Introduction
Despite achieving considerable gains in recent decades, women remain vastly underrepresented in national legislatures around the world. As of 2018, women accounted for just one-in-five elected lower house representatives (Inter-Parliamentary Union, Reference Union2018). While this figure is up from one-in-ten elected representatives in the late 1990s, the advances seem to have plateaued in recent years. In Canada, the proportion of women in the federal House of Commons has changed very little in the six elections since 1997 (increasing from 21 to 26 per cent in two decades). While the factors that contribute to the underrepresentation of women are complex and diverse, the comparative literature has typically focused on three distinct levels of analysis: the political system, the individual and the political party.
At the level of the political system (that is, country level), a number of factors have been identified as potential barriers to women's electoral success. These include the electoral system (Caul, Reference Caul1999; Krook et al., Reference Krook, Lovenduski and Squires2009; Kunovich and Paxton, Reference Kunovich and Paxton2005; Matland and Studlar, Reference Matland and Studlar1996; Schwindt-Bayer and Mishler, Reference Schwindt-Bayer and Mishler2005), the absence of national/constitutional quotas (Dahlerup, Reference Dahlerup2007; Krook, Reference Krook2004, Reference Krook2009; Franceschet et al., Reference Franceschet, Krook and Piscopo2012), the ideological composition of the party system (Caul, Reference Caul1999; Vandeleene, Reference Vandeleene2014) and socioeconomic and cultural factors (Inglehart and Norris, Reference Inglehart and Norris2003; Paxton and Kunovich, Reference Paxton and Kunovich2003). In general, proportional representation electoral systems, hard quotas, party systems with more (and competitive) leftist parties, and higher levels of educational attainment and employment among women in the population tend to result in more women being elected to Parliament.
At the level of the individual, a great deal of attention has been devoted to differences in nascent political ambition—that is, in a desire or interest in pursuing a career in politics. A considerable, and growing, body of research from Canada (Pruysers and Blais, Reference Pruysers and Blais2017; Pruysers and Blais, Reference Pruysers and Blais2018a, Reference Pruysers and Blais2018b), the United States (Holman and Schneider, Reference Holman and Schneider2018; Fox and Lawless, Reference Fox and Lawless2010; Lawless and Fox, Reference Lawless and Fox2005) and the United Kingdom (Allen and Cutts, Reference Allen and Cutts2018) reveals that men are significantly more interested in a political career than women and that women are significantly less confident in their political abilities and qualifications. Importantly, these gender gaps in ambition are coupled with the fact that women tend to respond less positively to recruitment efforts than men (Preece et al., Reference Preece, Stoddard and Fisher2016) and that party recruiters have been shown to hold stereotypical views regarding the political acumen of women candidates (Sanbonmatsu, Reference Sanbonmatsu2006).
The third level of analysis considers political parties as important gatekeepers to elected office (Caul, Reference Caul2001; Norris and Lovenduski, Reference Norris and Lovenduski1995). Caul (Reference Caul1999: 80), for example, notes that there is considerable variation among parties within countries: “Parties differ in the number of women they nominate, where they rank women on party lists, and the proportion of women that they send to parliament.” The major party-level factors that influence the representation of women include the composition of party selectorates (Rahat et al., Reference Rahat, Hazan and Katz2008; see also Pruysers et al., Reference Pruysers, Cross, Gauja, Rahat, Scarrow, Webb and Poguntke2017), candidate selection rules (Kenny, Reference Kenny2013; Krook, Reference Krook2010; Matland and Studlar, Reference Matland and Studlar1996; Vandeleene, Reference Vandeleene2014), party ideology (Caul, Reference Caul1999; Pruysers et al., Reference Pruysers, Cross, Gauja, Rahat, Scarrow, Webb and Poguntke2017) and the presence or absence of voluntary party quotas (Aldrich, Reference Aldrich2018; Childs, Reference Childs, Cross and Katz2013; Krook, Reference Krook2014) or all women's short-lists (Cutts et al., Reference Cutts, Childs and Fieldhouse2008). The evidence from the party level suggests that women tend to perform better in parties that have implemented voluntary party quotas and among parties of the left. The influence of smaller and centralized selectorates, however, is somewhat more contested.
