Worldwide, women are typically less likely than men to serve as chief political executives, that is, as presidents, premiers, and prime ministers (Jalalzai, Reference Jalalzai2013). This pattern also holds in Canada; despite recent parity cabinets in the federal and some provincial governments, women typically comprise less than one-third of cabinet ministers at any level of government (Trimble et al., Reference Trimble, Arscott, Tremblay, Trimble, Arscott and Tremblay2014), and between 2002 and 2008, no woman served as premier or prime minister at the provincial or federal level. Similarly, women in municipal politics are less likely to be mayors than are men (Tolley, Reference Tolley2011).
Given this, it is noteworthy how, starting in 2010, a considerable number of women were selected to serve as provincial premiers. Beginning in 1991, eight women have served in this office; of them, six were made premier first through an inter-election party-based appointment while their parties held government. During the same period, 50 men were selected premier, and a majority of them were made premier first through general election. What explains this gendered pattern?
One explanation is that women are more likely than men to be selected to lead parties in government that are in crisis or decline. This would be in keeping with what is known about women and political party leadership, both in Canada and in the comparative literature (Beckwith, Reference Beckwith2015; Cross and Blais, Reference Cross and Blais2012; O'Neill and Stewart, Reference O'Neill and Stewart2009). It also fits with the fact that several of the women premiers exited the office relatively quickly after they entered it. In addition to this, to date, no political party led by a woman in Canada has decisively won more than one election. Thus, we are presented with an interesting pattern: women are selected directly to the premier's office through a party leadership contest and sometimes lead their parties to majority mandates but then do not have the support to stay in office for an entire term or to be re-elected. The result is that, by the end 2017, only two woman premiers hold their own electoral mandates (Wynne and Notley), and neither is expected to be easily re-elected.
This pattern of party-based selection and quick exits raises a host of research questions about women's entrances and exits as premiers, their leadership styles and influence on policy, as well as media, public and internal party reactions to their presence. This paper contributes to a larger research agenda by examining one small facet of gender and the political executive in Canada: the contexts through which women are first selected to the premier's office. It uses the most common explanation found in the literature for women's emergence as leaders of electorally competitive parties as a starting point: that the parties are in crisis or in decline (Beckwith, Reference Beckwith2015; O'Brien, Reference O'Brien2015).Footnote 1 Results suggest that contrary to expectations, evidence supports this explanation for three of the eight women premiers, and that a fourth is explained by existing literature on gender and party leadership. However, for half of the women who have served as provincial premiers, evidence of their party's decline is at best partial or inconclusive. This highlights gender and the premier's office as a fruitful avenue for future research.
Why Premiers?
Premiers are powerful actors in the Canadian political system. Provinces are responsible for key programmes associated with the welfare state, including health care, education, welfare and child care, as well as all municipalities and major metropolitan centres. Provincial governments in Canada have a strong presence, considerable jurisdictional control and fiscal capacity, and the ability to design and deliver a large number of government services (O'Neill and Stewart, Reference O'Neill and Stewart2009; Simeon, Reference Simeon2002).
In addition to these powers, provincial legislatures in Canada, like their federal counterpart, are “not [systems] of checks and balances, but of fused, concentrated, centralized power,” where power rests in the political executive rather than the legislature (Franks, Reference Franks1987: 265; Russell, Reference Russell2008). At the federal level, this is characterized as “prime ministerial government,” where the chief executive dominates cabinet (Savoie, Reference Savoie1999, Reference Savoie2003). This pattern is replicated and magnified at the provincial level. Comparatively speaking, “even by Westminster standards, Canadian first ministers wield exceptional power,” due, in part, to the absence of institutional constraints within legislatures that are found in other cabinet-parliamentary systems (White, Reference White2005: 77). This leads Graham White to conclude that “Provincial government is indeed premiers’ government” (77). Add to this the electoral significance of party leaders for vote choice in Canada (Gidengil et al., Reference Gidengil, Nevitte, Blais, Everitt and Fournier2012; see also Poguntke and Webb, Reference Poguntke and Webb2005; Webb and Poguntke, Reference Webb and Poguntke2013), and it is clear that provincial premiers are some of the most important actors in Canadian politics.
