994; Schwartz, 2010) and the quantity of this resource is important for organizational choices made by potential members. This is especially true for low-resource niche sub-constituencies, for whom access costs are likely to be the primary concern. For high-resource interests, response costs of mainstream parties are more significant. If a lack of resources –in terms of petition signatures, spaces to hold meetings, administrative skills or public opinion – makes it questionable whether an organization could sustain itself for a reasonable period of time, then those concerns are clearly paramount. Whether politicians are responsive or not is irrelevant if activists will never get the chance to send their signal. If there is a surplus of resources, however, then the question becomes not whether activists will have a chance to say their piece, but rather what mainstream politicians’ responses will be. For organizational choice then, resources matter in terms of quantity more than type, and in many cases public opinion is the foundational resource for a new political issue/cleavage. The next section discusses how institutions affect both access costs and response costs in such a way as to influence activists’ choice of organizational mechanism.

When political activists support interest-group entry over party entry

The above discussion illustrates that niche activists can use multiple organizational mechanisms to achieve a policy goal. I now examine the policy-making process that determines when one or other organizational mechanism enjoys more influence. Burstein and Linton’s (2002) meta-study illustrates that most academic work privileges the role of parties in influencing policy, because of their proximity to elected officials. Though this article cannot provide a random sample of issues or countries upon which to test the proposition that interest groups can sometimes enjoy more influence than political parties, a formal model can demonstrate the theoretical conditions under which we should expect such results, and I argue that the environmental movement constitutes an appropriate test case for evaluating such circumstances.

Meguid (2008) argues that mainstream parties always find it costly to respond to new parties, but the benefits sometimes outweigh the costs. Mainstream parties can ignore new parties, can diverge from the new party position or can converge to it. Convergence and divergence are costly strategies, and will only be pursued if one mainstream party thinks that by raising the salience of a new issue it can split the core constituency of another rival mainstream party. I argue that responding to interest groups is less costly because it raises salience without conceding ownership – hence the ingenuity of SPD leader Gerhard Schröder in Niedersachsen, in appointing a Greenpeace executive to the environmental ministry rather than allowing the Green Party to take control of the same post. By responding to interest groups, major parties can appeal for the votes of those that formed the interest groups in a way that major parties cannot do if there are already new parties more credibly claiming to own the issue. However, the response costs that major parties have to pay to acknowledge the concerns of interest groups, though usually low, can approach the costs of responding to a new party under certain institutional conditions.

PR raises the response costs associated with parties and lowers the response costs associated with interest groups, while also lowering access costs for parties. Responding to new parties is most costly in pure PR systems, where new parties can gain dangerous blackmail potential (Downs, 1957; Laver and Shepsle, 1996; Lees, 2000; Lem and Dowling, 2006). Electoral potential is a function of seats more than votes, and Hug (2001) sums up current research as near-unanimous on the point that PR incentivizes new parties by instituting lower de facto thresholds for parties to win seats. However, PR has different effects on niche groups with high and low resources. In single-member districts with first-past-the-post electoral rules (SMD), access costs for new organizations are also much higher. Voters have an incentive to strategically vote for the ‘most likely’ preferred candidate, rather than the most preferred candidate, according to Duverger’s law (Amorim-Neto and Cox, 1998; Best, 2010; Clark and Golder, 2006; Cox, 1997; Hug, 2001; Lipset and Rokkan, 1967; Ordeshook and Shvetsova, 1994). Thus lower-resource niche groups will be put off by the high access costs for parties under SMD, and since access costs are their main concern they are marginally more encouraged to support new party entry under PR.

However, since PR raises response costs for parties and lowers them for interest groups, for high-resource niche groups more concerned with response costs, PR encourages support for interest group formation. Bawn and Rosenbluth (2006) discuss this point, implying that PR systems make it less costly for mainstream party politicians to respond to niche interest groups. Under the more consensual policy-making procedures of PR, especially coalition governments, individual parties are only held accountable for policy outcomes on their ‘owned’ issues and so no-one is accountable for keeping overall spending under control. PR thus tends to create governments that can afford to be more responsive to niche interest groups. This is a crucial point because as a game theoretic model outlines in the next section, without institutions that allow parties to ignore interest groups, even activists from low-resource niche groups will not necessarily support party entry in PR systems, because their demands will be incorporated even if they send the less costly signal of interest-group formation.

I argue that PR has its greatest effect on higher resource groups in corporatist systems. Tavits (2006) argues that corporatist interest-group systems are broad and inclusive, and so establishing new interest groups or parties is unlikely to force this coalition to budge. She finds that new parties are less likely to emerge in corporatist systems. However, I argue that corporatism can sometimes induce support for new party entry if a niche group has high resources in terms of public opinion. Corporatism can be a beneficial system for any interest group or party with the resources to be guaranteed a seat at the table, but despite the rhetoric of inclusiveness, corporatism sets access cost thresholds for influence that emerging groups often find it impossible to meet (Lijphart, 1999; Siaroff, 1999; Schmitter and Lehmbruch al., 1978; Spence et al., 2000; Tavit, 2006; Wiarda, 1997). Corporatism entails a state-granted monopoly on representation for a small number of centralized and hierarchical interest group organizations. This has important implications for niche activists, because the effect of corporatism on niche issue outcomes should be conditional on the resources of niche groups in any given period.

If two interests are orthogonal and one is more established than the other, then corporatism allows the more powerful group to shut out challenging groups. Almeida and Stearns (1998) summarize the grassroots environmental movement that formed around mercury poisoning in Minamata, Japan, and argue that corporatist characteristics negatively correlate with political opportunities. This is in contrast to literature (Bernauer and Koubi, 2009; Lijphart, 1999; Scruggs, 2001) claiming that corporatism has improved environmental outcomes by creating an inclusive policy-making process, because these analyses do not take into account the relative balance of power in political competition. For example, during the 1980s when environmentalism was emerging as a national political issue in developed democracies, most corporatist states were focused on finding policy solutions that would placate more established, politically significant interests, particularly business groups and trade unions (Burchill, 2002; Schmitter and Lehmbruch, 1978; Siaroff, 1999; Wiarda, 1997). Similarly, Tavits’s (2006) causal mechanism is that corporatism cartelizes access to policy-makers, and since this is easily conceptualized as an access cost I argue that this effect only applies to low-resource niche groups who are more concerned with access costs. For high-resource groups more focused on response costs, corporatism increases the cost for mainstream parties to respond from niche interest groups, and therefore should actually force niche activists towards supporting party entry.

