Political Party Candidate Selection: Methods and Consequences

Reuven Y Hazan & Gideon Rahat. Handbook of Party Politics. Editor: Richard S Katz & William Crotty. Sage Publications, 2006.

Whatever the electoral formula used in elections, candidate selection is one of the first things that political parties must do before they take place. Those who are eventually elected to office will be the successful candidates that the parties previously selected, and they are the ones who will determine much of how the party looks and what it does. That is, the results of the candidate selection process will affect the party a long time after the election itself is over.

Candidate selection is, according to Ranney (1981: 75), the ‘process by which a political party decides which of the persons legally eligible to hold an elective office will be designated on the ballot and in election communications as its recommended and supported candidate or list of candidates’. Candidate selection is, therefore, not the same as legislative recruitment; the latter is more comprehensive and actually includes the former. Legislative recruitment involves such aspects of the political system as the legal, electoral and party frameworks (Norris, 1996, 1997). Candidate selection may, indeed, be described as a ‘key stage’ in the recruitment process (Gallagher, 1988a: 2), or even as ‘the most important stage’ (Czudnowski, 1975: 219).

Candidate selection takes place almost entirely within particular parties. There are very few countries—e.g., Germany, Finland and Norway—where the legal system specifies criteria for candidate selection. Only in the United States does the legal system extensively regulate the process of candidate selection. In most countries, the parties themselves are allowed to determine the rules of the game for their selection of candidates. Candidate selection should, therefore, be seen as a particular and important aspect of legislative recruitment that takes place inside the party arena and is predominantly extralegal.

The following sections in this chapter elaborate why it is important for students of party politics to understand the mechanisms and dynamics of candidate selection; what the main factors are that delineate candidate selection methods; and how different candidate selection methods have significant consequences for central aspects of democracy, such as participation, representation, competition, and responsiveness.

Why Study Candidate Selection?

Until recently, candidate selection received relatively little attention. This dearth of scholarly literature has raised a formidable obstacle in the path of researchers who wish to undertake cross-national analyses of the subject. A few pioneering ventures, however, did take place, with initial attempts to produce a theory or a framework for analysis, but they remain few and far between (Duverger, 1959; Czudnowski, 1975; Epstein, 1980; Ranney, 1981; Gallagher and Marsh, 1988; Hazan and Pennings, 2001; Narud et al., 2002). This is partially due to the objective difficulties and obstacles one encounters in any attempt to conduct research on candidate selection—namely, the lack and inaccessibility of empirical data. It is not by chance that Gallagher and Marsh’s (1988) work on candidate selection calls it the ‘secret garden’ of politics. However, the more recent research into this subfield, particularly in the last decade, eschews many of the earlier assumptions, penetrates new grounds of empirical research, and shows that candidate selection has wide-ranging and significant implications for political parties, party members, leaders, and democratic governance.

Beyond being a significant stage in the recruitment process (Norris, this volume), candidate selection is also an important arena for internal party power struggles. Schattschneider’s (1942: 64) argument concerning this issue is worth citing in full:

Unless the party makes authoritative and effective nominations, it cannot stay in business, for dual or multiple party candidacies mean certain defeat. As far as elections are concerned, the united front of the party, the party concentration of numbers, can be brought about only by a binding nomination. The nominating process thus has become the crucial process of the party. The nature of the nominating process determines the nature of the party; he who can make the nominations is the owner of the party. This is therefore one of the best points at which to observe the distribution of power within the party.

Ranney (1981: 103) endorses this statement,

It is therefore not surprising that the most vital and hotly contested factional disputes in any party are the struggles that take place over the choice of its candidates; for what is at stake in such a struggle, as the opposing sides well know, is nothing less than control of the core of what the party stands for and does.

Gallagher (1988a: 3) takes it a step further, stating that ‘the contest over candidate selection is generally even more intense than the struggle over the party manifesto’. Indeed, after an election, what largely remains as the functioning core of almost any party is its office-holders -its successful candidates.

Thus, the importance of candidate selection methods for understanding party politics can be explained by a combination of the three elements elaborated above: First, candidate selection reflects and defines the character of a party and its internal power struggle. Second, it is relatively easy for parties to alter their candidate selection methods. Third, a change in candidate selection methods will affect party politics.