Expanding this analysis to provinces adds useful information for gender and politics research. To date, 12 women have served as chief political executives in Canada and the overwhelming majority (N = 11) are found at the subnational level. This would be missed if, like most studies of chief executive offices, we focused solely on the national level (see, for example, Jalalzai, Reference Jalalzai2008, 2011; O'Brien, Reference O'Brien2015). This study restricts the analysis to women in provincial premier's offices, given the constitutional differences between the provinces and the territories.Footnote 2 Studying women in the provincial premier's office represents one of the ways, if not the best, to assess women's access to chief political executive positions in Canada.
Who are the Women in the Provincial Premier's Office?
Table 1 shows the women who have served as provincial premiers in Canadian history, presented in the order in which they were sworn into office. The first woman provincial premier, Rita Johnston, was clearly a sacrificial lamb, as will be shown below, as her party was in a crisis when she was selected leader. The competitive position of the second, Catherine Callbeck of Prince Edward Island (PEI), appears very different: Callbeck was selected leader of a party in government, won a majority mandate in a subsequent general election and then resigned before the end of that term.
Table 1 Women Premiers in the Canadian Provinces

1See Courtney (Reference Courtney1995) for more information about the primary/convention leadership selection mechanism.
The bulk of women provincial premiers have been selected more recently. Of the six selected since 2010, four were first selected premier through internal party mechanisms, and three went on to lead their parties to majority electoral victories. Both Marois and Notley were first selected premier through general elections. Though those more recently selected premiers appear to be more stable and powerful than Johnston, none has yet to secure more than one general election victory. Dunderdale and Redford resigned in early 2014, while Marois led her party to its worst electoral defeat since the late 1970s. Clark technically led her party to re-election, but the election returned a minority legislature, and she did not secure the confidence of the legislative assembly.Footnote 3
Women, Party Leadership, and the Premier's Office
Table 1 shows that six of the eight, or 75 per cent of the women who have served as a provincial premier were first selected directly to the office through internal party mechanisms. This is striking for two reasons. First, this pattern is different from that of men who have been premiers. Excluding interim party leaders, 50 men have served as premiers in Canada since 1991.Footnote 4 Of these, 56 per cent were first selected provincial premier through general elections. This suggests the path to the premier's office in Canada is gendered: men who become provincial premier are most likely to do so through general election, but women who become provincial premier are, to date, most likely to do so through internal party mechanisms.
Second, comparative evidence shows that men's path to the provincial premier's office is the international norm. Typically, political executives in parliamentary democracies are appointed, at least for the first time, after a general election, rather than internal party processes between elections. One of the most comprehensive studies of party leadership selection in parliamentary democracies does not address when and why individuals may be selected for the dual role of party leader and chief political executive at the same time (see Cross and Blais, Reference Cross and Blais2012). Certainly, gendered analyses of women prime ministers highlight the importance of first being selected party leader, but even in these instances, the women in question (Angela Merkel, Margaret Thatcher) are first selected as party leader and then subsequently become prime minister through a general election (Beckwith, Reference Beckwith2015). This suggests that women's paths to the premier's office in Canada are different from men's and also, potentially, from the comparative norm.
Given this, it is perhaps unsurprising that even less is known about the gendered aspects of selection to the dual role of party leader and chief executive. Most research investigating gender and political careers focuses on women's presence and participation in legislatures (Childs and Krook, Reference Childs and Krook2008; Kathlene, Reference Kathlene1994; Trimble and Arscott, Reference Trimble and Arscott2003). Studies of women in provincial cabinets in Canada suggest that the most important factor leading to the appointment of women to cabinet is the proportion of women in the governing party (Studlar and Moncrief, Reference Studlar and Moncrief1997; but see also O'Brien et al., Reference O'Brien, Mendez, Peterson and Shin2015). Comparative research shows that women are more likely to serve as chief political executives when power is shared or constrained (Jalalzai, Reference Jalalzai2008, Reference Jalalzai2013), and that female-led governments do not necessarily promote women into cabinet (O'Brien et al., Reference O'Brien, Mendez, Peterson and Shin2015). And studies conclude that gender stereotypes have negative effects on women's ability to be directly elected to executive office (Huddy and Terkildsen, Reference Huddy and Terkildsen1993; see also Fox and Oxley, Reference Fox and Oxley2003; Streb et al., Reference Streb, Burrell, Frederick and Genovese2008).