Finally, decentralization decreases the response costs for major party politicians making concessions to new parties and especially to new interest groups, and also decreases the access costs associated with parties and especially interest groups. When more political power is devolved to localities it becomes easier for interest groups to reach politicians (Adams, 2007; Chandler, 1993) and secondary elections often allow new parties to win footholds with less effort and boost their profile. Politicians in a decentralized system are more likely to have regular contact with interest groups, and they are also more likely to be able to effectively influence local policy. If more decisions are made at a local level and cannot be overruled, groups will have to pay lower access costs to make a difference (Chhibber and Kollman, 1998; Lago and Martinez, 2011; Naoi and Krauss, 2009; Treisman, 2007). The stronger option of supporting party formation will not be necessary if local politicians have enough decentralized authority to pre-empt formation with policy concessions. The requirements of national access costs for national parties still apply (Cox, 1997; Fillipov et al., 2004), which further dis-incentivizes party formation relative to interest-group formation. Thus decentralization reduces the response costs associated both with parties and interest groups, but shifts the calculus for both high and low-resource activists towards interest groups due to the difference in access costs. The next section formalizes the institutional arguments made here.

A formal model of organizational choice

The model presented here shows that activists are more likely to support new party entry only if major parties can credibly commit to ignoring new interest groups. Major parties can only do this when political institutions have respectively raised two key parameters: the costs to activists of accessing policy-making using a new party, and the costs to major parties of responding to interest groups. Through these two cost parameters, the model describes the situation facing political activists deciding how to organize themselves. Participation by individual activists supports either new interest group entry or new party entry. The activists’ goal is to obtain policy concessions on a particular issue from major parties, who themselves only wish to grant those concessions under conditions of high public support. This public support serves as a simple proxy for niche group resources (Clark and Golder, 2006), and indeed is often the most important of those resources (O’Neill, 1997). Political activists can attempt to communicate that public support is high, and the section above argued that mainstream parties can ignore weaker (interest group) signals under some circumstances but not under others. The signalling game constructed below formalizes the situations under which mainstream parties can separate low-resource from high-resource niche sub-constituencies. The sequence is as follows:

Nature determines ‘high’ or ‘low’ public opinion on the new issue: $N_{1} = {H, L}$

Then, Player X (activists) supports a party or interest group: $X_{1} = {P T, I G}$

Then, Players Y and Z (major parties) simultaneously choose to ignore, converge to, or diverge from, X’s desired policy on the new issue: $Y_{1} = {I_{g}, C_{O}, D_{i}} = {I_{g}, C_{O}, D_{i}}$

Payoffs accrue.

The goal for the political activists (Player X) is to obtain P, which represents policy benefits. X pays the access costs for a party (C_P ) or an interest group (C_IG ). Y and Z can then either ignore the new issue, converge towards X’s view or diverge from it (Meguid, 2008). Any strategy from Y and/or Z, other than ignoring the issue, represents a victory for X because their issue has been acknowledged as salient by the actors that make policy (Budge and Farlie, 1983). Thus X will achieve their desired payoff of (P – C_P ) or (P – C_IG ) if and only if $I_{g} \notin (Y_{1} \cup Z_{1})$

Player Y is a major party on the same side of the ideological spectrum as X and Z is the most significant ideological rival to Y. So if X were a Green party, Y would likely be the dominant centre-left party and Z the major centre-right party – vice-versa if X were a new anti-immigration party (Arzheimer and Carter, 2006; Mudde, 2007; Norris, 2005). In more general terms, Y is the major party seen as most likely to lose votes to X, and Z is the major party that stands to benefit from a decline in the fortunes of Y. This one-sided situation is represented by Y paying –C_H if it ignores high-support demands (Ezrow, 2010; Meguid, 2008). Y and Z choose between Ignore, Diverge and Converge (I_g , D_i and C_o , respectively) as reactions to the new issue position of X. They receive an electoral payoff represented by the parameter E.

Thus Y can obtain this E parameter by converging, but will receive –E from diverging.

Z can capture E by diverging, or can receive –E by converging.

To deploy D_i or C_o is costly for Y and Z. They must issue statements, prepare press conferences, disseminate information to activists and research policy positions. Thus response costs against interest groups and against parties, C_RIG or C_RP respectively, are subtracted from the payoffs of Y and Z whenever they play the strategy of D_i or C_o against an interest group or a party.

A crucial aspect of the model, then, is whether these cost parameters are equal or unequal when Y or Z are playing their D_i or C_o strategies against interest groups or parties. That is, does, C_RIG = C_RP ? I have argued that in centralized, corporatist, SMD states the costs of responding to interest groups are likely to approach the costs of responding to parties. A signalling game illustrates that this is not necessarily ‘bad’ for mainstream parties, because under such conditions they can credibly commit to ignoring new interest groups. This means that only high-resource niche sub-constituencies, who can pay the higher access costs associated with parties, will gain policy concessions in the form of increased issue salience. Alternatively, in decentralized, non-corporatist, PR states the costs associated with D_i and C_o are likely to be lower when facing an interest group than a party. As shown in the equilibria below, this improves the situation for low-resource activists, and results in less support for new party entry. To reiterate, C as a payoff parameter represents costs, and so together with P (policy benefit) and E (electoral benefit) parameters the payoff functions are as follows:

X will receive the following payoff:

P if $I_{g} \notin (Y_{1} \cup Z_{1})$ , 0 otherwise, plus:

–C_PH if $X_{1} = {P ∖ N_{1} = H}$ ; –C_PL if $X_{1} = {P ∖ N_{1} = L}$ , otherwise:

–C_IGH if $X_{1} = {I G ∖ N_{1} = H}$ , –C_IGL if if $X_{1} = {I G ∖ N_{1} = L}$ .