An Analytical Framework for Studying Candidate Selection

In any analysis of candidate selection methods, the unit of analysis is a single party in a particular country at a specific time. Only in those cases where several parties in a particular country use similar methods (usually due to legal requirements), or where a single party uses a similar candidate selection method over time, can one begin to make generalizations about the candidate selection process.

The procedure for classifying candidate selection methods elaborated here is based on four criteria: the selectorate; candidacy; decentralization; and voting versus appointment (Rahat and Hazan, 2001).

Selectorate

The selectorate is the body that selects the candidates. It is, as Best and Cotta (2000: 11) argue, ‘an important intermediary actor in the process of recruitment’. The selectorate can be composed of one person or many people—up to the entire electorate of a given nation. On the inclusiveness to exclusiveness continuum, at one extreme the selectorate is the most inclusive, i.e., the entire electorate that has the right to vote in the general elections. At the other extreme, the selectorate—or rather the selector—is the most exclusive, i.e., a nominating entity of one leader.

Between these two extremes, the selectorate of each single party is classified according to its degree of inclusiveness. For example, American non-partisan primaries, in which every registered voter can vote for candidates from any party, are located near the inclusive end of the continuum (Ranney, 1981). American closed primaries, on the other hand, which demand voters’ registration according to their party affiliation before the day of the primaries, are located slightly away from the inclusive end. The exact location of American primaries will, therefore, depend on the restrictions that are defined by the different state laws (Kolodny and Katz, 1992).

The inclusive end of the continuum also includes examples from Iceland and Spain. According to Kristjánsson (1998), from 1971 on, several parties in Iceland adopted open primaries, where every citizen in a particular electoral district could participate. The Spanish Catalan Party opened its candidate selection to ‘registered “sympathizers”’—non-members who can register as party supporters without paying any membership fee (Hopkin, 2001).

Party primaries, in which the selectors are party members, would be located closer to the middle of the selectorate continuum (Gallagher, 1988a). More and more Western democracies allot their party members a significant role in candidate selection (Bille, 2001). The purest type of party primary is where the party members’ vote alone decides the composition and rank of the candidates. Less pure types allow the party members to select the party candidates from a short-list determined either by party agencies or by a nominating committee, and/or allow the party headquarters to veto certain candidates.

When the selectorate is an agency of the party, we find ourselves in the middle of the continuum. Inside the party, the relative size of each agency is a sign of its inclusiveness: conventions are usually larger than central committees, which in turn are usually larger than executive bodies, such as bureaus. As the size of the particular party agency gets smaller, we move further toward the exclusive pole of the continuum.

An extremely inclusive selectorate is, for example, a special nomination committee that is composed of a few leaders or their aficionados, and whose composition is ratified en bloc. The extreme end of the exclusive pole is defined by a selectorate comprised of a single individual. Israel’s ultra-orthodox religious parties serve as an example of such an extremely exclusive selectorate. In one party, a single rabbi was authorized to decide the composition and order of the party list (Rahat and Sher-Hadar, 1999).

Candidacy

Candidacy addresses the question of who can present himself or herself as the candidate of a particular party. Again we can posit an inclusiveness to exclusiveness continuum. At one end, the inclusive pole, every voter is eligible to stand as a party’s candidate. Some US states are close to this pole. At the exclusive pole, we encounter a series of restrictive conditions. Consider Obler’s (1974: 180) account of the requirements that applied to potential candidates in the Belgian Socialist Party.

While the exact requirements vary from one constituency to another, they generally stipulate that to be placed on the primary ballot aspirants must (1) have been a member of the Socialist party, trade union, co-operative and insurance association for at least five years prior to the primary; (2) have made annual minimum purchases from the Socialist co-op; (3) have been a regular subscriber to the party’s newspaper; (4) have sent his children to state rather than Catholic schools; and (5) have his wife and children enrolled in the appropriate women’s and youth organizations. These conditions, in effect, require that a candidate serve as a member of an activist subculture before he becomes eligible to run for Parliament. They involve a form of enforced socialization during which it is assumed (or hoped) that the aspirant will absorb the appropriate values and attitudes as well as a keen commitment to the party.

More common requirements are less demanding, such as a minimal length of membership prior to the presentation of candidacy and pledges of loyalty to the party. At times, parties will ignore their own candidacy regulations, largely due to electoral considerations. For example, even the exclusive Italian Communist Party included non-members as candidates (Wertman, 1988).