Women's selection to the dual role of party leader and premier is also atypical because few party leaders are women in the first place. Between 1965 and 2008, only 8 per cent of newly elected party leaders in parliamentary democracies were women (Cross and Blais, Reference Cross and Blais2012). Women's leadership numbers are low because competitive parties are less likely to select women leaders (O'Brien, Reference O'Brien2015). O'Neill and Stewart (Reference O'Neill and Stewart2009) note that in Canada between 1980 and 2005, 67 per cent of male leaders, but only 12 per cent of female leaders, headed major, electorally competitive parties. By 2005, “only one of the 21 women elected to lead a [provincial] political party in Canada has gone on to win an election” (O'Neill and Stewart, Reference O'Neill and Stewart2009: 749). This suggests that the legislative and electoral contexts faced by women in general, and female party leaders in particular, are hostile. Other Canadian scholars note that women lead “electorally decimated and moribund parties,” (Trimble and Arscott, Reference Trimble and Arscott2003: 77). Prior to 2010, only Catherine Callbeck had been selected leader for an electorally competitive party at the provincial level (Bashevkin, Reference Bashevkin2009); it was not until 2011 that a woman was selected party leader through a universal vote of party members (see Cross and Blais, Reference Cross and Blais2012).
This literature suggests, then, that women's paths to the premier's office are different from their male peers in Canada as well in other parliamentary democracies.
When Might Women Be Selected As Party Leaders and/or Provincial Premiers?
Research shows that the higher the political office, the fewer women will be found in them, especially when those offices are safe, secure, and electorally viable (Bashevkin, Reference Bashevkin1993, Reference Bashevkin2009). This suggests that the most likely explanation for women's selection as party leader/premier rests with party crisis and decline. Comparative research shows that parties that are losing support or are in crisis are more likely to select women leaders than are parties with stable support (Beckwith, Reference Beckwith2015; O'Brien, Reference O'Brien2015).
Crisis and decline create strategic contexts that discourage men who might otherwise contest party leadership. Women thus become competitive when they otherwise might not be. As strategic actors, women may see crisis or decline as producing a moment where they might be able to secure a leadership position that they may otherwise not have access to (see Beckwith, Reference Beckwith2015). A similar argument posits that women might be preferred over men as leaders when the crisis or decline was the result of men's leadership failures (Bruckmüller and Branscombe, Reference Bruckmüller and Branscombe2010). Women may be stereotyped as more moral or less corruptible than men (Fridkin Kahn, Reference Fridkin Khan1994; Huddy and Capelos, Reference Huddy, Capelos and Ottai2002), making them potentially more appealing to the party selectorate and, in theory, to the electorate overall.
What does it mean for a party to be in crisis or decline? Crisis can be defined, at least in part, by electoral loss (Beckwith, Reference Beckwith2015). However, losing an election alone does not mean a party is in crisis. Crisis implies an acute, intense, time-sensitive period of difficulty for a party or a government. Angela Merkel could be characterized as strategically using crisis to secure her party's leadership. After her party, the Christian Democratic Union (CDU), decisively lost the 1998 elections, the party became embroiled in a corruption scandal. These two factors ensured that men who would normally have contested the leadership of the CDU were removed from serious consideration, creating space for Angela Merkel to successfully seek leadership of the party (Beckwith, Reference Beckwith2015). Thus, crisis can emerge from significant electoral loss, major scandal or a serious split within a party.
Decline, by contrast, is perhaps best characterized as a more gradual loss of support. A party, or a party in government that is in decline, is on an unfavourable electoral trajectory and may lose both votes and seats in the election preceding the leadership contest (O'Brien, Reference O'Brien2015; O'Neill and Stewart, Reference O'Neill and Stewart2009). The British Conservatives’ experience leading to Margaret Thatcher's selection as party leader might be best characterized as decline. Edward Heath had been party leader since 1965. The party had lost two elections in 1974, creating an appetite for a challenge to Edward Heath's leadership. This may have been, at least in part, due to the perception that Heath's leadership was tired or unable to respond effectively to the party's perceived decline. Importantly for Thatcher, some contenders were removed from consideration, either by earlier scandal or by not coming forward to challenge Heath's leadership before Thatcher did. Other men only entered the contest after Thatcher's challenge removed Heath. Despite their belated entrance, Thatcher successfully secured a majority of votes on the second ballot (Beckwith, Reference Beckwith2015).Footnote 5
Measuring a party's gradual decline is far more challenging than identifying when a party is in acute crisis. It will be clear, for example, if a party spectacularly loses an election (and does not recover), or is embroiled in a serious scandal. By contrast, the best way to measure decline—the slow(er) loss of strength and support—is far less clear. Given this and the fact that 75 per cent of women enter the premier's office through an internal party process rather than a general election, the bulk of the empirical tests that follow focus on whether their parties were in decline when women were selected as leader/premier.