Y will receive the following payoff:

E if $Y_{1} = D$ , –E if $Y_{1} = C_{O}$ , 0 otherwise, plus:

–C_SP if $Y_{1} \notin {D_{i}, C_{O}} | X_{1} = {P}$ , otherwise:

–C_SIG if $Y_{1} \notin {D_{i}, C_{O}} | X_{1} = {I G}$ , otherwise –C_H if N₁ = {H}.

Z will receive the following payoff:

E if $Z_{1} = C_{O}$ , –E if $Z_{1} = D_{i}$ , 0 otherwise, plus:

–C_SP if $Z_{1} \notin {D_{i}, C_{O}} | X_{1} = {P}$ , otherwise:

–C_SIG if $Z_{1} \notin {D_{i}, C_{O}} | X_{1} = {I G}$ , otherwise:

–C_L if N₁ = {L}, otherwise 0.

The solution concept applied here is sequential equilibrium (Myerson, 1991), and by making reasonable assumptions about these parameters, comparative statics can be drawn out, as detailed in Appendix A.² I assign numerical payoffs and highlight two substantively interesting equilibria to illustrate the institutional argument. To go into more detail, we can use this model to construct a plausible facsimile of a decentralized, non-corporatist, PR country.³ Under numerical payoffs that fit these criteria, D_i is a dominant strategy for Z and C_o is a dominant strategy for Y. This sustains a pooling equilibrium with X choosing to support new interest-group entry whether public opinion is high or low. Hence, under a plausible set of assumptions, interest-group entry may find support even when public support is high – because major parties do not find it costly to respond to interest groups and so cannot credibly commit to ignoring them, and supporting party entry is more costly for X, so no actor has an incentive to defect from the pooling equilibrium. Activists will thus devote their participation to supporting new interest groups. To the extent that they do turn their attention to political parties in PR states then, it will be the result of major parties committing to ignore interest-group signals, which is easier in states with PR but which are also centralized and corporatist. Consider a country at the other extreme, that is, a highly corporatist and centralized SMD state. In such a country, a plausible preference profile would place much higher costs on responding to interest groups.⁴ At low levels of public support then, Y and Z are indifferent between ignoring parties and interest groups. Here, a separating equilibrium does indeed exist where with high public opinion, X supports party entry and under low public opinion X supports interest-group entry.

This simple game has illustrated the key intuition – parties and interest groups can be seen as signals of strength, and they will be substituted for one another depending on how the institutions of the state incentivize major parties to ignore one signal or another. Political institutions affect the costs and benefits of different organizational strategies not only through explicit barriers like vote thresholds and petition requirements, but also through their effect on political competition. The two equilibria also produce counter-intuitive predictions: in decentralized, non-corporatist PR states, if the policy concessions given to interest groups and parties are identical, and parties are more costly to form, then interest groups are the optimal strategy. The higher incidence of party entry in PR states (Best, 2010; Hug, 2001; Schofield and Caballero, 2011) should thus be explained not just by the effect of PR on parties, but also whether there are other institutions that have increased the cost of responding to interest groups. Without such institutions, activists should never have to form parties in PR systems. Their demands will be responded to even if they send the less costly signal of interest group formation. The next section demonstrates the empirical evidence that activists engage in substitutive behaviour when choosing which types of organizations to support.

Empirical evidence for substitution

The key implication of the preceding argument is that activists on niche issues should support different organizational mechanisms depending on their resources and the institutions of the state. The signalling model highlighted two equilibria that can be translated into observable implications. These are hypotheses one and two:

Hypothesis 1: In countries that are relatively decentralized, non-corporatist and have relatively proportional electoral rules, both high and low-resource niche interests will form interest groups.

Hypothesis 2: In countries that are relatively centralized, corporatist and have relatively dis-proportional electoral rules, low-resource niche interests will form interest groups whereas high-resource interest groups will form parties.

In order to test these implications, I use cross-national data on environmental concerns from the World Values Survey (WVS). This survey was conducted over a wide variety of developed democracies and included a set of questions on whether citizens were members of an environmental interest group or not, and which party was their first choice when they were deciding how to cast their vote.⁵ These questions were coded into a nominal dependent variable with four categories:

None: The respondent was not a member of an environmental interest group, nor did they mention a Green party as their first choice.

Interest Group: The respondent was a member of an environmental interest group, but did not mention a Green party as their first choice.

Party: The respondent was not a member of an environmental interest group, but did mention a Green party as their first choice (The USA, Japan and Norway have no observations on this category).

Both: The respondent was a member of an environmental interest group and mentioned a Green party as their first choice.

The sample of democratic countries where these questions were asked is shown in Table 1, which contains the count of each value of the dependent variable in that country.

Table 1.

Country	Years included	None (0)	Interest group (1)	Party (2)	Both (3)	Total
Australia	1995, 2005	8,612	1,448	628	320	11,008
Canada	2000	4,756	420	24	12	5,212
Finland	1996, 2005	4,988	464	556	116	6,124
Germany	1997	3,868	256	856	172	5,152
Japan	2000, 2005	6,248	244	0	0	6,492
New Zealand	1998, 2004	3,444	824	80	48	4,396
Norway	1996	3,528	272	0	0	3,800
Spain	1995, 2000	4,944	292	96	4	5,336
Sweden	1996, 1999	4,680	620	484	148	5,932
Switzerland	1996	2,388	556	52	60	3,056
USA	1995, 1999	7,084	1,856	0	0	8,940
Total	1989-2005	54,540	7,252	2,776	880	65,448

Category 2 in particular is a weak proxy for the phenomena of interest: supporting a political party. Ideally this category would include only active party volunteers, but this measure is not available for enough countries to test institutional hypotheses. Indeed, the sample here is still dictated by data availability and so includes only three pure PR countries (Norway, Finland and Sweden) and only two unitary countries (Norway and Sweden). The analysis here must therefore be treated as preliminary, but the key institutional hypotheses can still be tested. Each country’s institutional variables were taken from the Binghamton Institutions and Elections Project, with corporatism scores taken from the index of Siaroff (1999). These institutional variables are separated out into choice-specific variables.