Largely left out of this party-level story is what political parties do on the ground.Footnote 1 For many parties around the world, it is their local organization that plays a crucial role in identifying, recruiting and nominating candidates for the general election (see Cross, Reference Cross2016b; Hazan and Rahat, Reference Hazan and Rahat2010; Pruysers and Cross, Reference Pruysers and Cross2016). While parties have been identified as an important focal point of analysis, the vast majority of this literature considers the party at the national level with very little attention devoted to local organizations. In many ways, this literature portrays parties as unitary actors, as it focuses on national-level party rules governing candidate selection. Yet there is obvious variance within parties operating in both single and multi-member districts, as some electoral district associations (EDA) will choose a woman to be their candidate while others will not. It is the relationship between these differences among local party associations in geographically defined systems and the political representation of women that we are interested in.
While much of the existing literature explores the representational outcome of candidate nominations (that is, those who win the party's nomination and, subsequently, those who are elected), little attention has been paid to the nomination contest itself. The dependent variable in these analyses is typically whether a woman is selected to be the party's general election candidate (see, for instance, Cheng and Tavits, Reference Cheng and Tavits2011; O'Neill and Thomas, Reference O'Neill, Thomas, Pammett and Dornan2015; Pruysers et al., Reference Pruysers, Cross, Gauja, Rahat, Scarrow, Webb and Poguntke2017). However, the possibility of a woman being chosen as the general election candidate is dependent on a number of factors, including whether a woman sought the nomination. Research suggests that when women seek party nominations, they perform quite well—generally, on par with men (see, for example, Erickson, Reference Erickson and Megyery1991; Cross Reference Cross2016a). The real deficit appears to be in the number of women seeking nominations. While it is important to know what factors contribute to women winning, we also need to understand the factors that contribute to women running. Thus we adopt a unique approach and use whether or not a woman sought the local party nomination as our dependent variable.
The purpose of this article, then, is to explore those factors that are specific to the local nomination contest that may result in greater diversity in the candidate selection process and, by extension, the general election. In particular, we are interested in uncovering the district or riding-specific party factors that are related to women's participation in the candidate nomination stage. Why do some nomination races in a party have no women contestants, while others have many? In answering this question, we make two important contributions to the literature. First, we push the analysis further, in order to consider the nomination contest itself and not just the outcome. Second, we push the analysis “lower,” in order to consider the local determinants of a diverse candidate nomination race.
Theory and Hypotheses
Why should we be interested in political parties, especially their local organizations and activities? Political parties are important gatekeepers to elected office. Lilliefeldt (Reference Lilliefeldt2012: 194), for example, writes that “parties are the main architects of parliamentary representation.” That parties act as crucial gatekeepers to elected office is well established in the literature (Caul Reference Caul1999; Erickson, Reference Erickson and Megyery1991; Krook, Reference Krook2010; Lovenduski and Norris, Reference Lovenduski and Norris1993; Norris and Lovenduski, Reference Norris and Lovenduski1995; Norris, Reference Norris, Katz and Crotty2006), as in most countries winning a party nomination is the essential prerequisite for being elected to Parliament (Cross et al., Reference Cross, Kenig, Pruysers and Rahat2016). The nomination of party candidates determines the pool of individuals from which voters will cast their ballot and create their legislature and, as a result, the nomination has important implications for representation and the quality of democracy more generally.
Despite their important role as gatekeepers, parties are not unitary or necessarily coherent actors. As Carty (Reference Carty2004:6) explains, the traditional conception of a political party as “a single identifiable organization that some group can capture and command” may no longer accurately describe how parties actually operate and organize. Instead, parties may be better understood as engaging in a stratarchical bargain, in which a division of labour provides different “levels” of the party (that is, party on the ground, party in central office) with discrete functions to perform (Bolleyer, Reference Bolleyer2012; Carty, Reference Carty2002; Carty and Cross, Reference Carty and Cross2006; Eldersveld, Reference Eldersveld1964). The degree to which these “levels” have complete independence, however, is contested (Cross, Reference Cross2016b; Cross and Gauja, Reference Cross and Gauja2014). While interdependence (as opposed to mutual independence) may be a better way of describing the relationship, local party organizations should not be overlooked, especially when considering candidate nomination procedures and representational outcomes. After all, it is the local party association in many countries that identifies, recruits and ultimately nominates candidates to contest the general election.