Measuring Decline
As noted above, a party in decline is one that has an unfavourable electoral trajectory. Unlike a party in crisis, parties in decline could experience attrition in their support gradually or present weaknesses that are seen as harbingers of expected electoral defeat. Given this, accurately measuring the stability of a party's support is especially important.
In this article, decline is operationalized in five ways. For each, decline exists if a party's support appreciably drops; in the absence of clear evidence of that drop, the party's support could be characterized as stable. In order to avoid ex post facto logic, evidence must show that this decline precedes each woman's selection as party leader/premier; decline that occurs after women's selection cannot be contributing factor in their selection as leader/premier.
First, the party's performance in the election immediately prior to a woman's selection as leader/premier is assessed. Parties in decline lose votes, seats or both. This measurement of decline is available for every provincial party that selected a woman as leader/premier, and is presented in Table 2.
Table 2 Change in votes and seats for each woman selected party leader and premier simultaneously

Note: Cells report the change in vote share (percentage points) with election year in parentheses. Change in number of party seats is the change in the absolute number of seats held by the party from one election to the next. For each, negative numbers indicate losses; positive numbers indicate gains (author's calculations).
The second measure is party support stability. This is a rigorous measure that combines information from two past elections to assess how stable a party's support is over time, making it an ideal measure to diagnose whether a party is in decline. Its application at the federal level between 2004 and 2008 identified considerable declines in Liberal party support well in advance of the party losing seats it had once identified as safe (Bodet, Reference Bodet2013; Thomas and Bodet, Reference Thomas and Bodet2013). The difficulty with this measure is that it can only be generated when district boundaries are stable across time and/or elections are held using single member districts. This is problematic for BC and Newfoundland and LabradorFootnote 6 as well as PEI.Footnote 7 Given this, the strongholds/battlegrounds measure is calculated for Alberta in 2008 (prior to Redford's selection as party leader/premier) and Ontario in 2011 (before Wynne's selection as party leader/premier, see Table 3).
Table 3 Strongholds and Battlegrounds in Alberta and Ontario

Note: Cells report the N of seats. Bold cells report the party that went on to select a female leader/premier.
Because general elections occur in the provinces every four to five years, a party's support at the previous election might not identify erosion in its support, particularly if the party remains in government. In light of this, by-election results and contributions to party finances are used as the third and fourth indicators respectively. By-election results may highlight how voters feel about a party in government overall (see Loewen and Bastien, Reference Loewen and Bastien2010); for example, if a party in government loses by-elections in electoral districts it would typically win, this might suggest the party is in decline. All by-elections held between the previous election and the leadership selection for each woman premier are included here (Table 4). Similarly, though most of the research on campaign and party finance in Canada focuses on federal politics and parties (see Coletto et al., Reference Coletto, Jansen and Young2011; Cross, Reference Cross2004; Scarrow, Reference Scarrow2007), Canada's provincial political parties also finance themselves through contributions from individuals and businesses. This suggests that parties in decline should report lower levels of contributions in the years prior to the selection of a new leader. Contributions data are publicly available for parties selecting a woman as leader/premier since 2010, and are presented in Table 5.
Table 4 By-Election Results for Parties that Selected a Woman Leader/Premier

Sources: Elections PEI; Newfoundland and Labrador; Elections BC; Elections Alberta; Elections Ontario
Table 5 Total Contributions for Parties that Selected a Woman Leader/Premier

The fifth and final way decline is measured is through publicly available opinion polls.Footnote 8 Past election results could underplay the speed with which a party in government loses support. Given this, public opinion polls are a logical option to capture any shifts in a party's support that occur after a general election, but before a leadership contest. The veracity and quality of these polls is secondary, as these polls were published and formed the narrative around each political party and their selection of a woman as their leader/premier.Footnote 9 If polls informed the perception that the party was in decline when a woman was selected as party leader/premier, this could be taken as evidence that a woman was selected leader and premier of a party in decline (Figures 1–4).

