Gatekeeping by Central and Local Party Actors: Theory and Evidence from a Field Study of New Brunswick Nominations, 2017-2018

Quinn M. Albaugh∗ January 27, 2021

Abstract Under what circumstances do central and local party actors engage in gatekeeping to inﬂuence the outcomes of local nomination races? In this paper, I develop a theory of gatekeeping in Canadian parties. I present evidence for this theory with a ﬁeld study of New Brunswick provincial nominations (2017-2018), including participant-observation of 25 nominating conventions; 93 interviews with party insiders, nomination candidates, and other individuals with knowledge of the nominations process; and an original dataset of nominations for the 2018 New Brunswick election. The theory and results call into question the conventional wisdom that riding associations are the decisive actors in selecting local candidates in Canadian political parties. The extent of central party gatekeeping suggests a need to re-evaluate Carty’s claims about a “stratarchical bargain” within Canadian political parties and, ultimately, challenges the notion that Canadian party nominations are minimally democratic.

∗University of Toronto and Max Planck Institute for Religious and Ethnic Diversity. Please direct correspon- dence to [email protected]. I gratefully acknowledge support from a Social Sciences and Humanities Research Council of Canada Doctoral Award, along with grants from the Princeton Canadian Studies and the Mamdouha S. Bobst Centre for Peace and Justice. I would like to thank Chris Achen, Jeyhun Alizade, Elizabeth Baisley, Paul Frymer, Dara Strolovitch, and seminar participants at Princeton University for comments on previous versions of this research.

1 Since the first large-scale studies of candidate selection in Canada in the early 1990s, the guid- ing assumption of scholarship on candidate selection in Canadian political parties is that riding associations select candidates (Carty, 1991; Carty and Erickson, 1991; Erickson and Carty, 1991). This assumption underpins arguments that diversifying riding association executives should im- prove the representation of women and racialized minority groups (Cheng and Tavits, 2011; Cross and Pruysers, 2019; Medeiros, Forest, and Erl, 2019; Tolley, 2019; Tremblay and Pelletier, 2001). At the same time, scholars have documented several formal and informal means through which central party organizations – party leaders, their staff, and formal party organizations – increased power over candidate selection, including refusing to sign candidates’ nomination papers (Court- ney, 1978, 2015; Erickson and Carty, 1991), formal appointment powers for party leaders (Koop and Bittner, 2011), vetting or “green-light” processes that allow party committees to keep candidates off the ballot (Pruysers et al., 2017), and manipulating the rules or logistics of nominating conventions (Pruysers et al., 2017). However, we still know relatively little about why central and local party actors engage in gatekeeping – exerting influence over the choice of candidate during candidate selection to favour or disfavour particular candidates – over candidate selection in particular nomination races in particular ridings. Gatekeeping is critical to understanding which actors have influence over the nominations process, why Canadian parties nominate the candidates they do, and, ultimately, which individuals actually hold elected office. In this article, I argue that central party organizations exert much more influence over local candidate nominations than past work has acknowledged. They typically do so through informal gatekeeping practices, such as controlling the logistics of nominating conventions, rather than through formal powers of appointments and the leader’s veto. However, they are less effective in exerting control over nominations when riding associations are strong, particularly in relatively safe seats with retiring incumbents. I support this argument with a multi-method field study of candidate nominations in New Brunswick’s political parties across all ridings in advance of the 2018 provincial election. A field study allows for the gathering of in-depth qualitative evidence on particular nomination races. This type of qualitative evidence is rare in scholarship on candidate selection, since it is costly to obtain. However, it is invaluable for identifying mechanisms and ensuring that particular cases are correctly coded (George and Bennett, 2005). Although a study of federal nominations would speak most directly to past work on candidate selection in Canada, it is not generally feasible to gather in-depth data across a large number of ridings given their geographic size. For this reason, field studies of riding-level politics in Canada, such as Scarrow (1964), Sayers (1999), Koop (2011), Koop, Bastedo, and Blidook (2018), tend to rely on a small number of ridings. New Brunswick is small enough to allow for the collection of qualitative and quantitative data across all ridings. New Brunswick serves as a microcosm of party politics in Canada. New Brunswick is the only province whose party system has traditionally shared similar linguistic and religious pressures as the federal party system. Its provincial and federal parties tend to overlap substantially in terms of personnel (Koop, 2011). As a result, while these findings draw on provincial nominations, they

2 do so in the province most likely to generalize to federal politics and most able to speak to past work on candidate selection at the federal level. I review the conventional wisdom that local party organizations control candidate selection in Canada, along with increasing evidence that central party organizations have more power than previously ackonwledged. I outline a theory of gatekeekping by central and local party actors that explains many past findings on candidate selection. I sketch out a series of observable implications of this theory, for which guide the evidence presented in the paper. I lay out the research design, materials, and data collected. I present a series of qualitative and quantitative evidence that supports the theory. First, I present interview and participant-observation evidence that there is a “norm of interference” that limits the power. Second, I recount interview and participant- observation evidence that party actors nonetheless have “preferred candidates” that they support through the nominations process despite the norm of non-interference. Third, I demonstrate with interview evidnece that party actors have a range of different activities, which form a “gatekeeping repertoire,” through which they can influence the nominations process. Fourth, I use quantitative data on uncontested and contested nominations to show that, in line with the theory, contested nominations overwhelmingly emerge in competitive and safe non-incumbent ridings, particularly those in which central and local party actors have recruited different nomination candidates. Finally, I show that gatekeeping helps explain when disagreement between central and local actors over the choice of candidate does not lead to a contested nomination using a close study of “near-contested” nominations – cases in which multiple candidates expressed an interest in running at the same time but the party acclaimed one candidate at the nominating convention. This paper makes three main contributions. First, it brings new evidence to bear on debates over the influence of central and local party actors, including elite interviews, participant- observation and quantitative data across all ridings. This combination of materials suggest that central party actors, including party leaders’ offices and formal party organizations, are often able to influence the selection of local candidates behind the scenes to shape the overall candidate slate. This observation stands in stark contrast with the implicit assumption in past work that riding-level variables alone explain the nomination of women and members of racialized minority groups (Cheng and Tavits, 2011; Cross and Pruysers, 2019; Medeiros, Forest, and Erl, 2019; Tolley, 2019; Tremblay and Pelletier, 2001). Second, it suggests a need to re-evaluate the extent to Carty’s “franchise bargain” (Carty, 2002, 2004) and the concept of intra-party democracy (Cross and Katz, 2013; Cross et al., 2016) actually hold within contemporary Canadian parties. Although party actors often recognize the power of norms that bar party actors from “interfering” in candidate nominations and that support the right of local party members to choose candidates, party actors often circumvent these norms to exert influence over the candidate nominations process. As a result, the evidence for gatekeeping casts doubt on the ideas that riding associations autonomously choose candidates or that Canadian political parties are fully internally democratic. Third, the theory has testable implications for future work in federal and provincial politics, along with work on candidate selection in comparable countries such as the United Kingdom.

3 Who Controls Nominations in Canada?

On paper, Canadian political parties have traditionally used the most open candidate selection processes outside of American primary elections (Cross et al., 2016; Hazan and Rahat, 2010). In most federal and provincial political parties, the formal rules governing who is eligible to vote are even more permissive than general elections, save for membership dues. Individuals ages 14-17 and permanent residents are able to join political parties and vote at local meetings for that party’s candidate (Cross, 2004). These formal rules ostensibly give nomination voters – party members and supporters – the power to choose candidates for general elections. Despite these permissive formal rules, party members and supporters have relatively limited sway over the choice of nomination candidates in practice. One of the main reasons why party members join parties in the first place is to vote in nomination and leadership races (Young and Cross, 2002)). Few individuals are ongoing members of political parties, particularly Liberal and Conservative parties (Carty, 1991, pp. 36-39). The accepted strategy for winning a nomination race is to turn out family, friends, acquaintances, and their networks (Carty and Eagles, 2005; Sayers, 1999).1 These individuals have little reason to cast voters for candidates other than the ones who recruited them in the first place. As a result, nomination candidates generally choose their voters, rather than the other way around. More importantly, as recent work on Canadian (Pruysers et al., 2017), American (Cohen et al., 2009; Hassell, 2016, 2017; Masket, 2009), and comparative politics (Bjarneg˚ard,2013; Bjarneg˚ardand Zetterberg, 2016) has shown, different types of party actors can exert informal influences over the nomination process even when an expansive selectorate formally chooses the candidates. Since the 1990s, the conventional wisdom on candidate selection in Canada has been that riding association executives are the key actors in selecting local candidates because of these informal processes. In 1988, Carty and Erickson found that the majority of nominations went uncontested (Carty and Erickson, 1991, 120). This was true even in ridings where riding association presidents perceived the seat as “safe” for their political party (Carty and Erickson, 1991, 137). In a 1991 survey of riding association presidents, Carty found that “[f]orty percent of the constituency association presidents confirmed that in their riding ‘a core group of the association pretty well decide among themselves on who the candidate ought to be, then work to get that individual nominated” (Carty, 1991, p. 109). These findings are fairly clear signs of gatekeeping by local party actors. Thomas and Morden (2019) found similar of non-competitive nominations and gatekeeping activities in an recent report on federal nominations in Canada. Later in his career, Carty described Canadian parties as having a “franchise bargain” or “stratarchical bargain” in which leaders control policy and riding associations control candidate selection.(Carty, 2002, 2004). In more recent years, this assumption that riding association executives are the key actors for candidate selection has driven work on the importance of riding

1The main exception to these personalistic recruitment eﬀorts is organized membership drives by social conservative organizations. For a discussion, see Baisley (2020).