As Pruysers and Cross (Reference Pruysers and Cross2016) note, local party organizations not only nominate candidates but also take an active role in recruitment. According to Cross and Young (Reference Cross, Young, Bittner and Koop2013), 38 per cent of Liberal and New Democrat candidates for the 2008 Canadian general election were specifically recruited by the local party association and 22 per cent were recruited by both the local and national party organizations. The importance of the local association should not be surprising. A number of Canadian parties have a history of “freezing” nominations until local search committees have made significant efforts to identify women, as well as members of other underrepresented groups, willing to contest the nomination (Cross, Reference Cross, Pammett and Dornan2006; Pruysers and Cross, Reference Pruysers and Cross2016). In this regard, a local—and formal—candidate search can be an important element to the nomination contest. It is unclear, however, how well practised these formal searches are at the local level, as it has been long recognized that there can be differences between formal rules and informal practices. Moreover, since women are less likely to be “self-starters” or “political entrepreneurs” when it comes to political candidacy (Cross and Young, Reference Cross, Young, Bittner and Koop2013; see also Carroll and Sanbonmatsu, Reference Carroll and Sanbonmatsu2013), a formal candidate search should benefit potential women candidates. Thus, our first hypothesis:
H1:
Local party associations that conducted a formal candidate search will be more likely to have a woman in the nomination contest.
In addition to a formal candidate search, informal means of local recruitment and encouragement are likely important as well. Recent research suggests that women are significantly more likely to be (successfully) nominated in districts where a woman serves as president of the local constituency executive (Cheng and Tavits, Reference Cheng and Tavits2011; Cross, Reference Cross2016a). A similar finding is shown cross-nationally by Pruysers et al. (Reference Pruysers, Cross, Gauja, Rahat, Scarrow, Webb and Poguntke2017), who find that parties with more women on the party's national executive also nominate more women candidates. Cheng and Tavits (Reference Cheng and Tavits2011: 461–62) offer a number of reasons for the link between women on the executive and women's candidacy. First, having a woman in a position of party authority provides an encouraging signal to other women that they are welcome to be politically active in the party. In this respect, it should challenge the notion that women are “outsiders.” Second, gatekeepers often recruit from their own networks, and when the president is a woman, she is more likely to know, and therefore recruit, other qualified women. Indeed, Crowder-Meyer (Reference Crowder-Meyer2013) finds that female party recruiters are more likely to recruit women than are male party recruiters, and Tremblay and Pelletier (Reference Tremblay and Pelletier2001) report that women association presidents in Canada are more likely than their male counterparts to support measures that promote women's candidacy. Furthermore, Ashe and Stewart (Reference Ashe and Stewart2012) suggest that gatekeeper discrimination, and not an undersupply of aspirants, is at the core of underrepresentation. These studies not only document the importance of demand-side explanations but also provide strong evidence of demand for women candidates on behalf of local female party recruiters/association presidents.Footnote 2 If female aspirants in these districts recognize this demand, it is likely they will step forward in greater numbers. Thus, our second hypothesis:
H2:
Local party associations with a woman serving as president will be more likely to have a woman in the nomination contest.
Cheng and Tavits (Reference Cheng and Tavits2011: 469) note that other actors beyond the president likely play a role in encouraging women to step forward and seek candidacy but suggest that reliable data regarding the composition of the local executive more broadly would be “virtually impossible to obtain.” Recent data we have collected, however, provide us with a unique opportunity to explore the relationship between the gender composition of the local executive and those who seek political candidacy. Having women in other executive positions should theoretically serve the same or similar role as a local woman president: women on the executive should signal that the party is not hostile toward women and should therefore result in similar recruitment patterns. After all, the president does not act alone in governing the local association or engaging in potential candidate recruitment. More generally, women's political interest, efficacy and participation increase in districts where there are already women in politics or where there are women contesting elections (see, for example, Crowder-Meyer and Smith, Reference Crowder-Meyer and Smith2015). Seeing a variety of women on the local executive should spark a similar increase in engagement among potential female candidates. Thus, our third hypothesis:
H3:
Local party associations with a higher proportion of women on their executive board will be more likely to have a woman in the nomination contest.