Decentralization

A common error in studies that address candidate selection methods is that of considering decentralization and inclusiveness (and centralization and exclusiveness) as conceptually similar, or at least as describing the same dimension of candidate selection methods (Bowler et al., 1999; Shugart, 2001). Analytically, though, they are different. Decentralization could mean only that control over candidate selection has passed from the national oligarchy to a local oligarchy. For example, if the selectorate is decentralized from a national party conference of several thousand participants to ten local committees each consisting of a few dozen activists and leaders, the overall selectorate has been decentralized, but has not become more inclusive -and has actually become more exclusive.

Party selection methods may be seen as decentralized in two senses, which are parallel to the concepts Lijphart (1984) proposed when he dealt with the division of power in federal and unitary democratic regimes. Decentralization can be territorial, i.e., when local party selectorates nominate party candidates—such as a local leader, a party branch committee, or all party members or voters in an electoral district. Decentralization of the selection method can also be functional, i.e., ensuring representation for representatives of such groups as trade unions, women, or minorities.

Decentralization based on territorial mechanisms, in order to ensure regional and local representation, is rather straightforward. In many European cases, the selectorate at the district level plays the crucial role in candidate selection. The Norwegian case falls close to the territorial decentralization pole. National party agencies cannot veto a candidacy that is determined at the district level, and territorial representation is taken into account inside each district (Valen, 1988; Valen et al., 2002).

More complex mechanisms are required for ensuring functional representation via decentralization. There are two mechanisms commonly used. The first is the reserved place mechanism, which guarantees a minimal number of positions on the list (or minimal number of safe seats in the case of single-member districts) for candidates belonging to a distinct sector or social group. This mechanism implies the decentralization of candidacy alone. Candidates who are eligible for reserved places compete for their place on the list against all of the candidates and are selected by the same selectorate, and the reserved representation mechanism is implemented only if the candidates do not attain the reserved position or a higher one. Establishing quotas for women, a practice adopted by many parties, is one example.

The second mechanism used to ensure functional representation is the sectarian or social group district, where the candidates and the selectors are members of the same sector or social group. This mechanism decentralizes both candidacy and the selectorate. Belgium supplies us with examples of both of the functional representation mechanisms, which were used at the district level. In the Belgian Christian Social Party in 1961, the reserved place mechanism was used when it was decided that in some of the Brussels districts, Flemish and Francophone candidates would get every other seat on the party list. In 1965, separate intra-party subdistricts were actually established when Francophone and Flemish party members in these districts selected, separately, Francophone and Flemish candidates for parliament (Obler, 1974).

Voting versus Appointment

It is usually the case that in smaller and more exclusive selectorates, candidates are appointed, while larger selectorates usually vote in order to choose their candidates. However, a voting system can, theoretically, be used in a selectorate of two or more people, and appointments can take place in bodies that include several dozens of people. When the selection process includes a procedure by which votes determine whether someone is named as the party’s candidate in an election, and/or his or her position on the list, we are dealing with a voting procedure. It should be noted that while a voting procedure can be used by an appointment body of two people or more, it is not considered a voting system unless two conditions are filled: first, each candidacy must be determined exclusively by votes, and not, for example, by an agreed-upon list or an allocation that is ratified by a unanimous or majority vote; and second, the voting results must be presented officially to justify and legitimize the candidacy. When candidacy is determined without fulfilling these conditions, we refer to this as an appointment system. In a pure appointment system, candidates are appointed with no need for approval by any party agency except the nominating organ itself. In a pure voting system, all candidates are selected through a voting procedure, and no other selectorate can change the composition of the list.

Cases located between these extremes are called appointment-voting systems. Such is the en bloc ratification vote that was used in Belgium. In many constituencies, party members were asked either to vote for a ‘model list’—a list of candidates determined by a local party agency—or to express their preferences regarding the candidates. Only if more than 50% of party members did not ratify the model list were the other votes counted, and thus they did not carry much weight (Obler, 1974; de Winter, 1988). In Norway, appointments were more open to change in the ratification process. Lists that were recommended by a nominating committee were then ratified by a majority of a selected party agency, position by position (Valen, 1988).

