This article explores how candidates in the 2016–17 Conservative Party of Canada (CPC) leadership race campaigned on issues of moral traditionalism (for example, abortion and sexual diversity) and minority religious accommodation. Specifically, we examine the communication strategies that candidates used to navigate these divisive policy areas, which had support from various party factions but not necessarily from the majority of party members or the general public. We find that the overtness of candidates’ messages on religious issues is directly related to the level of public support for their positions. Those opposing the accommodation of minority religions or socially progressive policies on issues such as abortion or sexual diversity are more likely to express these views covertly, while those supporting such policies are more overt. We also find that candidates’ messaging changed following the close of membership sales. This article marks the first study of how party leadership candidates solicit support from particular party factions, or issue publics, and how candidate appeals evolve throughout a campaign. It also shows that despite the media attention given to the debate around minority religious accommodation during the CPC leadership race, contestation over the party's stance on issues such as sexual or reproductive rights remains very active.
Canadians are increasingly secular, with only one-third professing “deep devotion” to a religious practice (Angus Reid Institute, 2017). Yet Canada's politics continue to be shaped by faith perspectives. The CPC is home to a disproportionately large number of social and religious conservatives, making it challenging for the party to accommodate their preferences while not risking its broader electoral support (Rayside et al., Reference Rayside, Sabin and Thomas2017). While in government from 2006 to 2015, the CPC sought to bridge this divide through niche initiatives that signalled empathy with social conservatives but were less visible to the general public. These included tax measures to support traditional single-earner households, naming an ambassador for religious freedom and banning funding for abortion in Canada's foreign aid.
Internal party contests, however, provide different incentives. CPC leadership contestants only need to win the support of party members, and contestants can create a more supportive electorate by recruiting new party members. While national parties traditionally target the median voter, in direct leadership elections with multiple candidates, mobilizing—and appeasing—polarized constituencies (issue publics) within the party is a common strategy (Bawn et al., Reference Bawn, David and Seth2012). Such campaigning is also distinct, relying almost entirely on direct appeals to members rather than on public advertising. This combination of a smaller electorate, member recruitment and focussed appeals would appear to give socially or religiously conservative candidates more incentive to openly share their views in hopes of securing votes from like-minded members.
Yet social and religious conservatives, while more prominent than in the general population, do not compose the majority of CPC members (Farney, Reference Farney2012). Rival parties have used controversial statements from past leadership races to attack the CPC during general elections. Therefore, the strategies employed by socially or religiously conservative leadership candidates reflect an awareness that open appeals could have negative consequences within the race itself and over the longer term. Given this context, we argue that most CPC leadership candidates employed a Downsian strategy to avoid harming their political prospects: they structured their overt appeals to attract the median voter, while using covert appeals to signal to socially or religiously conservative members—a tactic common to conservative parties in the Canadian parliamentary system (Downs, Reference Downs1957; Nadeau and Bastien, Reference Nadeau, Bastien, Gagnon and Brian Tanguay2017).
To explore how such pressures shaped candidates’ behaviour during the 2016–17 CPC leadership race, we collected candidates’ email correspondence with party members and coded the position (supportive versus opposed) and openness (overt versus covert) of those messages signalling candidates’ stances on two issues of religious political contention: (1) the adoption of socially progressive policies (hereafter referred to as SPP) on issues such as sexual diversity, reproductive rights and assisted dying; and (2) policies supportive of the religious accommodation (hereafter referred to as RA) of minority faiths.
To examine how candidates reacted to updated information on the voting pool, we further disaggregated the findings into “early” and “late” campaign phases based on the deadline for membership sales. For each period, we compiled separate indexes summarizing each candidate's positioning and overtness on each issue. The characteristics of each message were then summarized in a dataset, enabling us to explore the relationship between policy stance and overtness through regression analysis.