We expect that the timing of the nomination should also matter in terms of who steps forward. Longer and earlier races likely signal to potential candidates that the nomination is indeed open for a competitive race and that the local association does not have a strongly preferred candidate. It is not uncommon for local party associations to orchestrate late or short nomination contests to ensure that their preferred local candidate goes unchallenged (Cross et al., Reference Cross, Kenig, Pruysers and Rahat2016; Cross Reference Cross2004). Members of a provincial (Manitoba) New Democratic party (NDP) characterized a recent nomination in the electoral district of Point Douglas as “sneaky” and “unethical” after the call for nominations was opened and closed in a matter of days with little warning or advertisement. Perhaps unsurprisingly, only a single candidate was able to complete the necessary paperwork to file for candidacy (Kavanagh, Reference Kavanagh2016). The absence of an obviously preferred candidate, as evidenced by an open and transparent contest, should therefore result in greater diversity in the candidate pool.
This, however, is not the only reason why the timing should matter. A typical explanation for women's lower rates of participation in party and electoral politics is resource disparity. This includes both socioeconomic resources such as time and money (Norris and Lovenduski, Reference Norris and Lovenduski1995) and political resources such as knowledge (Pruysers and Blais, Reference Pruysers and Blais2014), interest (Delli Carpini and Keeter, Reference Delli Carpini and Keeter1996) and efficacy (Burns et al., Reference Burns, Schlozman and Verba2001; Thomas, Reference Thomas2012). Given that women tend to have fewer of these resources, it should not be surprising that party members are significantly more likely to be men or that women candidates tend to have a shorter history with their party and attract significantly fewer volunteers.Footnote 3 As possible “outsiders,” women may need more time to mobilize support (that is, recruit new members) and assemble the necessary resources (financial and otherwise) to be successful candidates compared to men. The timing and length of the nomination campaign should therefore matter for who steps forward. In general, more accessible races should include more diversity.
Earlier nomination contests may also leave more time for women to organize their lives and arrange the necessary family and personal support systems to enable an entry into political life. This is especially relevant given that family and childcare obligations fall disproportionately on women (Milan et al., Reference Milan, Keown and Urquijo2011; see also Thomas and Bittner, Reference Thomas and Bittner2017). Thus, our fourth hypothesis:
H4:
Local party associations that hold longer and earlier nomination contests are more likely to have a woman in the contest than ridings where the nomination race is short and close to the general election.
Recent evidence demonstrates a clear gender gap in political ambition (Allen and Cutts, Reference Allen and Cutts2018; Fox and Lawless, Reference Fox and Lawless2005; Lawless and Fox, Reference Lawless and Fox2005; Pruysers and Blais, Reference Pruysers and Blais2017) and that women are more skeptical of party recruitment than men (Butler and Preece, Reference Butler and Preece2016). At least part of this gap is the result of competition/election aversion. When political office is more easily accessible, women tend to show similar levels of ambition as men (Kanthak and Woon, Reference Kanthak and Woon2015), whereas when the political office requires that they run against a competitive slate of candidates, women express less interest in running (Preece and Stoddard, Reference Preece and Stoddard2015). Competition and election aversion suggest that women may be more easily recruited to contest a local party nomination when they will face little intra-party competition. In other words, women may be more likely to contest the nomination in ridings where the local organization is not particularly strong or electorally successful. This would be consistent with recent evidence which suggests that women are nominated disproportionately in unwinnable or longshot districts and therefore fit the “sacrificial lambs” title (Thomas and Bodet, Reference Thomas and Bodet2013). Thus, our fifth hypothesis:
H5:
Local party associations located in constituencies where the party won the previous election or where the party association shows clear signs of health are less likely to have a woman in the contest than ridings where the party finished second or lower and where the party association is weaker.