Voting systems can be further distinguished on the basis of two elements. One is the position allocation formula, i.e., proportional representation (PR), semi-PR, semi-majoritarian, and majoritarian systems. The distinction among these four kinds of voting system is based on their potential level of proportionality. Proportional voting systems in this context will usually be personalized. For example, three of the four largest Irish parties in the 1980s—Fianna Fáil was the exception—used the single transferable vote (STV) system that was also used in general elections to determine the composition of their candidate lists. Semi-proportional systems are those in which the number of votes each selector has is smaller than the number of safe seats being contested. This is the intra-party version of a limited-vote electoral system. In a majoritarian system, the number of votes and safe seats/positions is equal. In many cases, every position is contested separately, making the system almost parallel to single-member district elections. Semi-majoritarian systems are defined as systems where the number of votes that each selector receives is higher than the number of safe seats contested. While such a system is majoritarian—as a majority block can be organized and can take over all of the safe positions—it is ‘semi’ in the sense that incentives for organizing a plurality or majority bloc vote are weaker than in the pure majoritarian case.

The second parameter distinguishes between single-round and multi-round selection methods. In the former, all safe positions are selected at one and the same time, whereas in the latter, the safe positions are filled gradually. The importance of this distinction lies in the opportunities to control and/or balance the composition of the lists that a gradual selection process gives the parties.

There is a connection between the voting system used in the final stage of the candidate selection process and the national electoral system. Where national elections are conducted in single-member districts, the voting system used in the candidate selection process must be majoritarian, in order to produce a single candidate. For example, the exhaustive ballot was used by the British Conservative and Labour parties in the final stage of their selection process, while the Liberals used the majoritarian method of the alternative vote (Denver, 1988). On the other hand, when general elections take place in multi-member districts, the voting system does not need to be majoritarian. For example, in Ireland the exhaustive ballot was used by Fianna Fáil in order to determine its candidate list, position by position, while the next three largest parties used a one-round STV method.

There is also a connection between the selectorate size and the use of either single-round or gradual selection. In smaller selectorates, it is possible to adopt either method. However, when the selectorate is larger—especially in those cases where it includes all party members or the entire electorate—logistics makes the use of a single round almost a must.

The Political Consequences of Candidate Selection Methods

As elaborated above, the classification of candidate selection methods can be based on four major dimensions: selectorate; candidacy; decentralization; and voting versus appointment. The different methods used along these dimensions produce different political consequences. For example, when it comes to the decentralization dimension, territorial decentralization of the candidate selection method could lead to increased responsiveness of members of parliament to the demands and grievances of their (newly created) constituency (Hazan, 1999)—similar to an electoral system founded on small districts. On the voting dimension, parties that use more proportional voting methods, or parties that use appointment methods, could control intra-party conflicts by balancing representation better than parties that use centralized methods and majoritarian voting systems. Concerning candidacy rules, parties could influence the composition of the parliamentary party group by adopting term limits or setting additional criteria for incumbents. Conversely, parties could make the life of incumbents easier by adopting an almost automatic reselection procedure.

The selectorate, however, determines the most significant and far-reaching consequences (Scarrow, 2000; Best and Cotta, 2000). Therefore, this section focuses on the political consequences of the inclusiveness of the selectorate. The impact of this foremost dimension will be assessed according to four important aspects of democracy: participation, representation, competition, and responsiveness.

Participation

When we talk about political participation, we must distinguish between the quantity of participants and the quality of their participation. It is obvious that in terms of quantity, the more inclusive selectorates are the more participatory ones. Indeed, political participation is at its lowest when a few party leaders, or their aficionados, produce the list of candidates. A selectorate that is composed of hundreds, or even thousands, of delegates (a party agency, such as a central committee or a convention) comes closer to the democratic ideal, especially since the party members usually select these delegates. A selectorate that includes tens, and even hundreds, of thousands of party members can be seen, from the quantitative perspective, as an open arena for political participation.

When looking beyond the numbers, in an attempt to analyze the quality of membership and its meaning, the picture becomes quite obscure. Citizens positively perceive the attempts of parties to enhance participation by adopting a more inclusive selectorate in their candidate selection processes. However, most of them do not even bother to join the parties in order to enjoy the now empowered benefits of membership. Numerous country studies, such as Britain (Webb, 2002), Germany (Scarrow, 2002), France (Knapp, 2002), the Scandinavian countries (Sundberg, 2002) and Ireland (Murphy and Farrell, 2002), show that parties try to meet the challenge of declining party membership by empowering party members, yet these efforts have failed to enlarge the number of party members significantly.