As per earlier discussion, I operationalize resources as the overall level of public opinion in a country. Though this neglects financial resources and informational resources that can be extremely important for interest groups, using the environmental movement as a test case suggests that public opinion is the most appropriate comparative measure. No environmental interest group in this period focused purely on financial contributions to influence politicians (Burchill, 2002; Müller-Rommel, 1989; O’Neill, 1997; Richardson and Rootes, 1995) and instead electoral threats were sent more conventionally through signalling the extent and direction of public opinion. The WVS asks respondents, on a four-point scale, how strongly they agree with a statement about their willingness to pay more taxes for environmental protection.⁶ The mean value of this variable in a given country-year was used as a measure of the extent of environmental public opinion, and is displayed in Table 2. As well as these key variables, the effective number of electoral parties at the most recent previous election is also controlled for, to the extent that crowded systems could dis-incentivize new party formation (Clark and Golder, 2006; Denemark and Bowler, 2002; Kitschelt, 1989; Norris, 2005; Van der Brug et al., 2000). The descriptive statistics for the institutional/system-level variables are listed in Table 2.

Table 2.

Country	PR	Siaroff	Opinion index	Federal	ENEP
Australia	0	3	2.709	1	2.911
Canada	0	1.875	2.610	1	3.770
Finland	1	4.375	2.552	1	5.320
Germany	0	4.125	2.745	1	3.740
Japan	0	3.625	2.589	1	3.794
New Zealand	0	2.375	2.528	1	4.146
Norway	1	4.625	2.877	0	4.740
Spain	0	2	2.635	1	3.237
Sweden	1	4.625	3.051	0	4.110
Switzerland	0	4.375	2.326	1	6.820
USA	0	2.125	2.603	1	2.112
Total	0.242	3.252	2.668	0.851	3.755

According to hypotheses 1 and 2, I should expect to see that decentralization, non-corporatism and PR should increase the probability that a respondent would be in category 1 (interest group participant) of the dependent variable rather than in category 2 (political party participant). High movement resources, operationalized as aggregate levels of public opinion, should increase the probability that a given respondent would be in category 2 of the dependent variable rather than in category 1, especially in corporatist, centralized, SMD systems. However, these institutions should not affect the overall probability that a respondent is in category 1, 2 or 3, versus category 0. Category zero is qualitatively different from the other categories, as it could include inactive environmentalists, non-environmentalists or anti-environmentalists. Yet in order to exclude the inappropriate respondents from the analysis predicting the choice between 1, 2 and 3, we must also predict the choice between zero and non-zero. Therefore including all survey respondents in the country, and controlling for factors that lead to environmental political participation is an important part of the research design.

An extensive literature on political participation argues that education leads to both political interest and political efficacy, and that most environmental activists are well educated (McAdam, 1989; Rosenstone and Hansen, 1993; Verba et al., 1978). Income, although highly correlated with education, provides for the free time to get involved in political activism and so should also be controlled for (Verba et al., 1978). Activists are also disproportionately male and disproportionately older, which is often connected to feelings of political efficacy, though Green activists tend to be younger and so age⁁2 is included alongside age to allow for these potential non-linearities (Müller-Rommel, 1989). Importantly, I also control for post-material sentiment (Dalton, 2002; Inglehart, 1977) in order to isolate not only those activists with the opportunity to make organizational choices, but also the willingness. Post-materialism is an opinion index where higher values indicate a higher propensity to care about issues such as environmental protection that are not considered to be economic priorities. The final variable I include in this stage is the year of the survey, to account for potential time trends influencing levels of activism. The model thus controls for the influence of education (coded in three categories from low to high), sex (coded male = 1, female =0), age, year, income (coded in ten categories from low to high) and strength of post-materialist opinion (coded in five categories from low to high) in terms of shifting people from category 0 to categories 1, 2 or 3. All of these variables were taken from the WVS, and the means for these individual-level variables are shown in Table 3.

Table 3.

Means of individual-level variables.

Country	Education	Gender	Income	Post- materialism	Age
Australia	2.244	0.514	5.343	2.639	45.612
Canada	2.021	0.563	5.117	2.741	46.105
Finland	1.841	0.501	4.117	2.627	44.791
Germany	1.981	0.530	5.278	2.657	43.590
Japan	2.179	0.501	5.011	2.252	47.885
New Zealand	2.293	0.520	6.486	2.344	46.390
Norway	2.044	0.493	5.559	2.458	43.278
Spain	1.534	0.490	4.449	2.496	45.474
Sweden	2.069	0.469	5.199	2.629	45.608
Switzerland	1.897	0.488	5.780	2.741	46.162
USA	2.248	0.530	6.097	2.367	45.446
Total	2.064	0.511	5.299	2.534	45.549

I estimate the effect of institutions on the dependent variable using a nested logit model.⁷ The nested logit allows for a discrete choice to be modelled in a way that relaxes the ‘independence of irrelevant alternatives’ (IIA) assumption that all alternatives are proportionately substituted for one another (Atherly et al., 2004; Heiss, 2002; Hensher and Greene, 2002; Train, 2003). Relaxing this assumption is appropriate for the situation here, where categories 1, 2, 3 are assumed to be substitutes for each other, but category 0 is a qualitatively different option. The nested logit model thus reflects the argument in this article that the set of individual-level variables listed above will predict the choice between ‘no participation’ and ‘participation’, and institutional variables predict the patterns of substitution between different forms of participation/action: ‘interest group’, ‘party’ or ‘both’.