Data and Case Selection
To address the question of why some nomination races within a party have women contestants while others do not, we draw upon data from a recent survey of constituency associations in Canada. Local association presidents completed the survey and provided considerable data on both the composition and political activities of their association.Footnote 4 Importantly, for our purposes, this includes information regarding the makeup of their local party executive (that is, gender of local association president and other officers), candidate recruitment efforts (that is, whether there was a formal candidate search) and local party health and vitality (that is, membership size and number of meetings). We integrated these data with administrative data gathered from Elections Canada reports. First, we collected data regarding the candidate nomination contest itself: specifically, the length of the contest (in days) and how long before the general election the nomination occurred (in days). Second, from these filings, we coded as our outcome variable whether a woman contested the nomination.Footnote 5 These two sources of data were then combined into a single dataset that includes our dependent variable (whether there was a woman in the nomination contest) and our key independent variables (local recruitment efforts, composition of local executive, campaign dynamics and the health and vibrancy of the local associationFootnote 6), as well as a variety of control variables (party, urban/rural district location,Footnote 7 and the competitiveness of the party in the district).
The constituency association survey was conducted online by the authors in February of 2016, only a few months after the October 2015 Canadian general election. Surveys were sent to all local associations for which valid contact information was available. The final dataset includes 255 valid responses, for a response rate of 24 per cent. Response rates, of course, varied by party: Liberals (27.6%), Conservatives (14.1%), New Democrats (28%) and Greens (27.2%).Footnote 8 The 24 per cent response rate is consistent with comparable surveys in the field. Importantly, the survey allowed us to collect data that have never before been collected, and therefore we are able to address questions that have never been asked. For purposes of this analysis, we remove associations with an incumbent seeking renomination, as these are typically “closed” nominations, rarely resulting in any challenge to the sitting member of Parliament. This leaves us with a total of 179 cases for analysis.
Why Canada? We chose Canada to study the relationship between local dynamics and the gender composition of nomination contestants for a number of reasons. Consistent with the single member plurality electoral system (SMP), Canadian parties nominate a candidate in each of the country's 338 electoral districts. This is important for at least two reasons. First, given that each party is nominating hundreds of candidates, this provides us with numerous cases to analyze. Second, candidate selection is devolved to constituency associations and therefore is done, for the most part, at the local level (Carty and Eagles, Reference Carty and Eagles2005; Sayers, Reference Sayers1999; Cross, Reference Cross, Everitt and O'Neill2002, Reference Cross, Pammett and Dornan2006; Cross and Pruysers, Reference Cross and Pruysers2017; see also Pruysers and Cross, Reference Pruysers and Cross2016). Third, and relatedly, there should be lots of variation in terms of local circumstances. Because of the stratarchical bargain (Carty, Reference Carty2002, Reference Carty2004; Carty and Cross, Reference Carty and Cross2006; Pruysers, Reference Pruysers2015), local associations have considerable freedom in how they operate and organize. The local nature of candidate selection in Canada, coupled with relative autonomy for party associations to operate with a local flavour, allows us to explore how grassroots contextual and organizational differences influence the composition of the race.
Context: Women and the 2015 Federal Election
The 2015 Canadian election saw more women candidates, and subsequently more women members of Parliament, than any other federal election in the country's history (though much of this can be attributed to the growing size of the House of Commons). A total of 533 women were nominated across the various parties and 88 were ultimately elected to Parliament (O'Neill and Thomas, Reference O'Neill, Thomas, Pammett and Dornan2015). Those women elected to the House of Commons include 50 Liberals, 18 New Democrats, 17 Conservatives, two Bloc, and one Green. While the 2015 election saw the most ever women candidates and elected MPs, our data from the nomination stage demonstrate a striking, though unsurprising, gender gap in terms of who steps forward to contest a party's nomination. A total of 496 women sought a nomination in one of the three principal parties. This group included 219 New Democrats, 178 Liberals and 99 Conservatives. An additional 144 women sought nomination in the Green party. Many more men sought a nomination in all of these parties, as women constituted a minority in each party: 44 per cent of NDP nomination contestants, 39 per cent of Greens, 31 per cent of Liberals and 22 per cent of Conservatives (see Figure 1).