Furthermore, many of those who join the parties do not satisfy even the minimum expectations of party members—to be loyal party voters and to be affiliated with the party for more than a short period. Research on Israeli party primaries found that many of the new party members joined the party (or rather were recruited) with the sole purpose of supporting a certain candidate. Approximately one-third of those who joined the party were not even aware of the fact that they were party members. One-tenth were actually members of more than one party. Moreover, many party members voted for other parties in the general elections (Rahat and Hazan, forthcoming). Studies of candidate selection in Canada show that most of those who joined the parties in order to participate in the candidate selection process were ‘instant members’. These members joined the party solely in order to select a leader, or a candidate for parliament, and left the party as soon as the selection was over (Carty and Blake, 1999; Carty, 2002; Malloy 2003). It appears that the parties, for their part, prefer to ‘improve membership statistics’ (Scarrow, 1994: 46) as a demonstration of their public ‘credibility’, rather than to embark on a search for measures to improve the quality of membership alongside its expansion.

The rate of turnout in party primaries is another interesting indicator of the (lack of) quality of a more inclusive selectorate. Kuitunen (2002: 74-5, 77) reports that turnout in primaries in Finland ranged from 20% to 63%. De Winter (1988: 26) reports a decline in the participation by Belgian party members from 51% in 1958 to 25% in 1985. According to Rahat and Hazan (forthcoming), turnout in primaries in Israel during the 1990s ranged from 51% to 75%, while Gallagher (1988b: 246) reports that in the Austrian People’s Party (ÖVP) in 1975 it was 63%. These statistics indicate two things: first, turnout in the party primaries is lower than in the general elections; and second, on average, about one-half of the dues-paying, empowered party members do not bother to collect the ‘merchandise’ that they paid for, and do not participate in their party’s candidate selection process.

A more inclusive policy is also likely to erode, in the long run, the loyal core of party activists. Enhanced, and equivalent, political participation in candidate selection damages the differential structure of rewards (or ‘selective incentives’) in parties when the privileges of long-time loyal activists are made equal to those of new, temporary and unfaithful registrants. This may be the actual aim of those who employ this policy (Katz, 2001), or may be an unintended consequence of it (Hazan, 2002).

Representation

Among the many notions of representation, two distinct concepts can be used to illustrate the influence of candidate selection. The first is the representation of ideas, as described by Pitkin (1976), which implies that the representatives reflect the political beliefs of their voters. The second is representation as presence, which relates to the descriptive characteristics of the representatives—whether their identity is similar to those whom they represent or not (Phillips, 1995). Both kinds of representation are relevant to candidate selection because parties—in their attempt to address the electorate, and to control intra-party conflicts—are likely to try to balance their list of candidates in terms of both notions of representation.

Smaller, exclusive selectorates will be more capable of balancing representation in both senses. When selection can be controlled by a party oligarchy that appoints candidates—and to a lesser extent, when voting takes place in a party agency and can be coordinated—there are more chances that different ideological and social groups (women, minorities, etc.) within the party will be allocated safe positions on the party list, or safe constituency seats. The parties themselves, according to their behavior, appear to validate this claim. The process of democratization of the candidate selection methods in western Europe (Hazan and Pennings, 2001) took place parallel to the increase in the use of representation correction mechanisms (Caul, 1999). Parties increasingly tend to restrict the choices of their more inclusive selectorates in order to ensure representation as presence, particularly that of women (Norris, this volume).

Competition

The candidate selection process pits against each other numerous candidates who aspire to be among those few that will compete in safe seats, or in safe list positions, during an election. There are various ways to measure and compare the level of competition in candidate selection. For example, one can calculate the average number of candidates who compete per safe seat or list position; alternatively, in cases where voting takes place, one can analyze the spread of the votes among the competing candidates. Incumbency, or the level of turnover, could serve as an indicator for the level of competition. For example, higher turnover could signify higher levels of competition.