Figure 1 Public Opinion in Newfoundland and Labrador
Analysis
Parties in crisis
Of the eight women listed in Table 1, two were selected to lead parties in crisis: Rita Johnston and Pauline Marois. Each became party leader during an acute period of difficulty for her party. In 1991, Bill Vander Zalm resigned as BC premier and Social Credit leader after he was charged with criminal breach of trust relating to the sale of some of his property (CBC Digital Archives, 2014; Paikin, Reference Paikin2003). This scandal produced an acute crisis for the BC Social Credit party. Johnston faced no male challengers and narrowly beat another woman (Grace McCarthy) for the party leadership (McElroy, Reference McElroy2017). She was in office for only seven months before her party suffered a devastating electoral defeat from which it has never recovered (Table 1; Elections BC, 2014b). Because of the magnitude of this loss, Johnston is perhaps best described as a sacrificial lamb for a party in crisis.
Similarly, the PQ was in crisis when it selected Pauline Marois to be its leader, though for different reasons. Marois’ successful bid for the PQ leadership built on its 2007 electoral loss.Footnote 10 This election marked the third in a row where the PQ lost support. In 1998, the party won government but lost the popular vote. In 2003, the party lost government and nearly 10 percentage points in popular support. By 2007, the party lost five more points, securing less than 30 per cent of the popular vote and finishing third for the first time since forming government in 1976. The scope and protracted nature of this loss created the crisis that preceded Marois' selection as leader. In keeping with the comparative literature, Marois was not challenged by competitive men, ran unopposed and was acclaimed leader.Footnote 11
Though Marois and Johnston fit with the comparative literature on gender and party leadership, they remain distinctive among women premiers in Canada. Most were selected leader and, by extension, provincial premier partway through an existing electoral mandate in the absence of an obvious scandal or crisis. The question remains whether these women were selected to lead parties in decline.
Which parties are in decline?
Data exist to test the assertion that women are more likely to be selected party leader/premier when their party is in decline for four women in the premiers’ offices: Kathy Dunderdale, Christy Clark, Alison Redford and Kathleen Wynne. Where possible, data are also reported for Catherine Callbeck.Footnote 12
Table 2 shows the change in the change in party vote and number of seats in the election leading up to and following a woman's simultaneous selection as party leader and premier.Footnote 13
Most parties do not appear to be in decline the election prior to selecting a woman as leader and premier simultaneously. Instead, support for most of these parties appears to be stable or increasing. The one exception to this appears to be Kathleen Wynne in Ontario. Though the Ontario Liberal party's support slipped by less than 5 percentage points in 2011 over 2007, the party lost 18 seats.
What is perhaps more telling is that nearly every party that simultaneously selected a woman as party leader and premier lost votes in the election following her selection. Some of these losses are devastating, as is the case with Johnston, while others appear to be moderate, as is the case with Dunderdale and Redford. What remains unknown from Table 2 is whether the selection of a woman as party leader/premier was the result of a shorter-term decline or the result of the woman's failures while in office. Still, Table 2 shows that most of these parties were not in decline the election before they selected a woman leader/premier.Footnote 14
Table 3 presents results from the second measure of party decline: party support stability. This measure uses information from the previous two elections to assess how many strongholds a party had. A party in a long, slow decline may be able to win a majority mandate even while its support is eroding. This measure speaks directly to how confident a party in government could be about retaining power at the next election, based on the long-term stability of its support.
Prior to Redford's selection as party leader, her party had 54 strongholds. This means that, based on the past two elections, the Alberta PCs could expect that their party would win 65 per cent of the seats available in the legislature. A party in decline could not have so many strongholds, suggesting that the Alberta PCs were not in long(er) term decline when Redford was selected as leader.