4 association executives for the recruitment and selection of women and racialized minority groups (Cheng and Tavits, 2011; Cross, 2018; Cross and Pruysers, 2019; Medeiros, Forest, and Erl, 2019; Tolley, 2019; Tremblay and Pelletier, 2001). These studies have consistently found that local actors matter for the nomination of women and racialized minority groups. Since they do not include information on the activities of other actors within party organizations, they implicitly assume that only riding-level activities matter. Contrary the prevailing assumption that riding associations are the central actors, there is mounting evidence that other actors within party organizations play an important role in nomination races. These roles are both formal and informal. Since the adoption of the Canada Elections Act in 1970, party leaders must sign candidates’ nomination papers (Courtney, 1978, 2015; Erickson and Carty, 1991). This law effectively gave party leaders a veto over local candidate nominations. Since 1970, most provinces have adopted some version of this requirement, as well (Albaugh, 2020, p. 45). This practice emerged out of Chrétien’sefforts to block social conservatives from winning local nomination races (Koene, 2008; Baisley, 2020). However, it quickly became a method for appointing preferred candidates, particularly star candidates and women candidates (Koop and Bittner, 2011). Finally, since the 1990s, formal party organizations have increasingly developed the ability to keep prospective nomination candidates off the ballot if they do not submit a detailed application and pass a review of their social media profiles and public communications (Albaugh, 2020; Marland and DeCillia, 2020; Pruysers et al., 2017). Even beyond these formal institutional developments, central party organizations play a major role in informal gatekeeping activities, including running their own candidate recruitment efforts (Cross and Young, 2013) and manipulating the logistics of nominating conventions to help their preferred candidates win (Pruysers et al., 2017). In order to settle debates over the relative power of different actors over candidate selection, it is important to pay attention to informal practices of influencing the nominations process – or gatekeeping. Gatekeeping was central to early research on nominations, particularly in American and British politics (Dallinger, 1903; Key, 1949, 1964; Ostrogorski, 1903). More recently, these informal practices have become central to American parties scholarship on “party control” over nominations (Cohen et al., 2009; Hassell, 2016, 2017; Masket, 2009) and feminist institutionalist scholarship on candidate selection and women’s representation (Bjarneg˚ard,2013; Kenny and Verge, 2012, 2016; Krook, 2010). However, in Canadian politics research, as in American and comparative politics research, scholars often discuss gatekeeping practices in relatively general terms, noting that gatekeeping takes place but not examining when, where, how, and why gatekeeping does and does not take place in particular cases. (For a good treatment of gatekeeping practices in Canadian federal parties, see Pruysers et al. (2017)). Ultimately, these questions require a theory of gatekeeping. In the next section, I present a theory of gatekeeping that draws on insights from rational choice institutionalism and sociological institutionalism (Hall and Taylor, 1996).

5 A Theory of Gatekeeping by Central and Local Party Actors

In this section, I outline a theory of gatekeeping by central and local party actors. I identify the relevant actors, their goals, and their capacity to engage in gatekeeping. This theory brings together many of the existing findings on candidate selection from Canada and elsewhere. It applies to parties that (1) run in district elections (rather than one nation-wide proportional representation system), (2) regularly win seats, and (3) have both a central leadership and local organizations. In any political party, the actors who are relevant “gatekeepers” depend on the structure of the party organization. For example, American parties research has tended to find that networks of national party organizations, donors, and elected officials tend to provide resources to primary election campaigns (Cohen et al., 2009; Hassell, 2016, 2017; Masket, 2009). In Canada, the gatekeepers include not just riding association executives – the central actors in past work on candidate selection in Canada – but also party leaders and their staff, members of the formal party organization including party executive directors, members of caucus, other nominated candidates, and retired politicians. For the sake of parsimony, I simplify the actors into two networks – central (party leaders’ offices, formal party organizations, and prominent non-local party members) and local (riding associations and prominent local party members). I call them networks to make the link explicit between this study and others that conceptualize political parties as networks (Koger, Masket, and Noel, 2009; Schwartz, 1990). The theoretical assumption that there are both central and local party networks who may come into conflict has deep roots in scholarship on Canadian parties (Carty, 1991; Erickson and Carty, 1991), and similar central-local party organization conflicts are also classic topics of study in other countries, such as the United Kingdom (Ranney, 1965; Rush, 1969) or Ireland (Gallagher, 1980). Central and local networks often have conflicting goals over candidate selection. The central network tends to prioritize winning the most seats (or a majority of seats, if possible), ensuring that prospective cabinet ministers, winning positive news coverage, and avoiding negative news coverage, especially about the leader. For these reasons, central networks may be more likely to prioritize the selection of women and racialized minority groups. (This claim fits with observations from work on candidate selection and intra-party democracy (Cross et al., 2016; Hazan and Rahat, 2010) and work on the professionalization or bureaucratization of political parties and women’s representation (Bjarneg˚ardand Zetterberg, 2016; Norris, 1996).) By contrast, local clusters tend to prioritize winning that particular seat and picking good constituency representatives who will pay attention to their ridings. Traditionally, policy considerations have not been central to local nomination races in Canada, particularly in the Liberal and Conservative parties (Carty, 1991; Sayers, 1999). The main exceptions are efforts by social conservatives to get “their own” nominated as candidates (Baisley, 2020; Cross, 2004; Koene, 2008). Electoral competition plays a major role in shaping the dynamics of candidate selection, as demonstrated by a considerable amount of scholarship on Canadian (Carty and Eagles, 2005; Sayers, 1999) and American politics (Hirano and Snyder Jr, 2019). Rational party actors value

6 Figure 1: Theoretical Relationship of Expected Electoral Performance and Electability Concerns

Electability Concerns

High

Low

Low Expected Performance High

7 “electability” – or positive “local candidate effects” (Blais, 2005; Stevens et al., 2019) or the “personal vote” (Cain, Ferejohn, and Fiorina, 1987) – the most when a choice of candidate can actually make a difference at the margin in whether the party wins an additional seat. They do not value electability at all when there is no reason to expect that the local candidate will make any difference in whether the party wins an additional seat, as in “hopeless” or “safe” ridings. Figure 1 displays the hypothesized relationship between the party’s expected performance in the next election and the degree to which local and central party clusters prioritize electability in selecting the local candidate. This figure does not draw on actual “data” but on a theoretical relationship. Of course, expected electoral performance is in large part a function of past electoral performance. As a result, party actors pay close attention to candidates’ past experiences (if any) running for elected office. Party actors tend to have well-defined preferences over different candidates for nominations. Not only do they have well-defined goals but also they frequently have knowledge of potential candidates. They may know potential candidates first-hand, may know someone else within the party or their local community who knows the candidates, and may have impressions of the candidates from their backgrounds. Party actors’ preferences are clearest when they have enough knowledge to believe that potential candidates differ in the extent to which the potential candidates can help the party actors realize their goals. The presence of incumbents plays a major role in shaping the preferences of party actors. Incumbents have uniquely strong relationships within party organizations. Due to their experience and social ties, incumbents are typically are capable of constructing campaign organizations to fend off nomination challenges, although they do not invariably win contested nominating conventions. When a candidate wins a nomination, that candidate may “stack” the executive of the local party organization with their supporters from the nomination campaign. These dedicated supporters will tend to support an incumbent. On top of these relationships, local and central clusters typically have electoral incentives to value incumbents. For one, it is hard to know which candidates are actually going to perform well in elections. Incumbents are proven winners (Erikson, 1971). Unless the incumbents are inattentive to their ridings or face major scandals or otherwise give party actors reasons to think they are a liability, central and local actors tend to agree about the need to keep incumbents in office. In the absence of incumbents, competitive nomination races tend to emerge when different actors within party organizations – particularly the central and local networks – prefer different candidates. In Canadian parties, central and local networks often engage in their own recruitment processes (Cross and Young, 2013). It is rare for individuals to become candidates without recruitment from either central or local networks (Cross and Young, 2013). When central and local actors recruit candidates, they develop preferences over potential candidates. Both central and local actors usually favour candidates they themselves recruited. When the central and local networks agree over their “preferred candidate” (Cohen et al., 2009; Hassell, 2016, 2017; Masket, 2009), they are usually able to orchestrate that individual’s nomination. Any nomination candidates who run against the preferred candidate of the central and local networks is likely to

8 be inexperienced and lose the nomination race. When they disagree, however, they either engage in competing gatekeeping efforts to influence the process behind the scenes or allow a competitive nomination process to resolve the conflict over who will become the candidate. Party actors engage in gatekeeping under a limited set of circumstances. First, no party actor will engage in gatekeeping if their preferred candidate is going to win the nomination unassisted. It is simply unnecessary to use power if the desired outcome will happen anyway. Second, if party actors from both clusters generally agree on a candidate, there is no need for them to exert influence. Third, gatekeeping is costly. In Canadian parties, there is a “norm of non-interference” in candidate nominations in which local associations are supposed to choose their own candidates without interference from outside actors, including party leaders or formal party organizations (Carty, 1991, 2002, 2004; Siegfried, 1906). Some gatekeeping activities violate this norm of non-interference and can provoke a backlash. The risk of backlash is particularly acute for party leaders, whose images are important for elections (Bittner, 2011, 2018). Although the central cluster is in principle always able to engage in gatekeeping over nominations, local clusters vary considerably in terms of their organizational strength. Figure 2 displays the theoretical relationship between organizational. Again, this figure does not represent data. In many ridings, there is no meaningful organization. Local party organizations tend to be stronger where parties perform better in general elections (Carty, 1991; Carty and Eagles, 2005; Sayers, 1999). This theoretical relationship has an S-shaped curve, for a few reasons. First, it takes some ”startup costs” to build a local organization. Second, incumbents play a major role in shaping the strength of local organizations. Apart from unusual cases, like the “Orange Wave,” winning candidates usually build strong local organizations through their nomination and general election campaigns campaigns. Local party organizations become much more active when there is an incumbent, as well (Koop, 2010). As a result, in ridings where the party has previously performed well, particularly ridings where the party elected a member in the previous election, the less power the party leader’s office has over candidate selection.