Figure 1 Nomination Contestant Sex by Political Party (2015 Election)
Turning our attention from all nominations to open nominations (a nomination where an incumbent was not seeking re-election), we find that a striking number of nomination races did not have a single woman contestant (54.4%). Again, we find important inter-party differences in this regard: 69 per cent of open Conservative nominations did not have any women in the race compared to 59 per cent of Greens, 53 per cent of Liberal nominations and 39 per cent of New Democratic Party nominations. In fact, our data reveal that there were some 15 ridings in which none of the three major parties had a single woman contesting the nomination. What is more, once the internal party nominations were complete, there were nearly 100 ridings across the country where the Liberals, Conservatives and New Democrats had not nominated a single woman to contest the general election, resulting in a local election between three men.Footnote 9
While women are indeed underrepresented in the nomination candidate pools, the evidence suggests they do not fare significantly worse than their male counterparts in these contests. In each party, the share of nominated women candidates almost identically matches the share of women in the nomination pool: 43 per cent of NDP general election candidates were women, as were 40 per cent of Greens, 31 per cent of Liberals and 20 per cent of Conservatives. While women might continue to be disproportionately nominated in districts that a party has little chance to win (Thomas and Bodet, Reference Thomas and Bodet2013), the evidence does suggest that nomination selectorates are not dramatically biased against women candidatesFootnote 10 and that at least part of the challenge for those seeking more equitable representation in legislatures is to encourage more women to seek party nominations (see Table 1). Thus, the presence of a woman candidate in the nomination race is our dependent variable.
Table 1. Women Nomination Contestants, Nominated Candidates and MPs (2015 General Election)
Results
We begin with a brief review of the bivariate findings concerning the relationship between our independent and dependent variables (Table 2). While all of the parties’ nomination rules and guidelines reference local party candidate search committees, we find that a considerable number of associations without an incumbent MP did not conduct a formal candidate search. This is true for all four parties, and as shown in Table 2, for almost one-third of local associations overall.Footnote 11 What is more, these formal searches are often linked, at least in principle, with specific diversity goals. Section 1.7 (a) of the 2013 Liberal nomination procedures, for example, asserts that no nomination meeting shall be issued until “the EDA has demonstrated to the satisfaction of the Provincial or Territorial Campaign Chair that the association has conducted an acceptable search for Nomination Contestants, including documented evidence of a thorough search for potential candidates who are female and who are reflective of the demographic and linguistic makeup of the local electorate” (Liberal Party, 2013). Even in the NDP, the party with the strongest requirements and apparent commitment to formal searches, slightly more than one-in-ten local party associations report that they did not have a formal search. This may be disconcerting, as local associations with a formal search committee were more likely to have a woman seek their nomination.
Table 2. Bivariate Relationship between Presence of a Woman Nomination Candidate and Characteristics of the Nomination Contest
There also appears to be a relationship between the presence of women in decision-making positions in the local party and the likelihood of a woman seeking the nomination. While only 27 per cent of our associations report having a woman president, those with a woman in the top position were more likely to attract a woman nomination candidate. Similar results are found regarding the presence of women in other local executive positions. The mean representation in our sample of women on local executives is 41 per cent. When we consider associations with a majority of these positions filled by women, we find that women are approximately 12 per cent more likely to have sought the nomination.
In terms then of our first three hypotheses, the bivariate relationships suggest that the actions of the local association in terms of affirmatively recruiting candidates and in placing women in positions of leadership do lead to a greater likelihood of having a woman seek the party nomination.
Regarding the timing of the nomination contest, the bivariate analyses again support our hypothesis. Perhaps as a result of fixed election dates, parties have been nominating their candidates earlier than in the past. Approximately one-third of candidates were nominated in 2014 for the October 2015 election. At the same time, these campaigns appear to be longer in duration than was traditionally the case. The mean duration in our sample is approximately four months (121 days). Both of these variables seem to have an effect on the likelihood of a woman seeking the nomination. A woman was present in the nomination contest in 54 per cent of the 2014 contests, compared with 44 per cent of those held in the election year (2015). Similarly, contests with a woman contender lasted 137 days on average, compared with 107 days for those with only male candidates.