Table 3 suggests the Ontario Liberals were less stable than the Alberta PCs when Wynne was selected leader. The Liberals held 47 strongholds, representing 44 per cent of the Ontario legislature. Though this represents the most strongholds of any party at the time, the Liberals still needed to win 7 of the 23 battlegrounds to form a majority government. This is a more tenuous position for an incumbent party in government and, when coupled with the losses in the election prior to Wynne's selection, it suggests the Liberals may have been experiencing longer-term decline before Wynne's selection.
More time-sensitive measures show if parties in government experience a shorter-term decline. Table 4 shows the number of by-elections held between the previous election and the year each woman was selected as party leader/premier. One challenge is that in most provinces, only one or two by-elections are held, making it difficult to identify a pattern. The Alberta PCs lost one by-election to the opposition Wildrose party prior to Redford's selection as PC leader. A post-hoc rationale may place a great deal of importance on a single by-election loss; placed appropriately in time order, though, this is weak evidence of a party in decline. Similar time-order issues exist for the Ontario Liberals: though they did not lose seats in by-elections prior to Wynne's selection in February 2013, by September, the Liberals had lost three seats to opposition parties. This fact is more suggestive of decline and matches the results found in Table 3. Newfoundland and Labrador's by-elections suggest that the PCs had strong support prior to Dunderdale's selection.
Table 5 outlines financial contributions made to political parties that selected women as their leaders while they held government and shows that none of these parties experienced consistent year-over-year declines in contributions in the three years prior to selecting women as leaders/premiers. Instead, the Newfoundland and Labrador PCs increased their contributions year over year prior to Dunderdale's selection. The other parties show some decline in donations, though these are not steady or linear for any of the parties. Contributions to the BC Liberals increased by 52 per cent two years before Clark's selection, but then decreased by 23 and 3 per cent respectively. Similarly, contributions to the Alberta PCs increased in the two years prior to Alison Redford's selection, but decreased by 21 per cent the year she was selected leader/premier. Finally, the Ontario Liberals exhibited a sharp decline in contributions the year prior to Wynne's selection as leader but then received more in contributions the year she was selected than they did in any of the previous three years. Again, this suggests the Ontario Liberals were a party in decline when Wynne was selected.
Though most parties posted declines in contributions prior to selecting women as leaders/premiers, Table 5 also shows that each party had considerable financial resources on hand. For example, though contributions to the Ontario Liberals were down prior to Wynne's selection, they still collected over $4 million in contributions in 2012. Similarly, though donations were down by nearly $3 million, the BC Liberals still collected over $9 million in donations the year before Clark was selected. While it may be tempting to conclude from these contributions data that these parties exhibited some financial decline prior to selecting women as leaders and premiers, it would not be fair to claim that this decline was steady or consistent over time.
Finally, public opinion polls may be able to provide evidence that a party was in decline prior to selecting a woman as leader/premier. Certainly, publicly available polls are of varying quality, and many fail to accurately predict a party's genuine level of support. Despite this, political media coverage is dominated by polls, often with little regard for their veracity (Matthews, Reference Matthews, Marland and Giasson2015; Pickup et al., Reference Pickup, Matthews, Jennings, Ford and Fisher2011). Polls are arguably reported at face value, becoming a large part of the public narrative, making them a legitimate way to identify if a party's support is in decline over time.
Historic polling data are available from Election Almanac (2017a, b, c, d). If there is indeed a scandal or unpopular previous leader that causes a party's support to drop off precipitously, it is plausible that this would be reflected in these data. Each of the four premiers in question, Dunderdale, Clark, Redford and Wynne, will be addressed in turn. Each figure starts with the election prior to each woman's selection as leader/premier, and continue at least until the next general election.Footnote 15
Figure 1 suggests the Newfoundland and Labrador PCs started to decline in popularity when Danny Williams resigned. Yet, when Dunderdale was selected as leader/premier, support for the party was close to 80 per cent. It is clear, though, that Dunderdale's selection did not stop the party's decline in popularity. Thus, though the party was in decline for most of Dunderdale's tenure as premier, the polls do not suggest that the party's decline was really underway prior to her selection, as would be the case if her selection could be explained by the party being in decline.