While incumbency is an advantage in every kind of candidate selection method, as it is in every type of electoral system (Somit et al., 1994), the differences in the inclusiveness of the selectorate are likely to create variations in the extent of this advantage. Smaller selectorates allow aspiring candidates a chance to be known and to personally contact their selectors. When the selectorate is inclusive, i.e., composed of party members at large, support cannot be based on personal affiliations and incumbency is thus likely to offer a larger advantage. This is mainly because, as public officials, incumbents enjoy publicity and the ability to demonstrate responsiveness to the demands of the selectorate, interest groups, financial supporters, etc. The American experience supplies clear evidence of the advantage of incumbents in primaries (a very inclusive selectorate). Between 1978 and 1992, in only 47 cases out of 3166 (1.5%) were incumbents seeking re-election to the House of Representatives defeated in the primaries, and in only 13 cases out of 236 (5.5%) were incumbents seeking re-election to the Senate defeated in the primaries (Jackson, 1994). As Maisel and Stone (2001: 43) point out, ‘It is clear that primary elections do not serve to stimulate more competition, and to the extent that competition is an essential ingredient of democracy, it is not clear they accomplish their intended purpose of enhancing U.S. democracy’.

If we accept the argument that party agencies are more competitive than primaries because of the shorter ‘distance’ between the candidates and selectors, then nomination committees (a very exclusive selectorate) can be expected to be even more competitive. This is because it is easier for each candidate to be in personal contact with each member of the committee than would be the case in a more inclusive selectorate. However, this prediction misses an important part of the picture. The nomination committee suffers—because of its small size and informal, non-transparent working procedures—from a lack of popular democratic legitimacy, i.e., a democratic deficit. The best strategy for the nomination committee to legitimize its decisions in the eyes of the party members, the party agencies and even the general public is to present a list that is largely composed of incumbents—i.e., a list that reflects the existing balance of power and will thus not encounter much antagonism. There will be changes, to demonstrate that something was changed and that the nomination committee is not a rubber stamp, but they will be minimal. The result is that party agencies, located in the middle of the inclusive-exclusive selectorate dimension, are likely to be the most competitive selectorates. The more exclusive selectorates, such as nominating committees, are likely to be the least competitive, while the most inclusive selectorates, such as primaries, are likely to be moderately competitive (Rahat and Hazan, 2005). More data and more empirical analysis are needed in order to validate and strengthen this assessment, because to date there are few studies of competition in the candidate selection process outside the inclusive American arena.

Responsiveness

Since a central motivation, and constraint, for the behavior of members of parliament (the successful candidates) is their wish to be reselected, they will pay special attention to the grievances and demands of their selectorate. The composition of the selectorate is thus likely to influence the behavioral patterns of parliamentarians and of their parent organization, the political party. Bowler (2000), for example, argues that the best explanation for the collective action of legislators is the nomination procedures in general, and who nominates in particular.

One approach claims that there are negative relationships between inclusiveness and party cohesion because the role of non-party actors in candidate selection rises with an increase in the inclusiveness of the selectorate, and also their importance as an object of responsiveness. Legislators who were selected by small nominating committees owe their positions to the party leadership, and are therefore likely to be first and foremost party players. Legislators who were selected by party agencies are likely to be party players at most times, but nevertheless will be somewhat differentiated in their efforts to promote the demands and interests of the groups within the party that serve as their power base. Legislators who were selected in primaries need the help of non-party actors in order to reach their massive, fluid, and somewhat apathetic audience. In the more inclusive selectorates, candidates need political ‘mediators’ who can supply the resources needed to address the large number of party members: capital holders who can supply the necessary finances to address a huge selectorate; interest group leaders who command the votes of hundreds, and even thousands, of members; and the mass media. All of these mediators have narrower—and sometimes different—interests and perceptions than the party as a whole. Facing the plurality of pressures that characterize the more inclusive selectorates, the cohesion of the parties is likely to decrease, as parliamentarians behave more like individuals than team players.

Comparing the level of party cohesion in the US Congress to levels of party cohesion in other Western democracies illustrates this point: the relatively low cohesion of US parties can be explained by their adherence to extremely inclusive candidate selection methods. Candidates select their party label with little to no say on the part of the party institutions. In most other cases, where cohesion is much higher, the parties as such have a say in candidate selection. First of all, party institutions often take part in candidate selection: filtering candidacies, validating their selection and sometimes playing the central role in their selection (Bille, 2001). Even when a growing number of parties allow their members a crucial role in candidate selection, membership, as such, is far more exclusive than in the USA (Epstein, 1980). As Gallagher (1988b: 271) argued, ‘It may not matter much, in this sense [level of party cohesion], which party agency selects candidates, but it does matter that some party agency selects them’. Second, Western European parties still put some formal and informal limits on candidacy. Third, US parties have lost most of their control over candidate selection. The most they can do—and even in this they are limited—is to endorse a certain candidate competing in the party primaries for the use of the party label (Jewell and Morehouse, 2001). Another case in point is that of the democratization of candidate selection methods in Israel, Iceland, and Taiwan. In all three countries, the adoption of a more inclusive selectorate led to a decrease in party cohesion (Rahat and Hazan, 2001; Kristjánsson, 1998; Baum and Robinson, 1999).