Observable Implications

This theory generates several observable implications. In this section, I outline several of these observable implications, the role they play in the theory, and the methods I use to verify them. I note, where possible, when past work corroborates the expectations of the theory.

Observable Implication 1: Party actors have preferences over prospective candidates for nominations.

If party actors do not have preferences over prospective candidates for nominations, there is no reason for them to engage in gatekeeping. The claim that party actors have “preferred candidates” has support in American research on the UCLA School of Parties (Cohen et al., 2009; Hassell, 2016, 2017; Masket, 2009). In this study, I not only demonstrate that party insiders have

9 Figure 2: Theoretical Relationship of Electoral Performance and Strength of Local Party Organi- zation

Organizational Strength

High

Low

Low Past Performance High

10 preferred candidates using interviews and participant-observation of nominating conventions but also that these preferences emerge through central and local recruitment processes. I triangulate evidence from participant-observation, conﬁdential interviews with party actors who engaged in gatekeeping, and corroborating interviews with nomination candidates and other actors within the party network to identify party actors’ preferred candidates in contested nomination races.

Observable Implication 2: Party actors recognize the existence of a norm of non- interference in candidate selection.

The norm of non-interference is an essential part of the theory because it explains why central party organizations do not generally use their most powerful gatekeeping activities, such as the “leader’s veto” or rejecting candidates during the vetting process. Canadian parties scholarship has long recognized the importance of local autonomy in candidate selection (Carty, 1991, 2002, 2004; Siegfried, 1906), which corroborates the existence of this norm. I show that party actors recognize the existence of this norm and the risk of backlash from journalists and other party actors using interviews and participant-observation evidence.

Observable Implication 3: Central and local party actors each have and use organizational resources to engage in gatekeeping in at least some ridings.

If party actors do not have organizational resources to engage in gatekeeping, or if they do not use them, then gatekeeping does not occur. I identify the organizational resources available to diﬀerent networks and actors within political parties and examine their frequency of use through a combination of interview and documentary evidence.

Observable Implication 4: Contested nominations should be more likely to occur in competitive and safe non-incumbent ridings than in “hopeless” and incumbent ridings.

By contested nominations, I mean nominations in which two or more candidates appear on the ballot. Under the theory, incumbency and electoral competitiveness play major role in explaining the dynamics of gatekeeping by central and local party networks. In incumbent ridings, central and local actors usually agree on the incumbent as the preferred candidate, since both of them benefit from holding the seat and avoiding conflict. In hopeless ridings, the central party network typically does the recruitment due to the weakness of the local organization. However, in competitive and safe seats, the central and local actors both have strong incentives, and these incentives are not always well-aligned. Although American work suggests that the relationship between perceived electoral competitiveness and competition in primary elections is positive and linear (Hirano and Snyder Jr, 2019), Canadian scholarship suggests that this relationship may not necessarily be linear (Carty, 1991; Carty and Erickson, 1991; Carty, Eagles, and Sayers, 2003; Carty and Eagles, 2005). That is, the difference between “hopeless” ridings and “competitive” may be much larger

11 than the one between “competitive” and “safe” ridings. These diﬀerent relationships may be a product of the greater capacity of Canadian party actors to engage in gatekeeping activities that prevent the emergence of contested nomination races in safe seats. I study the relationship between incumbency and electoral competitiveness using quantitative data on nomination races constructed from electoral results and political ﬁnance reports on nomination races.

Observable Implication 5a: If central and local actors generally agree on a preferred candidate, then that candidate is very likely to win the nomination.

Observable Implication 5b: If competing party networks disagree in their preferred candidates, then either (1) at least one network will engage in gatekeeping to prevent the nomination from becoming contested or (2) competing preferred candidates run against each other in a contested nomination.

In the theory, the degree to which central and local actors agree or disagree over their preferred candidates plays a major role in explaining the use of gatekeeping activities and the prevalence of contested nominations. When both networks agree, then their preferred candidate is very likely to win, since that candidate has the advantage of party resources and support over outsider candidates. This expectation parallels recent claims in American parties research. In the UCLA School of Parties in American politics, party actors “coordinate” on a single preferred candidate (Cohen et al., 2009; Hassell, 2016, 2017; Masket, 2009). The extent of this coordination helps explain which primary election candidates win the nomination. However, I ﬁnd that party actors often disagree over the choice of candidates. When they do, the central and local networks face a choice. They can either engage in gatekeeping or allow their preferred candidate to face a contested nomination. If they engage in gatekeeping, they may face a backlash from other actors within the party or critical news coverage. This risk of backlash is particularly acute for central networks due to the “norm of non-interference.” If they do not, then their preferred candidate may face a contested nomination race – and lose. I identify these patterns of agreement and disagreement using in two stages. I code the preferred candidates of the central and local party networks using evidence from interviews and participant-observation. I demonstrate that conﬂict between central and local party actors helps explain the use of gatekeeping activities to prevent competitive nominations and the emergence of competitive nomination races using interview evidence, participant-observation, and documentary evidence.

Observable Implication 6: When central party actors engage in gatekeeping, they decrease the chances of a contested nomination.

Finally, the theory suggests that gatekeeping activities can reduce the chances of a contested nomination. If gatekeeping is a meaningful concept, then there must be some evidence that gatekeeping can actually prevent certain candidates from running for the nomination. I present quantitative evidence that gatekeeping reduces the probability of a contested nomination through

12 an analysis of gatekeeping in cases of disagreement between central and local actors over their preferred candidates. The quantitative data provide general evidence for the existence of this pattern. After reviewing the quantitative evidence, I examine cases of “near-contested” nominations – that is, nomination races in which at least two candidates expressed a serious interest in running but the party acclaimed one candidate – for qualitative evidence of the eﬀective- ness of gatekeeping. These cases illustrate the mechanisms through which gatekeeping prevents contested nominations.

Data and Methodology

I illustrate this theory using a multi-method field study of New Brunswick politics. For the pur- poses of this study, New Brunswick serves as a feasible case for gathering qualitative information across the universe of ridings, due to its small geographic size. However, Canadian federal and provincial parties generally share similar candidate selection institutions (Cross, 2004), along with similar gatekeeping activities (Pruysers et al., 2017). Party actors also tend to learn lessons from other levels of government and other provinces. As a result, although this study only draws on material from New Brunswick, its findings are qlikely to generalize to other provinces and levels of government. I spent over 20 months conducting fieldwork in New Brunswick. During that time, I conducted 95 interviews with 69 individuals familiar with different aspects of candidate selection in New Brunswick’s political parties. These interviews included (1) current and former nomination candidates or potential nomination candidates, (2) party insiders (including party leaders, executive directors, riding association presidents, staffers involved in vetting candidates, and (3) interviews involved with feminist organizations who sought to increase women’s representation. Since the backroom politics of nominations is often sensitive, I made the decision in advance to conduct the interviews as confidential but quotable without attribution. I allowed respondents to go completely “off the record” as needed to gain additional insight into the nominations process. I conducted all but two interviews in English but offered interviewees a choice between English or French interviews. For additional details on the interview methodology, please see the supporting information. In addition to the interviews, I conducted participant-observation of 25 nominating conventions. These nominating conventions included all but two contested nominations across the five parties that had a chance of winning at least one seat in the Legislative Assembly – the Progressive-Conservative (PC) Party, the Liberal Party, the New Democratic Party (NDP), the Green Party, and the People’s Alliance of New Brunswick (PANB). The exceptions were the contested PANB nomination in Southwest Miramichi-Bay du Vin, which took place at the same time as a contested PC nomination in Fundy–The Isles–Saint John West, and the contested PC nomination in Albert, which took place at the same time as a contested Liberal nomination in Kent North. For full details on the ridings observed, please consult the Supporting Information.

13 Informal chats with party insiders at nominating conventions helped shed light on the informal rules and understandings, along with some insight into party insiders’ preferred candidates. The participant-observation work also helped in recruiting interview participants, since many interviewees were more likely to respond to an emailed or mailed recruitment letter if they had already met me. During the field research, I read newspaper coverage of nomination races, along with all the documents and information on nomination contests available on political parties’ websites. These documents , but they did have valuable information on particular nomination races that became major news stories, such as the legal dispute between prospective PC nomination candidate Chris Duffie and the Progressive-Conservative Party of New Brunswick (PCNB) over the way PCNB called the nominating convention in Carleton–York or Premier Brian Gallant’s decision to expel former Speaker Chris Collins from the Liberal caucus over accusations of sexual misconduct. I supplement the qualitative material with analysis of an original riding-level nominations dataset for the 2018 provincial election. I built the dataset by starting with riding-level electoral results from the 2014 and 2018 elections and political finance records on nomination races for the 2018 election.In 2015, New Brunswick became the first province to regulate nomination races under political finance law. Under the new political finance regulations, each party must file a Certificate of Nominating Convention that lists the date of the nominating convention, the location of the convention, the contestants for the nomination, and the name of the winner. Nomination candidates must also file disclosure reports that either declare that the funds raised for their nomination campaign were under $2,000 or detail their donations and expenses if they raise $2,000 or more. In addition to the publicly available data, I coded disagreement between party actors over their preferred candidates and the gatekeeping activities used in particular nomination races based on the interviews, participant-observation, and documentary evidence. I do not claim to have identified every case of disagreement and gatekeeping, since party actors are generally reluctant to discuss the backroom politics of nominations. While I identify many cases of disagreement and gatekeeping from questions about specific nomination races, party insiders may not have mentioned every instance of disagreement or gatekeeping. As a result, these known cases serve as lower bounds for the true extent of disagreement and gatekeeping. This possibility of understating the true extent of disagreement and gatekeeping is particularly salient for the PC Party, since PC insiders were more reticent to discuss these topics than Liberal insiders. By contrast, Liberal insiders from the Premier’s Office were willing to discuss any riding of interest to me so that they could emphasize their commitment to nominating 50 percent women in seats not held by Liberal incumbents (Albaugh, 2019). I have included detailed descriptions of the variable coding in the supporting information.