Figure 2 suggests the context was different for the BC Liberals prior to Christy Clark's selection as leader/premier. Following the general election in October 2011, BC Liberal support appears to have declined until Gordon Campbell resigned. At this point, the party's vote intention numbers were low enough that the Liberals would arguably have lost government in a general election. Yet, the party's support increased during the leadership campaign and appears to have been highest (around 40%) right around the time Clark was selected as leader/premier. A similar pattern is found for both the Alberta PCs (Figure 3) and the Ontario Liberals (Figure 4), though the increase in the Ontario Liberal party's support with Wynne's selection is muted, and the estimates of the Alberta PCs's support were uncertain. Where Alberta differs from BC and Ontario is that support for the party in government appears to have bottomed out well in advance of the previous premier's, Ed Stelmach's, resignation. However, in BC and Ontario, the party in government's support appears to have been lowest immediately prior to Campbell's and Dalton McGuinty's resignations as party leaders and premiers. By the time Wynne was selected leader, the Ontario Liberals’ support hovered around 30 per cent, suggesting the party would lose government in a general election. By contrast, with their parties polling around 40 per cent when they were selected, the same cannot necessarily be said for Clark or Redford.
Do these public opinion polls confirm that these parties were in decline when they selected women as their leaders/premiers? For Wynne and the Ontario Liberals, the answer is yes. The evidence is less clear, though, for the other parties. Only the Newfoundland and Labrador PCs show consistent, marked decline, and this appears primarily after Dunderdale was selected as leader/premier. Support for the other parties is unfavourable at some point before these women were selected as leader/premier, with the Ontario and BC Liberals polling below 40 per cent at some point and thus likely to lose if a general election were held then. Notably, in BC, Alberta and Ontario, party support rebounds after the leadership contest.Footnote 16
Overall, then, the evidence is mixed. By most of these measures, the Ontario Liberals appear to have been in decline when Wynne was selected leader. As such, her position was the most precarious, and she fits the literature's expectations about gender, leadership, and parties in decline. By contrast, both Clark and Redford's parties showed some signs of stable support and some signs of decline. Polls showed the BC Liberals might lose a general election if one were held before Clark was selected as leader. Yet, all parties were reasonably well financed during this period. In Redford's case, the Alberta PCs had a considerable number of electoral strongholds, suggesting they remained electorally competitive. The evidence for decline for these two parties is weak and mixed at best. There is no evidence to suggest that the Newfoundland and Labrador PCs were in decline when Dunderdale was selected; instead, the party's steady decline after Dunderdale's selection arguably has more to do with the vacuum created by the departure of a popular leader, or with her own leadership failures.Footnote 17
Neither crisis nor decline
Arguably, the crisis/decline narrative cannot explain how Rachel Notley came into the provincial premier's office. Instead, Notley's path fits the most identifiable pattern in the comparative literature on political parties: she was selected to lead a small, left-leaning, electorally uncompetitive party (Cross and Blais, Reference Cross and Blais2012; O'Neill and Stewart, Reference O'Neill and Stewart2009). Even though her party's election to government in the spring of 2015 was unexpected, the context surrounding her selection as party leader is in keeping with the existing generalizations found in the literature.
Discussion and Next Steps
This article set to out to identify the circumstances under which women have been selected to Canada's premier's offices. The literature suggests that women are most likely to be selected to these positions when their parties are in crisis or decline. Of the women who have served as Canadian premiers, four cases clearly fit this literature: Rita Johnston was selected leader and premier when her party was in crisis, while Pauline Marois was selected leader after her party experienced protracted decline to the point of crisis. The Ontario Liberals were, across multiple indicators, in decline when Kathleen Wynne was selected as leader. And, as leader of a small, left-leaning party, Rachel Notley's selection as leader is the case that best fits the literature. In other words, the comparative literature can explain half of these cases.
What, then, explains the other four premiers? They do not easily fit with the comparative literature investigating women's access to chief political executive offices. Parties selecting women as leaders, especially while in government, are predicted to be in decline. Yet this cannot readily explain why most of these women were selected simultaneously as party leader and premier. For Callbeck, Dunderdale, Clark and Redford, the evidence that their parties were in decline when they were selected leader is at best partial and inconclusive. Indeed, a critical reader may conclude these parties were not in steady decline when they selected women to the premier's office.