A second approach, based on the logic of the cartel party model (Mair, 1994, 1997; Katz and Mair, 1995; Katz, 2001) and on the Canadian experience (Carty 2004), claims that inclusive selectorates actually increase the power of party elites and help preserve party cohesion. An increase in nominal power at the base of the party is achieved at the expense of the middle-level activists, as they are the ones who might be able to coordinate an effective challenge to the autonomy of the party leaders. The rationale behind this approach is that the less intense (atomistic, unorganized, unstable) audience of party members is more likely to take cues from the highly visible party leadership. Empirical support for this approach comes from the Canadian case of inclusive selectorates and high party cohesion, and from other cases in which the opening up of candidate selection did not lead to a decrease in cohesion. However, as in the case of competition, the study of the impact of candidate selection methods on responsiveness, and in particular on party cohesion, is still in an embryonic phase. More data and empirical analysis, such as Hix (2004), are needed to assess its ramifications comprehensively.

Intra-Party Democracy

The political consequences of the selectorate dimension in candidate selection reveal that more democracy in one dimension does not necessarily lead to more democracy in other dimensions. For example, parties that use small nomination committees can ensure representation, but are problematic in the participatory sense. On the other hand, parties that select their candidates through primaries enjoy high levels of participation, but can hardly balance representation. This indicates that in order to be considered truly ‘more’ democratic, parties should look for a candidate selection method that could optimally balance different democratic dimensions, such as participation, representation, competition, and responsiveness. It is the challenge of future research to clarify the relationship between these dimensions, based on the much needed empirical research, so that the study of candidate selection can supply a solid base for improved theoretical and political understanding of democracy in general and democracy within parties in particular.

Challenges for Future Research

The major challenge for the study of candidate selection methods is to bring it closer to the state that the study of electoral system was in approximately 40 years ago, when Rae’s (1967) seminal work The Political Consequences of Electoral Laws was published. That is, we need cross-party and cross-national empirical studies of the political consequences of candidate selection methods.

Achieving this is by no means easy. Existing theoretical frameworks, particularly those concerning party politics, provide substantial propositions for the study of candidate selection. The lack of cross-national empirical studies is, nonetheless, the Achilles heel of any attempt to make further progress. The problem with a cross-national empirical study is that it requires familiarity with local politics and accessibility (in terms of language, as well as in other more basic terms) to intra-party data.

Candidate selection methods not only affect party politics, they can also reflect party politics. There is, therefore, a need to analyze candidate selection methods both as a dependent and as an independent variable. Prominent examples of treating candidate selection as a dependent variable, in the rather exceptional case of the United States, range from Key (1949) to Ware (2002). The possible links between candidate selection and such variables as the structure of government, the electoral system, the political culture, and the nature of the party were already suggested long ago (Gallagher and Marsh, 1988), but were only recently put to a cross-national empirical test (Lundell, 2004). There is a dire need for additional data, before any conclusive findings can be reached.

An additional path is the study of the politics of reform of candidate selection methods -why, when, and in what circumstances parties change (or preserve) their candidate selection methods. Several studies have suggested various explanations for such changes, for example, the democratization of candidate selection methods (Katz, 2001; Scarrow, 1999; Hazan and Pennings, 2001). Yet there remains a need for a more integrative look at the phenomenon, one that would account not only for overall trends but also for the differences among parties and among nations that may result from interactions at the inter- and intra-party levels.

Sailing the uncharted waters of candidate selection could help us better understand the nature of party membership, the kind of candidates selected, the dynamics exhibited within the party, the power and performance of the party in parliament, and our overall ability to evaluate party politics. Behind closed party doors, this ‘secret garden’ of politics is still largely unexplored.