14 The Idea of Preferred Candidates

I began by verifying that party actors have preferred candidates (Observable Implication 1). I discussed this claim extensively with my interviewees. I found that party insiders overwhelmingly recognize that there is a “preferred candidate” or a “favourite.” Nomination candidates often did not think of the process that way, but backroom operatives in both the Liberal and PC parties, along with several nomination candidates, almost unanimously confirmed that party actors did have preferred candidates. Indeed, once I laid out the idea, one former Liberal member ran through each time they ran with me in terms of whether they were the leader’s preferred candidate.2 Importantly, my interviewees highlighted that central and local networks may have different preferred candidates because of their different goals. For example, in Kent North, the preferred candidate of the central network was Dan Murphy, former executive director of the New Brunswick Liberal Association, while the preferred candidate of the riding association was Emery Comeau, former executive assistant to outgoing incumbent Bertrand LeBlanc. The central network viewed Kent North as a good place to pick up a competent minister and privately supported Dan Mur- phy. However, Emery Comeau had deep connections within the riding based on his experience handling case work and his connections with the many members of the Kent North Liberal riding association who were part of the Rogersville-Kouchibouguac Liberal riding association before the 2013 redistribution. This analysis stands in contrast the assumption in American parties research that there is a single, “true” party network that has preferred candidates (Cohen et al., 2009; Hassell, 2016, 2017; Masket, 2009). Preferred candidates tend to emerge through recruitment. Both central and local networks recruit candidates in New Brunswick parties – as in Canadian federal parties (Cross and Young, 2013). Sometimes both networks view the same recruited candidate positively, and there is only one preferred candidate. However, when the two networks disagree, the result is either gatekeeping by one network to force the other network to accept the candidate or a contested nomination in which two candidates have substantial resources to run a nomination race. While party actors overwhelmingly acknowledged that party actors in general have their preferred candidates, many of them were reluctant to divulge which candidates were the preferred candidates of the central or local networks. “The party would never admit to having a favourite....You don’t do it that blatantly.”3 In a rare exception, a Liberal insider at the Premier’s Office ran through all the preferred candidates of the central network in contested nominations.4 Even though party actors clearly do have preferences over potential candidates, as predicted by the theory, they generally do not want to talk openly with outsiders about their preferences. For an explanation of this reluctance, I turn to the next major component of the theory, the norm of non-interference. 2Interview 52, January 15, 2019. 3Interview 22, February 28, 2018. 4Interview 29, multiple dates.

15 The Norm of Non-Interference

Many party actors, particularly in the central party network, face a “norm of non-interference” that constrains their ability to engage in gatekeeping in candidate selection (see Observable Implication 2 above). Across many interviews, party insiders emphasized that the norm was for the party leader’s office, the party’s legislative office, party headquarters, the party’s legislative caucus and the president of the local riding association to remain neutral and not “interfere” in any nomination contest.5 This norm dates back well over a century (Siegfried, 1906), but it has become even stronger with the emergence of norms around intra-party democracy in recent decades (Bille, 2001; Cross and Katz, 2013). Within this theory of gatekeeping, the norm of non-interference explains why actors with strong formal gatekeeping powers, particularly party leaders, do not make use of those powers. Instead, both central and local party networks rely on private, informal means to influence the nomination race, such as controlling the logistics of nominating conventions (Pruysers et al., 2017). The norm of non-interference is not only a matter of principle. It is an enforced social norm. Both party actors and news media enforce this norm. When riding associations perceive “interference” from the centre, they can resist it by bolting from the party.6 For example, the Liberal riding association executive in Moncton Centre resigned after Premier Brian Gallant expelled their member, Speaker Chris Collins, from caucus over allegations of sexual misconduct (Poitras, 2018). Mass resignations deprive the central party organization of a pre-existing local organization for the general election. Journalists play a major role in enforcing this norm, as well. A Liberal from the Premier’s office said, “we [political staffers] are all terrified of the story [of manipulation] in the newspaper.”7 PC insiders likewise agreed that interference risked a backlash, though they downplayed the occurrence of this phenomenon.8 Given this norm, party leaders generally do not use their most powerful gatekeeping powers, such as appointments or refusing to sign a candidates nomination papers for the general election. The Liberal Party gave Premier Brian Gallant the ability to appoint up to three candidates without a vote of party members, primarily for the purpose of increasing women’s representation. In no case did the premier’s office actually do so. Likewise, no party leader used the leader’s veto. These gatekeeping activities not only violate the norm of non-interference but also risk antagonizing a potential candidate who could go to the news media with a story that could damage the party leader. The main exception to this forbearance in using powerful formal gatekeeping activities is blocking candidates after the vetting process. All the major parties have a vetting (or “green- light”) process in which party insiders review a questionnaire completed by the candidate, the results of a criminal background check and a credit check, and a systematic examination of their

5For example, Interview 1, November 15, 2017; Interview 19, Session 2, February 26, 2018. 6Interview 63, August 16, 2019; Interview 69, September 11, 2019. 7Interview 58, July 25, 2019. 8Interview 10, Session 1, December 4, 2017; Interview 19, Session 2, February 26, 2018.

16 social media, press coverage, and publicly available writing. If these processes raise “red flags,” then the party can formally disqualify the candidate – engaging in ballot control. In the two largest parties, a committee that does not include the party leader or their staff reviews the material collected during the vetting process. Since the party leader has no direct involvement in the vetting process, it helps insulate the leader from criticism by blocking undesirable candidates from running.9 However, even disqualifying prospective candidates during the vetting process can risk a media backlash. For example, a disqualified candidate went after the PC Party in the press after their vetting committee disqualified her as a candidate for the Saint Croix by-election. Instead, party actors generally rely on private, informal mechanisms to influence the nominations process. A former Liberal candidate said, “[t]hey [party actors] work behind the scenes. They talk to a few key organizers...but they can’t push it very hard.”10 As a senior Liberal insider described the approach, “[y]ou try to do it without being caught. You quietly, aggressively recruit people you want. You give them advice. Maybe you call a few good organizers. You pick a date at a time that suits your preferred candidate. If you are in a larger, rural riding, you pick a location that suits your preferred candidate.”11. The same insider indicated that these activities “all have plausible deniability. . . .You do things you can deny. If a leader calls someone to tell them not to run, they can’t say they were drunk and called the wrong person.”12 These observations fit with observations of informal influences in candidate nominations in federal and provincial politics (Koop, 2010; Pruysers et al., 2017). Informal gatekeeping activities are often not as powerful as formal activities, but they are often more desirable. While they do not necessarily guarantee success, they avoid the impression that party actors subject to the norm of non-interference are in fact involved in the nominations process. Even when formal rules offer blunt instruments like ballot control or the leader’s veto, party actors usually prefer to use more subtle activities – or none at all. (If the preferred candidate is going to win anyway, there is no need for gatekeeping.) The real power lies in not needing to pick up the blunt instruments in the first place.

The Gatekeeping Repertoire

In order for gatekeeping to be a meaningful concept, party actors must both have and use organizational resources to inﬂuence the candidate selection process (Observable Implication 3). In this section, I outline the gatekeeping activities available to the central and local party networks. I take these activities as being part of “gatekeeping repertoires.” This meaning of repertoire draws on Tilly’s concept of repertoires of contention in social movements scholarship (Tilly, 1986).

9Interview 1, November 15, 2017; Interview 2, Session 2, November 27, 2017. 10Interview 8, December 1, 2017. 11Interview 58, July 25, 2019. 12Interview 58, July 25, 2019.

17 Repertoires of contention include a range of customary protest activities, such as petitions or sit- ins that change over time. Tilly borrowed the idea from musical repertoires. As he put it,

These varieties of action constitute a repertoire in something like the theatrical or musical sense of the word; but the repertoire in question resembles that of comme- diadell’arte or jazz more than that of a strictly classical ensemble: people know the general rules of performance more or less well and vary the performance to meet the purpose at hand (Tilly, 1986, 390).

I take central and local actors as having a range of different gatekeeping repertoires. The central network’s gatekeeping repertoire has actions that are more useful in preventing a contested nomination, while the local network’s gatekeeping repertoire has more actions that are useful in helping a preferred candidate win a contested nomination. Party insiders generally know when, where, and how to use their varieties of gatekeeping and combine activities to obtain their desired outcomes. Table 1 displays a summary of these activities by the actors able to engage in them and their frequency I observed them through interviews and participant-observation. I include counts of each activity where possible. I have organized them by the party actors or networks that can make use of them. Within each set of actors, I organize them from the most common to the least common. Any party actor can, in principle, engage in certain limited gatekeeping activities. These activities include encouraging or discouraging prospective candidates from running, providing advice to preferred candidates, referring volunteers or organizers to nomination campaigns, spreading gossip or rumours about nomination races, or donating under $100 – the legal disclosure threshold – to a nomination campaign. These activities generally do not violate the norm of non-interference (as understood by party insiders) and usually take place in private spaces. Many of these activities are not very powerful. They do not guarantee that a preferred candidate can win. They have most power when a network of party actors works together to make them work, as American parties scholarship suggests (Cohen et al., 2009; Hassell, 2016, 2017; Masket, 2009). While any party actor can in principle engage in these actions, not all party actors are equally effective in doing so. For example, party insiders with more connections and influence are likely to be more effective in referring volunteers and organizers to nomination campaigns or mounting a successful whisper campaign. Likewise, party actors with experience with nomination campaigns are likely to provide better advice than those who do not. Central party actors have a variety of gatekeeping actions through which they can prevent contested nominations, but they have limited opportunities to support a nomination campaign directly. Central party actors cannot openly campaign on behalf of their preferred candidates due to the norm of non-interference. Despite the norm of non-interference, they have a range of gatekeeping activities. Party leaders ultimately have control over the dates of nominating conventions. By custom, the leader speaks at every nominating convention. The party leader can dictate the

18 Table 1: Gatekeeping Activities in New Brunswick’s Two Major Parties

Actor Activity Frequency Any Party Actor Encouraging or discouraging prospective Ubiquitous candidates to run Providing advice to preferred candidates Ubiquitous Referring volunteers and organizers to Common nomination campaigns Spreading gossip or rumours about nominations Rare (1) Donating under $100 to a nomination (Not Observed) campaign

Central Actors Vetoing and approving convention logistics Common Refusing to allow a prospective candidate’s Uncommon (11) name to appear on the nomination ballot (ballot control) Refusing to sign a candidate’s nomination (Not Observed) papers (leader’s veto) Appointing candidates without a convention (Not Observed) Oﬀering prospective candidates jobs not to run (Alleged) Stuﬃng ballot boxes (Not Observed)

Local Actors Proposing convention logistics Ubiquitous Volunteering on nomination campaigns Common Publicly endorsing preferred candidates Uncommon (2) Donating $100 or more to a nomination Rare (1) campaign Note: Disparaged activities are in italics.