This conclusion is an uneasy counterpoint to conventional wisdom and post-hoc rationalizations. More research is required to assess how robust these findings are beyond this small number of cases, as well as to determine why existing generalizations do not fit these Canadian cases, as well as what these Canadian cases can contribute to the comparative literature. Admittedly, this exploration raises more questions than it answers. If these parties were not in obvious decline by these measures, then what best explains how these women entered the premier's office? Given the short period of time some of these women served as premier, it may be tempting to use ex post facto logic to claim their parties must have been in prior decline. Yet the evidence presented here does not comport with that post hoc rationalization. It is possible that none of these measures adequately capture a party in decline, though it is perhaps implausible that there is an empirical measure of decline that would capture something by-elections, polls, contributions, and vote and seat changes do not.
How much of how these women entered the premier's office is about popular prior leaders who define their parties, such as Joe Ghiz in PEI or Danny Williams in Newfoundland and Labrador? Certainly, these parties are not in decline as defined by the literature, but it is also clear that the resignation of popular leaders that arguably defined their parties (and governments) could precipitate a decline in the parties. It is possible, for example, that popular leaders who define their parties purge or otherwise remove competitive men from leadership contests that follow their resignation. Future research could assess this, as well as how much of women's selection as party leader/premier could be explained by leaders who did not necessarily define their parties, such as Ed Stelmach, Dalton McGuinty or Gordon Campbell.
Future research could also examine how women approach leadership contests, as well as how they lead their parties in election campaigns and in government. It would be particularly interesting to see if the campaign styles of women leading their parties in an election campaign for the first time do so differently if they are in government rather than in opposition. The role of caucus and other internal party organizations may be particularly important, as negative reactions to leadership styles are highlighted as reasons why some women face (gendered) caucus revolts (CBC News, 2014).
It would be interesting to see how the context in which women are selected affects their leadership and government styles, as well as their exits from office. Certainly, the context in which some women enter party leadership roles structure how they leave it (O'Brien, Reference O'Brien2015; O'Neill and Stewart, Reference O'Neill and Stewart2009). The same is arguably true for women premiers. This study demonstrates that some women become leader/premier when their parties do not appear, at least by these indicators, to be in decline. Given this, it is surprising that their tenure in the premier's office is comparatively short.Footnote 18 Generally speaking, in parliamentary democracies, party leaders’ tenure averages somewhere between 2 and 10 years (Cross and Blais, Reference Cross and Blais2012). When all Canadian party leaders are considered, the average tenure does not appear to be conditioned by leader gender. That said, women heading major parties had considerably shorter terms as leaders (2.4 years) than did their male peers (5.3 years, see O'Neill and Stewart, Reference O'Neill and Stewart2009). Women in the premier's office are, to date, little different from women heading major parties: on average, their tenure is about 31 months, or about 2.5 years (see Table 1).
Studies could also investigate the role of the media; preliminary findings suggest that women premiers receive about as much coverage as do men (Thomas et al., Reference Thomas, Harell, Gosselin, Marland, Lawlor and Giassonforthcoming). Women premiers may still receive different coverage compared to their male peers, both during election campaigns and during their time in government. A content analysis of media reports of polls may shine new light on the context when women are selected as party leaders/premiers, as well as when, as premiers, they make an early exit from the office.
Finally, the Canadian territories offer a unique context to study the effect of institutions on women premiers. Two women led “consensus governments.” Northern women's educational attainment and active recruitment into territorial public services act to reduce the structural barriers to women's involvement in politics, though important cultural barriers remain (White, Reference White, Trimble, Arscott and Tremblay2013). All representatives elected to the legislatures in the Northwest Territories and Nunavut are independents without party ties, and the premier is selected from among their members. This divorces the chief executive from political parties, while also allowing us to test hypotheses that contend that women are more likely than men to prefer consensus over contentious politics. Studies investigating institutional effects must note that chief executives in the territories are not as powerful as their provincial counterparts, as consensus government constrains those premiers in very different ways than their southern counterparts. Importantly, both of the territorial legislatures that govern through consensus—the Northwest Territories and Nunavut—have selected women premiers who served a single term.
Ultimately, the goal of the study is to explain if women premiers were selected by parties that were in crisis or decline. Indeed, because half of the cases cannot yet be easily explained, gender and Canada's provincial premiers' offices is a fruitful area for future investigation, for scholars of Canadian and comparative politics alike.