The coefficients from the random-utility-consistent nested logit model are displayed in Table 4. The dis-similarity parameter (an inverse measure of the correlation of error terms associated with alternatives within a nest) does not allow us to reject the assumption that this choice can be modelled in a way that assumes IIA, but does show that the model is consistent with utility maximizing behaviour from respondents at all ranges of the independent variables (Train, 2003). Due to the data structure and non-linear estimation, the coefficients and standard errors in such a model do not adequately convey the information relevant for testing the hypotheses. More important for drawing inference are the predicted probabilities.

Table 4.

A Nested Logit Model of Organizational Choice.

‘Allocation’	Variable	Coefficient	Standard Error
Action (1\|2\|3)	Education Category	0.422***	0.032
	Male	0.204***	0.043
	Income Category	0.027***	0.009
	Post-Materialism	0.314***	0.019
	Age	−0.008	0.008
	Age⁁2	0	0
	Year	−0.051***	0.007
	Dis-similarity Parameter	0.949	0.212
I.G. (1)	PR	0.009	0.133
	Federal	−0.344	0.291
	Corporatism	−0.175**	0.068
	Resources	−1.019**	0.439
	Constant	101.391***	14.32
Party (2)	PR	0.239	0.159
	Federal	3.67***	0.635
	Corporatism	0.882***	0.148
	Resources	6.295***	1.403
	ENEP	0.253**	0.118
	Constant	72.61***	14.92
Both (3)	PR	0.01	0.235
	Federal	2.496***	0.501
	Corporatism	0.814***	0.158
	Resources	3.821***	0.883
	Constant	80.589***	14.439
Observations = 65448		N = 16362
Wald χ² (20) = 763.38***		Log-Likelihood = -8835.3703

*** = p < 0.01, ** = p < 0.05, * = p < 0.1.

Therefore the following tables show predicted probabilities, broken down by institutional context. Table 5 uses out-of-sample predictions to illustrate hypotheses 1 and 2. I compare the predicted actions of activists with high and low resources under institutions that generate first pooling and then separating equilibria according to the game above. In order to show that even institutions that are usually thought to encourage party entry – such as federalism and PR – only have that effect in countries where other institutions allow major parties to commit to ignoring interest groups, I focus on testing the effect of corporatism in federal and PR countries. Under low corporatism, which I argue helps generate pooling, the predicted probabilities do indeed show that, for movements with both low and high resources, the most likely organizational mechanism was an interest group. Under high levels of corporatism, however, for movements with low resources the most likely organizational mechanism is still an interest group, but movements with high resources will form parties. Thus it is clear from Table 5 that even under federalism and PR, countries experience high levels of environmental party activity only if they simultaneously feature high levels of corporatism and high levels of pro-environmental public opinion.⁸

Table 5.

Conditional Probabilities of Organizational Choice.

Conditional Out-of-Sample Probabilities:
Example of Pooling to Separating Equilibrium
	Low Resources (Index = 2.4)		High Resources (Index = 2.8)
	Federal = 1,	Federal = 1,	Federal = 1,	Federal = 1,
	PR = 1,	PR = 1,	PR = 1,	PR = 1,
Pr(DV)	Corporatism = 2	Corporatism = 4	Corporatism = 2	Corporatism = 4
1 (IG)	0.968	0.796	0.685	0.212
2 (PTY)	0.019	0.131	0.254	0.653
3 (BOTH)	0.012	0.073	0.06	0.135

Table 6 goes on to interpret the coefficients more concretely for other institutions showing specific countries using in-sample rather than out-of-sample predicted probabilities. So, for example, New Zealand and Germany have similar levels of public opinion, but Germany has much higher levels of corporatism and also party activity, just as predicted by hypothesis 2. The effect of PR on increasing the likelihood that activists will support party entry is highest in highly corporatist states with lower resources, reinforcing the conclusions of the formal model. These differences are statistically significant at the 95 percent level.⁹

Table 6.

Absolute In-Sample Probabilities of Organizational Choice, with Mean 95% CIs.

Country	Average Prob (None)	Average Prob (IG)	Average Prob (Party)	Average Prob (Both)
Australia	0.827 (0.807-0.846)	0.112 (0.098-0.125)	0.045 (0.038-0.052)	0.017 (0.013-0.021)
Canada	0.849 (0.832-0.865)	0.140 (0.124-0.155)	0.008 (0.005-0.01)	0.004 (0.002-0.005)
Finland	0.814 (0.788-0.841)	0.076 (0.061-0.09)	0.091 (0.073-0.108)	0.019 (0.012-0.026)
Germany	0.743 (0.715-0.771)	0.074 (0.06-0.88)	0.143 (0.123-0.163)	0.040 (0.03-0.049)
Japan	0.874 (0.859-0.89)	0.084 (0.073-0.09)	0.029 (0.022-0.035)	0.013 (0.01-0.016)
New Zealand	0.850 (0.834-0.867)	0.137 (0.122-0.152)	0.009 (0.006-0.011)	0.004 (0.003-0.005)
Norway	0.859 (0.84-0.878)	0.107 (0.09-0.124)	0.024 (0.018-0.03)	0.010 (0.006-0.013)
Spain	0.864 (0.846-0.882)	0.115 (0.1-0.131)	0.016 (0.01-0.021)	0.005 (0.003-0.007)
Sweden	0.833 (0.812-0.855)	0.082 (0.69-0.095)	0.066 (0.052-0.079)	0.019 (0.012-0.025)
Switzerland	0.827 (0.801-0.853)	0.132 (0.11-0.155)	0.030 (0.019-0.041)	0.011 (0.005-0.016)
USA	0.832 (0.815-0.85)	0.156 (0.139-0.017)	0.007 (0.004-0.009)	0.005 (0.003-0.007)
Total	0.833 (0.813-0.854)	0.111 (0.096-0.126)	0.042 (0.034-0.051)	0.013 (0.009-0.017)

The findings in Tables 5 and 6 suggest that niche activists see interest groups and parties as substitutable strategies for policy influence, and that they choose between those options based on which is institutionally optimal.