19 logistics by being “unavailable” except at the time that suits the central party network’s preferred candidate. This control over the logistics allows them to call “snap conventions” or “early conventions” to benefit their preferred candidates or to hold off calling a convention to allow their preferred candidates time to campaign. This practice was more common in Liberal nominations as part of “affirmative gatekeeping” on behalf of preferred women candidates (Albaugh, 2019). Party vetting committees, which do not include the leaders, can keep candidates off the ballot at nominating conventions by rejecting their application during the vetting process. I call this “ballot control.” This gatekeeping activity was more common in PC nominations (9) than Liberal nominations (2) for the 2018 election.13 Party leaders can refuse to sign a candidate’s nomination papers, which serves as a “leader’s veto” over all local candidates. This practice famously dates back to Stanfield’s refusal to sign Moncton Mayor Leonard Jones’ nomination papers over his opposition to bilingualism and adversarial relationship with Acadians (Erickson and Carty, 1991; Poitras, 2004). Party leaders can appoint candidates without a convention. On September 16, 2017, the Liberal Party’s Board of Directors approved a document entitled“Rules for Nominating Conventions” that gave Gallant the power to appoint up to three women as candidates without a nominating convention. Despite the obvious power of the leader’s veto and appointments, no party leader used these powers except in cases where the party did not have a candidate right before the deadline to file nomination papers. Party leaders and formal party organizations have the ability to offer prospective candidates material incentives not to run. Chris Duffie, a prospective PC candidate in Carleton-York, alleged that PC leader Blaine Higgs and his chief-of-staff, Dominic Cardy, offered him a job not to run against incumbent Carl Urquhart.14 Finally, central party networks usually assist with managing the voting at nominating conventions, which allows them a chance to stuff the ballot boxes at nominating conventions. While I did not observe this in New Brunswick, allegations of ballot box stuffing in Ontario PC nominations emerged in 2017 and 2018. Local party actors have fewer activities through which they can candidates from running in the first place, but they have more activities through which they can help their preferred candidates win nominations. Many local actors, particularly members of the boards of riding associations and former members of the Legislative Assembly (MLAs), are not subject to the norm of non- interference. As a result, they can support their preferred candidates by proposing dates and locations of nominating conventions that would benefit their preferred candidates, volunteering on nominating campaigns, publicly endorsing preferred candidates, and donating $100 of more to a nominating convention. I saw observed the endorsement of a former PC MLA at the PC nominating convention in Carleton-Victoria, and a former PC MLA donated to Jill Green’s campaign for the PC nomination in Fredericton North in 2018. This analysis supports the idea that central and local party actors can and do have organiza-

13For the reference to nine cases in the PCs, see Poitras (2019). The two Liberal cases were Kevin Arseneau in Kent North and Chris Duffie Carleton-York, both of which received considerable media attention in the provincial press. 14Chris Duffie’s affidavit, The Court of Queen’s Bench (Fredericton), FC-159-2017.

20 tional resources through which they engage in gatekeeping in at least some ridings (Observable Implication 3).

Explaining Contested Nominations

In this section, I explain why some nominations became contested using the quantitative data on Liberal and PC nominations for the 2018 New Brunswick Election. First, I show that contested nominations are more likely to occur in non-incumbent ridings and ridings where the party is more competitive (Observable Implication 4). Second, I demonstrate that nominations go uncontested when party actors do not disagree over their preferred candidates (Observable Implication 5a). Third, I show that disagreement between diﬀerent party actors leads either to the use of gatekeeping by central party actors or to the emergence of contested nominations (Observable Implication 5b). Finally, I present quantitative evidence that gatekeeping by central party actors reduces the likelihood of contested nominations (Observable Implication 6). Many of my outcomes of interest – disagreement among party actors over preferred candidates, gatekeeping, contested nominations – are binary. I rely primarily on simple cross-tabulations, for two reasons. First, due to the sequential relationship between district contexts in the general election, disagreement over preferred candidates, gatekeeping, and contested nominations, each predicted bivariate relationship is important to demonstrate on its own. Second, due to the small number of ridings, cases of disagreement over preferred candidates, and contested nominations, the close relationship between intra-party disagreement over preferred candidates and contested nominations in the theory and the data make it diﬃcult to detect other relationships in multivariate analyses. Still, I include logit models estimated with penalized maximum likelihood – which is more appropriate in small samples (Rainey and McCaskey, N.d.) – in the Supporting Information.

Table 2: Electoral Competitiveness, Internal Disagreement, and Contested Nominations for the 2018 New Brunswick Election, by Party

Party Type of District Disagreement (N) Contested (N) Total Liberal Incumbent 5% (1) 0% (0) 20 Safe 80% (4) 60% (3) 5 Competitive 17% (2) 27% (3) 11 Hopeless 8% (1) 0% (0) 13

PC Incumbent 15% (3) 5% (1) 20 Safe 0% (0) 0% (0) 0 Competitive 33% (4) 33% (4) 12 Hopeless 12% (2) 6% (1) 17

21 Table 2 displays the frequency of disagreement among party actors over preferred candidates and contested nominations across four types of districts. Based on the theoretical framework outlined above, I group ridings in which an incumbent is seeking-re-election, then I sub-divide non-incumbent ridings into three categories of perceived electoral competitiveness – safe, competitive, and hopeless. These categories of electoral competitiveness draw on the general perceptions of party insiders and New Brunswick politics expert from my ﬁeld research. Bodet (2013) notes, party insiders tend to view the competitiveness of ridings in subjective categories, rather than continuous variables. These subjective perceptions incorporate information about riding histories, local disgruntlement with incumbents, and other information that objective measures of competitiveness may miss. Finally, subjective perceptions of competitiveness better align with the theory (see Figure 1 above). As an additional check, I include graphical displays of these relationships using the party’s margin of victory or loss in the Supporting Information. As Table 2 shows, both disagreement over preferred candidates and contested nominations are rare in incumbent seats. In non-incumbent seats, the greater the party’s chances of winning the riding, the more likely there is disagreement within the party over preferred candidates, and the more likely the nomination is contested. (As an additional check, I replicate the results using the party’s win margin in 2014 as a measure of competitiveness in the Supporting Information.) This is true for both the Liberals and the PCs.

Table 3: Disagreement Over Preferred Candidates and Contested Nominations for the 2018 New Brunswick Election, by Party

Party Disagreement Uncontested Contested Total Liberal Yes 38% (3) 63% (5) 8 No 100% (41) 0% (0) 41

PC Yes 44% (4) 56% (5) 9 No 98% (39) 2% (1) 40

Next, I examine the extent to which disagreement over preferred candidates leads to contested nominations. Table 3 shows a cross-tabulation of the relationship between pre-convention intra-party disagreement over preferred candidates and a contested nomination. As it shows, disagreement between different actors within the is a very strong predictor of contested nominations. Both these cross-tabulations are highly statistically significant (for the Liberals, χ2 = 22.47 and p < 0.001, and for the PCs, χ2 = 19.25 and p < 0.001). When central and local party actors generally agree on a preferred candidate, the nomination usually goes uncontested (Ob- servable Implication 5a). However, disagreement does not always lead to a contested nomination. This cross-tabulation illustrates that contested nominations are generally the product of disagreement between different actors within the party organization in recruitment, rather than outside

22 Table 4: The Use of Gatekeeping Activities and Contested Nominations, Cases of Disagreement Over Preferred Candidates Only, by Party

Party Gatekeeping Uncontested (N) Contested (N) Total Liberal Yes 3 2 5 No 0 3 3

PC Yes 4 1 5 No 0 4 4 candidates seeking the nomination.15 I turn to the prediction that disagreement will result either in gatekeeping or a contested nomination (Observable Implication 5b). Since this claim only applies in cases in which competing party networks disagree over their choice of candidate, I restrict the analysis to cases in which I observed disagreement in my field research. As Table 4 shows, when different networks in the Liberal and PC parties disagreed over their choice of a preferred candidate, this disagreement produced either gatekeeping or a contested nomination. In this table, positive cases of gatekeeping by central party actors include (1) , (2), (3), or (4) keeping prospective candidates off the ballot by rejecting them during the vetting process (ballot control). Finally, I examine the claim that gatekeeping by central party actors reduces the probability of a contested nomination (Observable Implication 6). Table 4 also illustrates the effectiveness of gatekeeping in cases of intra-party disagreement over preferred candidates. It is important to restrict the analysis to cases of intra-party disagreement because party actors tend not to engage in gatekeeping unless there is some reason why they perceive an interest in influence the nomination. Since gatekeeping and contested nominations both tend to occur in cases of disagreement, similar bivariate analyses on the full dataset actually suggest that gatekeeping increases contested nominations. This result is a bias of not controlling for intra-party disagreement. Although the number of cases is small, Table 4 shows that contested nominations are less likely when central party networks engage in gatekeeping. This pattern exists in both parties. However, gatekeeping does not always work. In both parties, contested nominations still occurred despite the use of gatekeeping. Despite the small number of cases in this table, these cross-tabulations are statistically significant to p < 0.1 (for the Liberals, χ2 = 2.88 and p < 0.1, for the PCs, χ2 = 5.76 and p < 0.05). As an additional test of idea that gatekeeping reduces the probability

15The exception is in the data is the PC nomination Fredericton North. Although I observed informal support from the central party organization for Jill Green at the nominating convention, I did not see any evidence in my interviews or participant-observation of informal support from party actors for the two other nomination candidates. I code the case as one of disagreement in order to be cautious in my generalizations, though I suspect that the case ﬁts with the theoretical expectation the prior intra-party disagreement over preferred candidates explains contested nominations.