The bivariate findings provide mixed evidence regarding the relationship between the likelihood of the presence of a woman nomination contestant and the vitality of the local party association. Associations with and without a woman contestant both held approximately seven meetings in the year prior to the election. However, women were more likely to contest nominations in associations that saw considerable membership mobilization in the year of the nomination. The later contests saw twice the growth in membership numbers than did those with no women contestants. This may reflect the fact that women nomination candidates are more likely to be challenged when seeking a nomination and thus the need to mobilize support (Cross, Reference Cross2016a).Footnote 12
The remaining question is how all of these factors fit together. We move now to a more robust multivariate analysis. Table 3 provides the logistic regression results. Contrary to H1, we find no evidence that engaging in a formal candidate search results in more gender diversity among participants. Given that formal searches are typically tied to diversity goals, we expected associations that held a formal search to be more likely to have a woman in the contest. Table 3, however, reveals that associations that report holding a formal candidate search are no more likely, when controlling for other factors, to have a woman contestant than those associations that did not conduct a formal search.
Table 3. Local Determinants of a Nomination Race including a Woman Contestant
Note: Liberal party is the reference category for “party.” Excludes all incumbents from the analysis.
* p < 0.05
** p < 0.01
Consistent with Cheng and Tavits (Reference Cheng and Tavits2011) and Pruysers et al. (Reference Pruysers, Cross, Gauja, Rahat, Scarrow, Webb and Poguntke2017), we provide further evidence that the gender of the local-party gatekeeper matters. When women are in positions of party authority—in this case, as local party association presidents—the chances of having a woman in the nomination race increase significantly. The odds ratio of 3.5 indicates that the likelihood is more than three times higher when the local president is a woman, compared to when the association president is a man. We can therefore confirm H2 and the importance of having women at the top of the local party executive.
Interestingly, however, our multivariate analysis finds no evidence that the makeup of the broader local party executive matters in this regard (H3). While we measure this as a continuous variable in Table 3, we also explored the possibility that there could be a threshold effect (allowing for a higher number of critical actors; see Childs and Krook, Reference Childs and Krook2008) wherein women are more likely to contest a nomination when the local party association has a certain number (percentage) of women on the executive. However, even when exploring a variety of thresholds (25%, 33%, and so forth), we find no meaningful effect for the composition of the executive as a whole. The analysis suggests that the president is the driving force. Nonetheless, it is important to bear in mind that local presidents often come through the party ranks, serving in lower executive positions before assuming the presidency. EDAs with a woman president report significantly greater presence of women on their executives than do those with a male president (48.9% compared to 38.5%). Thus the fact that only one-in-four presidents is a woman likely reflects the gender gap found further down the executive and in EDA memberships generally.Footnote 13
We also find that some of the more “administrative” aspects of the contest matter, supporting H4. In general, it appears that the more accessible the contest is, the more likely there will be a woman in the race. Specifically, the longer the nomination takes place before the general election, and the longer the nomination contest lasts, the more likely a woman is to compete. We suspect that a longer campaign window allows women, who have been historically marginalized and excluded from party politics, with the time needed to mobilize support and mount a competitive campaign. Longer and earlier contests may also provide nontraditional contestants with confidence that there is not a preferred candidate. There are many reports, in each election, of parties delaying and holding very short nomination contests to allegedly ensure the nomination of a favoured candidate (see Cross et al., Reference Cross, Kenig, Pruysers and Rahat2016).
In terms of H5, we find that women are indeed strategic when choosing where to run, just not entirely as we expected. While the literature suggests that women win nominations disproportionally in ridings where the party did poorly in the previous election (Thomas and Bodet, Reference Thomas and Bodet2013), our results reveal a more complicated story in terms of entry into the race. When there is no incumbent seeking re-election, nominations in ridings where the party won the previous election are far more likely to have a woman in the race, compared to nominations in ridings where the party lost the previous election. The effect is both significant and substantive. Such races, however, are numerically limited, as incumbents rarely leave office before defeat. Moreover, we find no significant relationship between overall party competitiveness in the district and women's entry into the contest. The effect of competitive party dynamics therefore appears to be limited to those rare cases where the party holds the seat but has no incumbent. Likewise, healthy and vibrant associations, as measured by membership growth and local activity (number of association meetings), do not have the same influence as previous electoral victory. Contrary to H5, then, while women are more likely to contest nominations where their party won the previous election, they are no less likely to seek the nomination in ridings where the party is showing concrete signs of grassroots vitality.