Conclusion

I have argued that interest groups and parties are substitutable organizational mechanisms for activists seeking policy influence. The institutions of a state play a large role in determining which of these options is optimal for the purpose, because institutions directly affect access costs (to activists) and response costs (to mainstream parties). A formal signalling model captures this argument, and results were confirmed on cross-national survey data regarding participation in different types of environmental organizations. Table 7 summarizes the argument of the article about the effect of institutions on organizational choice.

Table 7.

Institutional Conclusions In The Formal Model.

	PR Elections	Decentralization	Corporatism
Effect On Interest Groups?	Decreases response costs.	Decreases access costs, decreases response costs.	Increases access costs, greatly increases response costs
Effect On Parties?	Greatly decreases access costs, greatly increases response costs.	Decreases access costs, decreases response costs.	Greatly increases access costs, increases response costs
Effect On Overall Organizational Choice, Given Higher/Lower Public Opinion?	Lower Public Opinion: Parties become slightly less likely relative to under SMD.	Lower Public Opinion: Parties become less likely relative to under centralization.	Lower Public Opinion: Parties become less likely relative to under pluralism.
	Higher Public Opinion: Parties become less likely relative to under SMD.	Higher Public Opinion: Parties become less likely relative to under centralization.	Higher Public Opinion: Parties become much more likely relative to under pluralism.

However, several important extensions remain. These results could be challenged either by relaxing the assumption of identical policy concessions or of the desire of activists to achieve policy at the expense of office. This second possibility points to another critique of this model. The frequent occurrences of staggered entry (interest groups which shortly afterwards became parties) as well as the presence of tandem operations between parties and interest groups could also be taken as evidence that modelling them as substitutable strategies is to mislabel an evolutionary process. This alternative viewpoint, though, cannot account for cross-national variation in relative allocations of activist resources to either parties or interest groups, which is indeed directly predicted by a theory of substitution. An important next step will be to examine actual instances of new party entry. The total number of parties in a system should reflect not only the number of cleavages and the permissiveness of electoral institutions, but also the influence of institutions that increase or decrease the incentives for activists to support interest groups rather than parties. By examining multiple cleavages rather than just the issue of the environment, it would be possible to illustrate the applicability of this theory of organizational choice to predicting party system size (Best, 2010; Cox, 1997; Ordeshook and Shvetsova, 1994).

I argue that the institutional theory of organizational substitution outlined here helps explain cross-national variation in why different interests have emerged in different organizational forms and gone on to have different levels of success. The way politics can proceed as organized in representative democracies can serve to amplify the influence of a small active constituency – provided they choose their organizational strategy right. Parties are clearly not the only translation mechanism for turning public opinion into public policy, and a game theoretic analysis of the effects of institutions on political competition helps explain when one organizational type will be chosen over another as the optimal translation mechanism for a new issue. As well as allowing for better prediction of when new parties emerge, this also highlights the short-sightedness of major parties that prize stability and ignore new issues, especially if all major parties are biased towards ignoring issues on one side of the ideological spectrum. The analysis here implies that suppressing environmental issues, or refusing to cooperate with far-right groups, may only drive those activists towards different and more radical tactics. Thus in order to understand the optimal strategies for both activists and their major party opponents, it is important to identify the full range of options available to such actors

Footnotes

Acknowledgements

I am grateful to Andrea Berg, Dave Clark, Olga Shvetsova, Florence So and four anonymous reviewers for their helpful suggestions. Any remaining errors are the sole responsibility of the author.

Funding

This research received no specific grant from any funding agency in the public, commercial or not-for-profit sectors.

Notes

Appendix A

Equilibrium A : {IG, IG; C_O, C_O; D_i, D_i }. Given that divergence benefits Z and convergence hurts Z, and vice versa for Y, D_i is a dominant strategy for Z if E > C_SP > 0, and C_o is a dominant strategy for Y if E > C_SP and C_SIG < C_L , and both of these relations are true by assumption in this scenario. Given this, the dominant strategy for X is always to form an interest group if C_IGH < C_PH and C_IGL < C_IGH , which again are true by assumption. If the initial nature moved is assumed to be 50/50, which I assume for simplicity, then both Y and Z are updating their beliefs in accordance with Bayes’s rule if they believe that both high and low are equally likely after seeing the IG signal from X. One example of numerical payoffs that fit these criteria and that were used to calculate this equilibrium are P = 3, C_PL = 3, C_IGL = 2, C_PH = 1, C_IGH = 0 for X and C_H = 1, E = 2, C_SP = 1, C_SIG = 0, C_L = 0 for Y and Z.

Equilibrium B : {P, IG; C_O, [α(I_g)+ (1-α)(C_O)]; D_i, I_g} . If C_SP = C_SIG then Y’s mixed strategy must include α such that X is indifferent between P and IG if N = {L}. Such a value of α is rational for Y if a number of conditions hold. First, E – C_SIG = C_L . Second, X never plays IG after high public support, and Z diverges from parties but ignores interest groups. This hypothesized equilibrium means all players know the location of their decision nodes if X = {P}, but if X = {IG} then Y is playing a mixed strategy, so Z updates in accordance with Bayes’ rule: of the decision nodes after {Low, IG}, Y has played I_g with probability α and C_O with probability (1-α), and after this update Z will have no incentive to defect if E = C_SIG . Again given that N = {0.5H, 0.5L} and numerical payoffs P = 4, C_PL = 3, C_IGL = 2, C_PH = 1, C_IGH = 0 for X, and C_H = 3, E = 2, C_SP = 2, C_SIG = 2, C_L = 0 for Y and Z, then α = 0.25.

Appendix B

Resources: On a scale from 1 to 4, respondents went from strongly disagreeing (1) to strongly agreeing (4) with the following statement: ‘I would agree to an increase in taxes if the extra money were used to prevent environmental pollution’.

Interest groups: Respondents were coded as being members of an environmental interest group if they answered ‘yes’ or ‘active’ or ‘inactive’ to whichever of the following questions was asked in their country (though some states did not include the ‘animal rights’ label in their wording):

Please look carefully at the following list of voluntary organizations and activities and say…which, if any, do you belong to?