23 of a contested nomination, I turn to a qualitative examination of “near-contested” nominations.

Central Party Gatekeeping in Near-Contested Nominations

I examine why party actors engaged in gatekeeping and how their gatekeeping activities prevented contested nominations (as Observable Implication 6 suggests) through case studies of “near- contested” nominations. By near-contested nominations, I mean cases in which two or more nomination candidates expressed an interest in running at the same time but the nomination went uncontested. I focus on near-contested nominations for three reasons. First, near-contested are the most likely cases to find gatekeeping activities. When only one person wants the nomination, party actors have no reason to engage in gatekeeping, as it will not affect the outcome. Likewise, when a nomination becomes contested, it is possible that it went uncontested because party actors did not engage in gatekeeping activities to prevent a contested nomination in the first place. Second, near-contested nominations are the most likely cases to illustrate that gatekeeping activities can meaningfully prevent a prospective candidate from running the nomination. If gatekeeping does not explain “near-contested” nominations, then it is not a very useful concept. Third, these near- contested nominations help substantiate claims about why party actors engage in gatekeeping through a close study of mechanisms. I observed five cases of near-contested nominations. I display these cases in Table 5. For each nomination, I present the party, the riding, the riding context (incumbent seats, safe non-incumbent seats, competitive non-incumbent seats, and hopeless non-incumbent seats), the names of the prospective nomination candidates (when their names are public information), and the gaetkeeping activities observed. I indicate the central party organization’s preferred candidates in bold and candidates blocked or deterred from running using a dagger (†). I observed two near-contested Liberal nominations. In Shediac-Beaubassin-Cap Pelé,one of the safest Liberal seats in New Brunswick, the Premier’s Office attempted to recruit a woman to run.16 However, Shediac Mayor Jacques Leblanc began campaigning for the nomination as soon as it became clear that longtime Liberal incumbent Victor Boudreau was retiring.17 Ultimately, Leblanc deterred the central network’s preferred candidate from running.18 The central party network went along with pressure from the riding association to call a nomination for January 27. This snap convention was enough to deter another interested candidate, Pascal Haché,from running. According to Haché’spublic Facebook page, “There comes a time when we must make difficult decisions in our lives. I had to make a decision, quite probably the most difficult decision [of my life], before the holidays. I came to the conclusion that I will not stand as a candidate at the Liberal convention that will take place on January 27. It would be nearly impossible for

16Interview 29, Session 1, April 3, 2018; Interview 58, Session 1, July 25, 2019. 17Interview 23, Session 1, February 28, 2018. 18Interview 29, Session 1, April 3, 2018; Interview 58, Session 1, July 25, 2019.

24 Table 5: Gatekeeping Activities in Near-Contested Nominations for the 2018 New Brunswick Election

Party Riding Riding Context Nomination Candidates Gatekeeping Activities Liberal Shediac-Beaubassin- Safe Jacques Leblanc Snap Convention, Cap Pel´e Pascal Hach´e† Discouragement Unnamed candidate†

Liberal Fredericton South Competitive Susan Holt Snap Convention Unnamed candidate†

PC Carleton-York Incumbent Carl Urquhart Snap Convention, Chris Duﬃe† Early Convention, Discouragement

PC Moncton Northwest Incumbent Ernie Steeves Snap Convention, Don Moore† Discouragement

PC Saint Croix Competitive Greg Thompson Discouragement Cynthia Barlett†

25 me to launch myself into this adventure [a nomination campaign] this winter with the imminent arrival of our son Zacharie who is supposed to be born in the next few weeks (January 28).”19 Meanwhile, in Fredericton South, the central party network called a snap convention to help preferred candidate Susan Holt win the nomination.20 Although a man expressed an interest in running before the holidays, the Premier’s Office’s arranged for a nominating convention on relatively short notice in January. The man decided not to run. After this snap convention call, Susan Holt won the nomination by acclamation. For the PCs, I found three near-contested nominations. The most well-documented case is the PC nomination in Carleton-York. There, Chris Duffie sought to challenge incumbent PC MLA Carl Urquhart. Duffie previously ran for a PC nomination for the 2010 election. He expressed an interest in running as the PC candidate in Carleton-York after Urquhart retired. Urquhart took him as a protégéand gave him “exposure” by involving him in constituency work.21 Until late in 2016, the understanding was that Urquhart was not going to run in the next election.22 Urquhart changed his mind, and Duffie had a falling out with Urquhart over that decision. In 2017, Duffie spent months campaigning for the nomination.23 One PC insider suggested that Duffie would have been “the favourite” if Urquhart had not run again.24 However, when Duffie decided to challenge Urquhart, central and local PC insiders took several actions to prevent him from appearing on the ballot at the nominating convention. PC central party insiders made it very clear in interviews that they did not want Duffie to run, especially against a popular incumbent.25 In April 2017, Fredericton party actors tried to dissuade Duffie from running against a sitting MLA.26 According to an interviewee in the room at the time, one of them suggested that Duffie could run in another riding, such as Fredericton South.27 Fredericton South was a much less desirable seat, since Green Party leader David Coon had won it in a four-way race in 2014. Duffie claimed that he had a meeting with PC Leader Blaine Higgs and his Chief of Staff Dominic Cardy in which Cardy offered Duffie a job not to run, though Cardy denied this.28

19This is a translation from Pascal Haché’sFacebook page, January 18, 2018 (accessed 6 March 2019). Here is the original: “Vient un temps oùnous devons prendre des décisionsdifficiiles dans notre vie. J’ai dûen prendre une avant les fêtes,fort probablement LA plus difficile, et en venir àla conclusion que je ne me présenteraispas à l’investiture libérale de Shédiac–Beaubassin–Cap Peléqui aura lieu le 27 janvier. Il étaitpractiquement impossible pour moi de me lancer dans cette aventure avec une nomination cet hiver, avec la venue imminent de notre fils Zacharie qui devrait faire son arrivéedans nos vies dans les prochaines semaines (28 janvier). 20Interview 20, Session 2, February 16, 2018 21Interview 17, January 10, 2018; Interview 39, May 28, 2018. 22Interview 17, January 10, 2018; Interview 39, May 28, 2018. 23Interview 17, January 10, 2018; Interview 19, Session 1, January 16, 2018; Interview 22, Session 1, February 23, 2018. 24Interview 38, May 28, 2018. 25Interview 19, Session 1, January 16, 2018; Interview 22, Session 1, February 23, 2018. 26Interview 17, January 10, 2018; Interview 19, Session 1, January 16, 2018. 27Interview 17, January 10, 2018. 28See Cardy and Duffie’s affidavits, Court of Queen’s Bench (Fredericton), FC-159-2017.

26 On May 4, 2017, Progressive-Conservative (PC) Party of New Brunswick president and acting executive director Don Moore sent out an email to its members advising them of upcoming nomination contests – the first email communication of the 2017-2018 nominations cycle. It did not mention that May 5 was the deadline for submitting paperwork to the party to run in the riding of Carleton–York. Instead, on May 4, the party organization notified potential nomination contestants and the party membership of the convention through an ad on page C3 of the Life section of the Fredericton Daily Gleaner.29 The ad called the convention for May 19, 2017-—the Friday of Victoria Day long weekend. The party set the location as Canterbury High School—-a relatively small venue in a rural community in the southwestern part of the riding where PC leader Blaine Higgs had attended. Duffie did not find out about the nomination call until he ran into someone at Costco on May 6. He sent an email to PC party president and acting executive director Don Moore that day declaring his intent to run.30 The following Monday, May 8, he went to the PC Party headquarters to try to submit his paperwork to contest the nomination, along with membership forms and payment for dues. A few members of his campaign team went with him. Progressive- Conservative party president and acting executive director Don Moore called the police when party employees let him know about the situation. In the end, the police escorted Duffie into the office – alone – to drop off his documents. The party organization rejected his application to contest the nomination for being late. Duffie later brought a court case against the Progressive-Conservative Party of New Brunswick for violating its formal rules, which he lost. After Duffie lost the court case, he sought the Liberal nomination in the same riding. De- spite support from the Liberal riding association, the Liberal green-light committee rejected his application to run during the vetting process, in large part due to statements he made that were critical of Premier Gallant on social media. Although the Carleton-York case is the most well-documented, there were two other near- contested PC nominations. In Moncton Northwest, former party president Don Moore filed paperwork to run against incumbent PC MLA Ernie Steeves.31 According to one interviewee, central party actors met with Moore to convince him that it was not in his best interest to run against a sitting incumbent, since it could ruin his chances of running in the future.32 Then,