Finally, while they serve as controls in our analysis, it is interesting to note that we find no significant differences between urban and rural electoral districts, and no major party differences either. Once removing incumbents and controlling for a variety of local factors, for example, the Conservatives are just as likely to have a woman contesting a local nomination as are the Liberals.
Discussion and Conclusion
Women remain underrepresented in national legislatures not just in Canada but around the world. Despite a growing body of literature that seeks to uncover the numerous causes and potential solutions to this underrepresentation, the local activities of political parties have largely been left unexplored. We address this limitation in the current literature with data from an original survey of Canadian local party associations, as well as with extensive nomination data gathered from Elections Canada. While research exploring the factors that are related to women winning a party nomination is indeed important, women can only win a nomination if they are in the contest. A further original contribution of this article is, therefore, our use of nomination contestants, as opposed to successfully nominated candidates, as our dependent variable. Our results reveal that what parties do at the local level matters for representational outcomes. We highlight a number of practical solutions for parties seeking to have more diversity in their candidate pool. Although parties can implement quotas, develop informal targets and construct progressive and favourable candidate selection rules, and while these types of reforms have been shown to increase the number of women in the party, we demonstrate that there are other solutions available to parties as well: diversifying other positions within the party, and adjusting the timing and length of the nomination race.
Building on the work of Cheng and Tavits (Reference Cheng and Tavits2011), Tremblay and Pelletier (Reference Tremblay and Pelletier2001), and Cross (Reference Cross2016a), we show that the gender of the local gatekeeper matters. In particular, women are more likely to enter a nomination race when the local association is led by a woman. While party association presidents may not be personally responsible for candidate recruitment in every case (though an overwhelming majority report that they are involved), they can informally encourage women to enter the race, as well as signal to potential candidates that they would be welcome within the party generally and the local association specifically. Our results are also consistent with recent experimental evidence suggesting that women respond less positively to potential party recruitment when the recruiter is a man (Pruysers and Blais, Reference Pruysers and Blais2018a) and with evidence from the United States suggesting that women party officials are more likely to recruit women than are male party officials (Crowder-Meyer, Reference Crowder-Meyer2013). The results therefore suggest that parties seeking to diversify their slate of candidates would do well to diversify other positions in their party.
Our results suggest that the administration of the nomination also matters in terms of who steps forward. Indeed, an earlier nomination date and a longer campaign both appear to encourage women to enter the race. This is likely due to several factors. An earlier nomination gives potential candidates more time to emerge and prepare. As women continue to do more in terms of childcare and unpaid family responsibilities, an earlier nomination provides more time to prepare for the general election. Likewise, a longer campaign allows “outsider”candidates time to recruit new members and mobilize support. Related to this, longer nomination contests provide the appearance that the party is open to a contest and that there is not a “preferred candidate.” This, of course, suggests that there are minor administrative changes that parties can make to the nomination process that will diversify the candidate pool. Importantly, these kinds of changes do not substantively change the process. Such changes, however, are somewhat complicated due to the relationship between the grassroots and the parties’ centre. While perhaps seemingly mundane, these would require greater central/local cooperation and coordination, as both levels of the party have key roles to play in the candidate selection process (Cross, Reference Cross2016b; Pruysers and Cross, Reference Pruysers and Cross2016).
Overall the results point to a number of ways in which parties can make relatively minor changes to enhance the inclusiveness of the nomination process. Though this is an important contribution, there are two limitations and avenues for future research that are worth briefly discussing. First, our data rely on a sample of constituency associations rather than the entire population. While our response rate is consistent with that found in the literature, it does raise potential concerns of generalizability. As always, replication is important, and we encourage scholars not only in Canada but in other countries to conduct similar analyses focusing on the interaction between party politics at the grassroots and gendered intra-party nomination dynamics. In particular, more research is needed that explores the nomination contest and dynamics as we do, and not solely the final outcome. Second, our data do not allow us to consider differences between different groups of women. This is an important limitation, as recent evidence suggests that the experience of women in politics is not universal and that women of colour are often discouraged from running for political office (Carroll and Sanbonmatsu, Reference Carroll and Sanbonmatsu2013). An important challenge for future research is therefore to collect data that allows for an intersectional approach to be built into the research design and analysis (but see Tolley, Reference Tolley2017).