[Conservation, environment, animal rights groups]

Now I am going to read out a list of voluntary organizations; for each one, could you tell me whether you are a member, an active member, an inactive member or not a member of that type of organization?

[Environmental organization]

Parties: The following parties were coded as Green, by country, in response to the question: ‘If there were a national election tomorrow, for which party on this list would you vote?’

Australia – The Greens

Canada – The Greens

Finland – Green League

Germany – The Greens and the Democratic Ecologist Party

Japan – New Party Sakigake

New Zealand – The Greens

Norway – None

Spain – The Greens

Sweden - Miljopartiet

Switzerland – Swiss Greens and Swiss Ecology

USA – None

References

Adams

Clark

Ezrow

Glasgow

(2006) Are niche parties fundamentally different from mainstream parties? The causes and electoral consequences of Western European parties policy shifts, 1976–98. American Journal of Political Science 50: 513–529.

Adams

(2007) Citizen Lobbyists. Philadelphia, PA: Temple University Press.

Alchian

Demsetz

(1972) Production, information costs, and economic organization, American Economic Review 62: 777–795.

Aldrich

(1995) Why Parties? Chicago, IL: University of Chicago Press.

Almeida

Brewster

Stearns L

(1998) Political opportunities and local grassroots environmental movements. Social Problems 45: 37–60.

Amorim Neto

Cox

(1998) Electoral institutions, cleavage structures, and the number of parties. American Journal of Political Science 41: 149–174.

Arzheimer

Carter

(2006) Political opportunity structures and right-wing extremist party success. European Journal of Political Research 45: 419–433.

Atherly

Dowd

Feldman

(2004) The effect of benefits, premiums, and health risks on health plan choice in the Medicare program. Health Services Research 39: 849–866.

Austen-Smith

(1993) Information and influence: Lobbying for agendas and votes. American Journal of Political Science 37: 799–833.

10.

Bawn

Rosenbluth

(2006) Short versus long coalitions. American Journal of Political Science 50: 250–265.

11.

Becker

(1983) A theory of competition among pressure groups for political influence. Quarterly Journal of Economics 98: 371–400.

12.

Bernauer

Koubi

(2009) Effects of political institutions on air quality. Ecological Economics 68: 1355–1365.

13.

Best

(2010) Increasing irrationality? The equilibrium relationships between electoral and legislative party system size. Electoral Studies 29: 105–116.

14.

Budge

Farlie

(1983) Explaining and Predicting Elections: Issues Effects and Party Strategies in Twenty-Three Democracies. London: George, Allen and Urwin.

15.

Burchill

(2002) The Evolution of Green Politics: Development and Change Within European Green Parties. London: Earthscan.

16.

Burstein

Linton

(2002) The impact of political parties, interest groups, and social movement organizations on public policy. Social Forces 81: 381–408.

17.

Chandler

(ed.) (1993) Local Government in Liberal Democracies. London: Routledge.

18.

Chhibber

Kollman

(1998) Party aggregation and the number of parties in India and the United States. American Political Science Review 92(3): 329–342.

19.

Chong

(1991) Collective Action and the Civil Rights Movement. Chicago, IL: University of Chicago Press.

20.

Clark

Golder

(2006) Rehabilitating Duverger’s theory. Comparative Political Studies 39: 679–708.

21.

Clark

Wilson

(1961) Incentive systems: A theory of organizations. Administrative Science Quarterly 6: 129–166.

22.

Collins

Jr (2007) Lobbyists before the U.S. Supreme Court: Investigating the influence of amicus curiae briefs. Political Research Quarterly 60: 55–70.

23.

Cox

(1997) Making Votes Count. Cambridge: Cambridge University Press.

24.

Cox

McCubbins

(2005) Setting the Agenda. Cambridge: Cambridge University Press.

25.

Cyert

March

(1992) A Behavioural Theory of the Firm. New York: Wiley–Blackwell.

26.

Dalton

(2002) Citizen Politics. London: Chatham House.

27.

Denemark

Bowler

(2002) Minor parties and protest votes in Australia and New Zealand. Electoral Studies 21: 47–67.

28.

Downs

(1957) An Economic Theory of Democracy. New York: Harper and Row.

29.

Ehrlich

(2011) Access Points: An Institutional Theory of Policy Bias and Policy Complexity. Oxford: Oxford University Press.

30.

Etzioni

(1964) Modern Organizations. Englewood Cliffs, NJ: Prentice-Hall.

31.

Ezrow

(2008) On the inverse relationship between votes and proximity for niche parties. European Journal of Political Research 47: 206–220.

32.

Ezrow

(2010) Linking Citizens and Parties. Oxford: Oxford University Press.

33.

Filippov

Ordeshook

Shvetsova

(2004) Designing Federalism. Cambridge: Cambridge University Press.

34.

Frohlich

Oppenheimer

Young

(1971) Political Leadership and Collective Goods. Princeton, NJ: Princeton University Press.

35.

Greene

(2003) Econometric Analysis. Upper Saddle River NJ: Pearson Education.

36.

Grossman

Helpman

(2001) Special Interest Politics. Cambridge, MA: MIT Press.

37.

Hall

Deardorff

(2006) Lobbying as legislative subsidy. American Political Science Review 100: 69–84.

38.

Heiss

(2002) Structural choice analysis with nested logit models. STATA Journal 2: 227–252.

39.

Hensher

Greene

(2002) Specification and estimation of the nested logit model. Transportation Research Part B 36: 1–17.

40.

Hirschmann

(1970) Exit, Voice, and Loyalty: Responses to Decline in Firms, Organizations, and States. Cambridge, MA: Harvard University Press.

41.

Hug

(2001) Altering Party Systems. Ann Arbor, MI: University of Michigan Press.

42.

Imig

Tarrow

(eds) (2001) Contentious Europeans: Protest and Politics in an Emerging Polity. Lanham: Rowman and Littlefield.

43.