29Duffie argued that putting it in the Daily Gleaner rather than the Telegraph-Journal (Saint John) was an attempt to make it so people would not hear about the nominating convention. Many New Brunswick politicos, even in Fredericton, read the Saint John Telegraph-Journal rather than the Daily Gleaner (Belliveau, 1980, 180). Stewart Fairgrieve, former executive director of the Progressive-Conservative Party of New Brunswick (PCNB), argued in a court affidavit that the party could just have easily put it in the Woodstock Bugle Observer, which has even lower circulation numbers but serves the northwestern portion of the riding. Even if it was not a deliberate act of gatekeeping, it appears to have contributed to the convention taking Duffie off-guard. See the records of the Court of Queen’s Bench (Fredericton), FC-159-2017. 30This email is available in the records of the Court of Queen’s Bench (Fredericton), FC-159-2017. 31This information is a matter of public record, since it appears on the Certificate of Nominating Convention for the PC nomination in Moncton Northwest. 32Interview 22, Session 2, March 9, 2018. It was not in the central party network’s interests to have a contested

27 the central party organization arranged for a snap convention. According to a Moncton PC insider, Moore withdrew from the race because he knew he was not going to unseat an incumbent candidate without time to sign up his supporters as members.33 In Saint Croix, Cynthia Bar- lett announced her interest in running for the nomination, but later longtime former MP Greg Thompson decided to run. Since Thompson won every election in the federal riding that includes Saint Croix from 1988 to 2008, PC insiders overwhelmingly viewed him as the best chance to win the riding. According to a Saint Croix PC familiar with the case, party insiders met with Cynthia to discourage her from running, and she withdrew before the convention because of this discouragement.34 Although I focus on gatekeeping in near-contested nominations in this section, gatekeeping also occurs in other cases. For example, sometimes contested nominations still occur despite gatekeeping efforts. For example, in the Southwest Miramichi–Bay du Vin Liberal nomination, the Premier’s office wanted to make sure that their preferred candidate, Karen Howe, won the nomination to ensure that the party met the Premier’s goal of nominating fifty percent women in seats not held by Liberal incumbents.35 The riding association and the nomination candidates had agreed amongst themselves on an April convention to avoid winter weather but allow members from the the lobster-fishing community of Baie-Sainte-Anne to participate before the start of fishing season.36 However, the Premier’s office arranged for a snap convention in January once they caught word that Andy Hardy, president and a vocal critic of Brian Gallant, was planning on running for the Liberal nomination.37 According to two local Liberals, Hardy had the support of a member of the riding association executive who had experience shepherding preferred candidates through past contested nominations.38 The snap convention was not enough to deter Andy Hardy from entering the race and submitting his nomination paperwork on time. At the convention, Hardy defeated Howe 88-49. Party actors may engage in gatekeeping even if there is no alternative candidate interested in running. For example, the PCs rejected 2014 People’s Alliance as a candidate in Kent North during a time when the PCs were attempting to find ways to find candidates and win in pre- dominantly francophone areas.39 In the end, central party actors selected Katie Robertson as the candidate for Kent North at last minute because they still did not have a nominated candidate. For an alternative example, in at least six ridings, the Liberal Premier’s Office arranged for early nominating conventions in ridings where women had agreed to run to help them win their

nomination, since Moncton Northwest was swing riding. Party insiders noted that a divisive nomination could hurt the party in a general election and that attempts to challenge sitting incumbents are usually divisive 33Interview 22, Session 2, March 9, 2018. 34Interview 49, September 6, 2018. 35Interview 29, Session 2, May 14, 2018; Interview 35, April 13, 2018 36Interview 34, April 12, 2018; Interview 35, April 13, 2018. 37Interview 29, Session 2, May 14, 2018; Interview 35, April 13, 2018 38Interview 34, April 12, 2018; Interview 35, April 13, 2018. 39Interview 44, Session 3, January 18, 2019. For additional evidence on the rejection, see MacIntosh (2018).

28 nominations uncontested before any men decided to run.40 All six of these cases were successful. Ultimately, central party actors’ gatekeeping activities explain the ﬁve cases of near-contested nominations I examine in this section, along with several cases that do not qualify as near- contested nominations. These case studies illustrate not only the use of gatekeeping activities but also the reasons and consequences of those gatekeeping activities in practice. Although central party actors generally have the power to block any prospective candidate from running for any reason by keeping that candidate oﬀ the ballot at nominating conventions, or by refusing to sign a candidate’s nomination papers, or (in the case of the Liberals) by appointing a candidate outright, they do not make use of these powers. They violate the norm of non-interference, and they typically garner negative news coverage. As a result, central party actors rely on other gatekeeping activities that are less reliable – and sometimes these gatekeeping activities do not do enough to help the central party network’s preferred candidates win nominations.

Discussion

In this paper, I present a theory of gatekeeping that draws on rational choice and sociological institutionalism. This theory generates several observable implications. I present evidence for these observable implications from 20 months of fieldwork in New Brunswick; participant-observation of 25 nominating conventions; 93 interviews with 69 party insiders, nomination candidates, and other individuals familiar with various aspects of the nominations process; and an original dataset of nominations for the 2018 New Brunswick election constructed from electoral results, political finance data, and my fieldwork. The data provide previously uncollected quantitative variables that allow me to demonstrate the extent of gatekeeping and the role of intra-party disagreement and gatekeeping in predicting contested nominations. Although this study draws on material from a New Brunswick provincial election, the theoretical expectations outlined in this paper are likely to travel outside of New Brunswick. Federal and provincial parties in Canada typically share similar formal rules around nomination races and similar gatekeeping activities, such as controlling the logistics of nominating conventions. For this reason, the claims about gatekeeping in this paper are particularly likely to travel to federal politics and provincial politics in other provinces. While the theory aims to explain gatekeeping in Canadian party nominations, it generates new theoretical claims for scholars of candidate selection in other countries, including the United Kingdom and the United States. In particular, it provides an explanation of why different actors within the party organization may support different preferred candidates and how their gatekeeping activities shape which candidates actually win party nominations. The extent of gatekeeping by central party actors shown in this paper suggests a need to re- evaluate research on political parties, candidate selection, and the representation of marginalized

40Interview 20, Session 2, February 16, 2018; Interview 40, June 28, 2018; Interview 42, July 6, 2018; Interview 58, July 25, 2019; Interview 59, July 25, 2019

29 groups in Canada. If central party actors engage in gatekeeping on behalf of their preferred candidates, then models of candidate selection that focus on riding association executives – such as models of how the gender or race of riding association executives predict the gender or race of nominated candidates (Cheng and Tavits, 2011; Cross and Pruysers, 2019; Medeiros, Forest, and Erl, 2019; Tolley, 2019) – omit theoretically important variables. Gatekeeping by central party actors likewise challenges Carty’s ideas of a “stratarchical bargain” or a “franchise model” of parties, both of which assert a division of labour between central and local party organizations in which central party actors set policy and local party actors select candidates (Carty, 2002, 2004).41 Ultimately, gatekeeping casts doubt on the fairness of intra-party elections, which is essential to intra-party democracy.

41For a similar view on Carty’s ideas, see Cross (2018).

30 References

Albaugh, Quinn M. 2019. “Affirmative Gatekeeping: Informal Influences on Behalf of Women Candidates.” Paper presented at the annual meeting of the American Political Science Associ- ation, Washington, DC. Albaugh, Quinn M. 2020. Gatekeeping and Group Representation. Doctoral Dissertation, Prince- ton University. Baisley, Katherine Elizabeth Ann. 2020. Interest Group Strategies and Party Position Change: The Case of Canadian Social Conservatives. Doctoral Dissertation, Princeton University. Belliveau, John Edward. 1980. Little Louis and the Giant KC. Hantsport, NS: Lancelot Press. Bille, Lars. 2001. “Democratizing a Democratic Procedure: Myth or Reality? Candidate Selection in Western European Parties, 1960-1990.” Party Politics 7(3): 363–380. Bittner, Amanda. 2011. Platform or Personality?: The Role of Party Leaders in Elections. Oxford, UK: Oxford University Press. Bittner, Amanda. 2018. “Leaders always mattered: The persistence of personality in Canadian elections.” Electoral Studies 54: 297–302. Bjarneg˚ard,Elin. 2013. Gender, Informal Institutions and Political Recruitment: Explaining Male Dominance in Parliamentary Representation. London, UK: Springer. Bjarneg˚ard,Elin, and PärZetterberg. 2016. “Political Parties and Gender Quota Implementation: The Role of Bureaucratized Candidate Selection Procedures.” Comparative Politics 48(3): 393– 417. Blais, André.2005. “Accounting for the Electoral Success of the Liberal Party in Canada: Presi- dential address to the Canadian Political Science Association London, Ontario June 3, 2005.” Canadian Journal of Political Science 38(4): 821–840. Bodet, Marc André.2013. “Strongholds and battlegrounds: measuring party support stability in Canada.” Canadian Journal of Political Science 46(3): 575–596. Cain, Bruce E., John A. Ferejohn, and Morris P. Fiorina. 1987. The Personal Vote: Constituency Service and Electoral Independence. Cambridge, MA: Harvard University Press. Carty, R. Kenn, D. Munroe Eagles, and Anthony Sayers. 2003. “Candidates and Local Campaigns Are there Just Four Canadian Types?” Party Politics 9(5): 619–636. Carty, R. Kenneth. 1991. Canadian Political Parties in the Constituencies. Toronto, ON: Dundurn Press.

31 Carty, R. Kenneth. 2002. “The Politics of Tecumseh Corners: Canadian Political Parties as Franchise Organizations.” Canadian Journal of Political Science 35(4): 723–745.

Carty, R. Kenneth. 2004. “Parties as Franchise Systems: The Stratarchical Organizational Im- perative.” Party Politics 10(1): 5–24.

Carty, R. Kenneth, and Lynda Erickson. 1991. “Candidate Nomination in Canada’s National Political Parties.” In Canadian Political Parties: Leaders, Candidates and Organization, ed. Herman Bakvis. Toronto, ON: Dundurn Press.

Carty, R. Kenneth, and Munroe Eagles. 2005. Politics is Local: National Politics at the Grassroots. Don Mills, ON: Oxford University Press.

Cheng, Christine, and Margit Tavits. 2011. “Informal Inﬂuences in Selecting Female Political Candidates.” Political Research Quarterly 64(2): 460–471.

Cohen, Marty, David Karol, Hans Noel, and John Zaller. 2009. The Party Decides: Presidential Nominations Before and After Reform. Chicago, IL: University of Chicago Press.

Courtney, John C. 1978. “Recognition of Canadian Political Parties in Parliament and in Law.” Canadian Journal of Political Science 11(1): 33–60.

Courtney, John C. 2015. “Canada’s National Parties: From Private to Public Institutions.” In Parties and Party Systems: Structure and Context, eds. Richard Johnston, and Campbell Sharman. Vancouver, BC: UBC Press , 202–223.

Cross, William. 2004. Political Parties. Vancouver, BC: UBC Press.

Cross, William. 2018. “Understanding Power-Sharing within Political Parties: Stratarchy as Mutual Interdependence between the Party in the Centre and the Party on the Ground.” Government and Opposition 53(2): 205–230.

Cross, William, and Lisa Young. 2013. “Candidate Recruitment in Canada: The Role of Political Parties.” In Parties, Elections, and the Future of Canadian Politics, eds. Amanda Bittner, and Royce Koop. Vancouver, BC: UBC Press , 24–45.

Cross, William P., and Richard S. Katz. 2013. The Challenges of Intra-Party Democracy. Toronto, Canada: Oxford University Press.

Cross, William P., and Scott Pruysers. 2019. “The Local Determinants of Representation: Party Constituency Associations, Candidate Nomination and Gender.” Canadian Journal of Political Science 52(3): 557–574.

32 Cross, William P., Ofer Kenig, Scott Pruysers, and Gideon Rahat. 2016. Promise and Challenge of Party Primary Elections: A Comparative Perspective. Montreal, QC and Kingston, ON: McGill-Queen’s University Press.

Dallinger, Frederick William. 1903. Nominations for Elective Oﬃce in the United States. New York, NY: Longmans, Green, and Company.

Erickson, Lynda, and R Kenneth Carty. 1991. “Parties and Candidate Selection in the 1988 Canadian General Election.” Canadian Journal of Political Science 24(2): 331–349.

Erikson, Robert S. 1971. “The Advantage of Incumbency in Congressional Elections.” Polity 3(3): 395–405. Publisher: The University of Massachusetts Press.

Gallagher, Michael. 1980. “Candidate selection in Ireland: The impact of localism and the electoral system.” British Journal of Political Science 10(4): 489–503. Publisher: JSTOR.

George, Alexander L., and Andrew Bennett. 2005. Case Studies and Theory Development in the Social Sciences. Cambridge, MA: MIT Press.

Hall, Peter A, and Rosemary CR Taylor. 1996. “Political science and the three new institution- alisms.” Political studies 44(5): 936–957.

Hassell, Hans J. G. 2016. “Party Control of Party Primaries: Party Inﬂuence in Nominations for the U.S. Senate.” The Journal of Politics 78(1): 75–87.

Hassell, Hans J. G. 2017. The Party’s Primary: Control of Congressional Nominations. Cambridge, UK: Cambridge University Press.

Hazan, Reuven Y., and Gideon Rahat. 2010. Democracy within Parties: Candidate Selection Methods and Their Political Consequences. Oxford, UK: Oxford University Press.

Hirano, Shigeo, and James M Snyder Jr. 2019. Primary Elections in the United States. Cambridge, UK: Cambridge University Press.

Kenny, Meryl, and T`aniaVerge. 2012. “Decentralization, Political Parties, and Women’s Repre- sentation: Evidence from Spain and Britain.” Publius: The Journal of Federalism 43(1): 109– 128.

Kenny, Meryl, and T`aniaVerge. 2016. “Opening Up the Black Box: Gender and Candidate Selection in a New Era.” Government and Opposition 51(3): 351–369.

Key, Vladimir. 1964. Politics, Parties, and Pressure Groups. Fifth ed. New York, NY: Thomas Y. Crowell and Company.

33 Key, V.O. 1949. Southern Politics: In State and Nation. New York, NY: Knopf.

Koene, Miriam. 2008. “Targeted Representation? An Analysis of the Appointment of Liberal Candidates in the 1993 and 1997 Federal Elections.” Past Imperfect 7: 55–86.

Koger, Gregory, Seth Masket, and Hans Noel. 2009. “Partisan Webs: Information Exchange and Party Networks.” British Journal of Political Science 39(3): 633–653.

Koop, Royce. 2010. “Professionalism, Sociability and the Liberal Party in the Constituencies.” Canadian Journal of Political Science 43(4): 893–913.

Koop, Royce. 2011. Grassroots Liberals: Organizing for Local and National Politics. University of British Columbia Press.

Koop, Royce, and Amanda Bittner. 2011. “Parachuted into Parliament: Candidate Nomination, Appointed Candidates, and Legislative Roles in Canada.” Journal of Elections, Public Opinion & Parties 21(4): 431–452.

Koop, Royce, Heather Bastedo, and Kelly Blidook. 2018. Representation in Action: Canadian MPs in the Constituencies. Vancouver, BC: UBC Press.

Krook, Mona Lena. 2010. “Beyond Supply and Demand: A Feminist-Institutionalist Theory of Candidate Selection.” Political Research Quarterly 63(4): 707–720.

MacIntosh, Jo¨el.2018. “I unfortunately have been refused as a candidate for the Progres- sive Conservative Party of New Brunswick. I want to take a moment and thank every- one for their support over the years and in the past few months 1/2.” Twitter.com. https://twitter.com/JMacIntoshNB/status/1017232046199070720 (accessed 30 April 2020).

Marland, Alex, and Brooks DeCillia. 2020. “Reputation and Brand Management by Political Parties: Party Vetting of Election Candidates in Canada.” Journal of Nonproﬁt & Public Sector Marketing 32(4): 342–363. Publisher: Taylor & Francis.

Masket, Seth E. 2009. No Middle Ground: How Informal Party Organizations Control Nomina- tions and Polarize Legislatures. Ann Arbor, MI: University of Michigan Press.

Medeiros, Mike, Benjamin Forest, and Chris Erl. 2019. “Where Women Stand: Parliamentary Candidate Selection in Canada.” Politics, Groups, and Identities 7(2): 389–400.

Norris, Pippa. 1996. “Legislative Recruitment.” In Comparing Democracies: Elections and Voting in Global Perspective, eds. Lawrence LeDuc, Richard G. Niemi, and Pippa Norris. London, UK: Sage.

34 Ostrogorski, Moisei. 1903. Democracy and the Organization of Political Parties. Vol. 2. London, UK: Macmillan.

Poitras, Jacques. 2004. The Right Fight: Bernard Lord and the Conservative dilemma. Freder- icton, NB: Goose Lane Editions.

Poitras, Jacques. 2018. Liberals in Moncton Centre need time to ’heal,’ says would-be candidate. Publication Title: CBC News.

Poitras, Jacques. 2019. “PC Riding President Resigns After Party Bars Potential St. Croix candidate.” CBC News, December 31. Available online at https://www.cbc.ca/news/canada/new- brunswick/st-croix-riding-byelection-1.5411647 (accessed 23 June 2020).

Pruysers, Scott, William Cross, Anika Gauja, and Gideon Rahat. 2017. “Candidate Selection Rules and Democratic Outcomes: The Impact of Parties on Women’s Representation.” In Organizing Political Parties: Representation, Participation, and Power. Oxford, UK: Oxford University Press , 208–233.

Rainey, Carlisle, and Kelly McCaskey. N.d. “Estimating Logit Models with Small Samples.” Political Science Research and Methods: Forthcoming.

Ranney, Austin. 1965. Pathways to Parliament: Candidate Selection in Britain. London, UK: Macmillan.

Rush, Michael. 1969. The selection of Parliamentary candidates. Nelson’s political science library London, UK: Nelson.

Sayers, Anthony M. 1999. Parties, Candidates, and Constituency Campaigns in Canadian Elec- tions. Vancouver, BC: UBC Press.

Scarrow, Howard A. 1964. “Nomination and Local Party Organization in Canada: A Case Study.” The Western Political Quarterly 17(1): 55–62.

Schwartz, Mildred A. 1990. The Party Network: The Robust Organization of Illinois Republicans. Madison, WI: University of Wisconsin Press.

Siegfried, Andr´e.1906. Le Canada, les deux races: probl`emespolitiques contemporains. Paris, A. Colin.

Stevens, Benjamin Allen, Md Mujahedul Islam, Roosmarijn de Geus, Jonah Goldberg, John R. McAndrews, Alex Mierke-Zatwarnicki, Peter John Loewen, and Daniel Rubenson. 2019. “Local Candidate Eﬀects in Canadian Elections.” Canadian Journal of Political Science 52(1): 83–96.

35 Thomas, Paul E. J., and Michael Morden. 2019. Party Favours: How Federal Election Candidates Are Chosen. Toronto, ON: Samara Centre for Democracy.

Tilly, Charles. 1986. The Contentious French: Four Centuries of Popular Struggle. Cambridge, MA: Harvard University Press.

Tolley, Erin. 2019. “Who You know: Local Party Presidents and Minority Candidate Emergence.” Electoral Studies: Forthcoming.

Tremblay, Manon, and R´ejeanPelletier. 2001. “More Women Constituency Party Presidents A Strategy for Increasing the Number of Women Candidates in Canada?” Party Politics 7(2): 157–190.

Young, Lisa, and William Cross. 2002. “Incentives to Membership in Canadian Political Parties.” Political Research Quarterly 55(3): 547–569.