Inglehart

(1977) The Silent Revolution. Princeton, NJ: Princeton University Press.

44.

Keeley

(1988) A Social Contract Theory of Organizations. Notre Dame, IN: University of Notre Dame Press.

45.

Kitschelt

(1989) The Logics of Party Formation. Ithaca, NY: Cornell University Press.

46.

Kollman

(1997) Inviting friends to lobby. American Journal of Political Science 41: 519–544.

47.

Lago

Martínez

(2011) Why new parties? Party Politics 17(3): 3–20.

48.

Laver

Shepsle

(1996) Making and Breaking Governments. Cambridge: Cambridge University Press.

49.

Lawson

Merkl

(eds) (1988) When Parties Fail. Princeton, NJ: Princeton University Press.

50.

Lees

(2000) The Red–Green Coalition in Germany. Manchester: Manchester University Press.

51.

Lem

Dowling

(2006) Picking their spots. Political Research Quarterly 59: 471–480.

52.

Lijphart

(1999) Patterns of Democracy. New Haven, CT: Yale University Press.

53.

Lipset

Rokkan

1967 Party Systems and Voter Alignments. Toronto: Free Press.

54.

Lohmann

(1998) An informational rationale for the power of special interests. American Political Science Review 92: 809–827.

55.

McAdam

(1989) The biographical consequences of activism. American Sociological Review 54: 744–760.

56.

McAdam

McCarthy

Zald

(eds) (1996) Comparative Perspectives on Social Movements. Cambridge: Cambridge University Press.

57.

McCarthy

Zald

(1977) Resource mobilization and social movements. American Journal of Sociology 82: 1212–1241.

58.

McIntosh

(1983) Private use of a public forum: A long range view of the dispute processing role of courts. American Political Science Review 77: 991–1010.

59.

Meguid

(2008) Party Competition Between Unequals. Cambridge: Cambridge University Press.

60.

Michels

(1912) Political Parties. New York: Free Press.

61.

Müller-Rommel

(ed.) (1989) New Politics in Western Europe. Boulder, CO: Westview.

62.

Mudde

(2007) Populist Radical Right Parties in Europe. Cambridge: Cambridge University Press.

63.

Myerson

(1991) Game Theory. Cambridge, MA: Harvard University Press.

64.

Naoi

Krauss

(2009) Who lobbies whom? American Journal of Political Science 53: 874–892.

65.

Norris

(2005) Radical Right. New York: Cambridge University Press.

66.

Nownes

Neeley

(1996) Public interest group entrepreneurship and theories of group mobilization. Political Research Quarterly 49: 119–146.

67.

Olson

(1965) The Logic of Collective Action. Cambridge MA: Harvard University Press.

68.

ONeill

(1997) Green Parties and Political Change in Contemporary Europe. Aldershot: Ashgate.

69.

Ordeshook

Shvetsova

(1994) Ethnic heterogeneity, district magnitude, and the number of parties. American Journal of Political Science 38: 100–123.

70.

Penn

(2009) A model of far-sighted voting. American Journal of Political Science 53: 36–54.

71.

Przeworski

Sprague

(1986) Paper Stones. Chicago, IL: University of Chicago Press.

72.

Richardson

Rootes

(eds) (1995) The Green Challenge. New York: Routledge.

73.

Rosenstone

Hansen

(1993) Mobilization, Participation, and American Democracy. New York: Macmillan.

74.

Salisbury

(1969) An exchange theory of interest groups. Midwest Journal of Political Science 13: 1–32.

75.

Schlesinger

(1984) On the theory of party organization. Journal of Politics 46: 369–400.

76.

Schmitter

Lehmbruch

(eds) (1978) Trends Toward Corporatist Intermediation. London/Beverly Hills: Sage.

77.

Schwartz

(2006) Party Movements in the USA and Canada. New York: Lanham, Rowman and Littlefield.

78.

Schwartz

(2010) Interactions between social movements and US political parties. Party Politics 16: 587–607.

79.

Schofield Cabellero

(eds) (2011) The Political Economy of Democratic Institutions. New York: Springer

80.

Scruggs

(2001) Is there really a link between neo-corporatism and environmental performance? British Journal of Political Science 31: 686–692.

81.

Siaroff

(1999) Corporatism in 24 industrial democracies. European Journal of Political Research 36: 175–205.

82.

Snyder

Ting

(2002) An informational rationale for political parties. American Journal of Political Science 46: 90–110.

83.

Somer-Topcu

Adams

(2009) Moderate now, win votes later. Journal of Politics 71: 678–692.

84.

Spence

Jeurissen

Rutherfoord

(2000) Small business and the environment in the UK and the Netherlands. Business Ethics Quarterly 10: 945–965.

85.

Spoon

J-J

(2009) Holding their own. Party Politics 15: 615–634.

86.

Tarrow

(1989) Struggle, Politics, and Reform: Collective Action, Social Movements, and Cycles of Protest. Ithaca, NY: Cornell University Press.

87.

Tavits

(2006) Party system change. Party Politics 12: 99–119.

88.

Train

(2003) Discrete Choice Methods with Simulation. Cambridge: Cambridge University Press.

89.

Treisman

(2007) The Architecture of Government. Cambridge: Cambridge University Press.

90.

Van der Brug

Fennema

Tillie

(2000) Anti-immigrant parties in Europe. European Journal of Political Research 37: 77–102.

91.

Van der Heijden

(1999) Environmental movements, ecological modernisation and political opportunity structures. Environmental Politics 1: 199–221.

92.

Verba

Nie

Kim

(1978) Participation and Political Equality. Cambridge: Cambridge University Press.

93.

Verba

Lehman Schlozman

Brady

(1995) Voice and Equality. Cambridge, MA: Harvard University Press.

94.

Walker

(1983) The origin and maintenance of interest groups in America. American Political Science Review 77: 390–406.

95.

Wiarda

(1997) Corporatism and Comparative Politics. Armonk, NY: M. E. Sharpe.

96.

Wilson

(1973) Political Organizations. New York: Basic Books.