The results show a clear correlation between candidates’ issue positions and the openness of their messaging: those opposing RA or SPP almost always did so covertly, while those supporting such policies were more open. Candidates also appeared to adjust their strategies over time. Regardless of their policy position, all candidates were more covert in their messaging on RA during the second phase of the campaign. The volume of messages opposing SPP also increased in the late campaign phase, both in absolute terms and relative to messages on RA.
We begin with a review of the literature on candidate issue ownership and campaign communications within leadership races, followed by a discussion of the relationship between faith communities and the CPC. We then provide an overview of the 2016–17 CPC leadership race, its candidates and issues related to social and religious conservatism. Next we present our method for gathering and coding CPC campaign emails and then present the results of the study.
Issue Ownership and Political Communication in Party Leadership Elections
Political parties bring together individuals with diverse interests to compete in elections. While supporters typically share some values, most parties contain multiple factions, or issue publics, with different priorities (Krosnick, Reference Krosnick1990). For example, Flanagan (Reference Flanagan2011) credited the CPC's 2011 federal election win to a coalition between western Canadian and Ontarian populists, traditional Tories in Ontario and Atlantic Canada, and “ethnic” new Canadian voters, who all shared a general economic and social conservatism.
Research on issue ownership demonstrates that parties are more than bundles of policy preferences; they also attract voters using issue reputations that reflect their campaign promises and performance in office (Budge and Farlie, Reference Budge and Farlie1983; Petrocik, Reference Petrocik1996). Like parties, individual candidates can also be issue owners, especially in leadership contests and primaries, and the differences between the issues owned by candidates and leaders can vary considerably (Tromborg, Reference Tromborg2018). Leadership candidates typically require support from multiple issue publics to win. Leader selection can also have profound effects on a party's direction and policy commitments. As such, leadership contests present competing issue publics a chance to shape their party's policy agendas. Leadership candidates, in turn, vie to own issues that will mobilize these publics, communicating their support for particular issues and highlighting their past performance on the subject.
In the 2016–17 CPC leadership race, candidates held positions on such issues as supply management, climate change, immigration and moral traditionalism that varied dramatically from past party policies and from each other. However, candidates face strategic dilemmas when choosing which issue publics to target: if certain publics hold controversial views, then courting their support risks alienating the broader electorate. Candidates may therefore employ covert or coded language in their appeals to minimize the risks to their public profiles. They may also rely on external endorsements, such as the evaluations of candidates’ voting records by the pro-life advocacy group Campaign Life Coalition, to communicate this issue ownership (Campaign Life Coalition, 2017c). This reliance on coded appeals also appears consistent with past research on leadership campaign messaging, which finds that candidates with ambiguous policy positions are more successful (Loewen and Rubenson, Reference Loewen and Rubenson2011). In this case, only those familiar with the covert messaging would know the candidate's position, leaving others uncertain.
Canadian party leaders were historically chosen by their parliamentary caucuses. The move to selection by members, first through delegated conventions and then mass membership votes, makes leadership campaigns a form of party primary elections (Cross et al., Reference Cross, Ofer, Scott and Gideon2016). Allowing more members to vote for party leaders also changed the nature of campaigning. As Cross et al. note, primary candidates “must reach thousands, tens of thousands, and sometimes hundreds of thousands of potential selectors across large geographic expanses. . . . While primaries are intra-party affairs, they more closely resemble general elections” (Reference Cross, Ofer, Scott and Gideon2016: 136).
The tight fundraising and spending limits imposed by the Canada Elections Act, combined with members’ geographic dispersion, make it impractical for candidates to rely on traditional media. Instead, Canada's modern party leadership campaigns rely on lower-cost tools, including direct mailings, websites, social media, email and candidate events. However, while recent years have seen a surge in interest in political marketing and communication in Canada (see, for example, Marland et al., Reference Marland, Giasson and Small2014), research on political communication by candidates for Canadian party leadership contests remains limited and focussed on the period before the rise of mass-membership votes and online campaigning (see, for example, Loewen and Rubenson, Reference Loewen and Rubenson2011).
In our study, we examine candidates’ emails to supporters. The use of emails for political communication has been explored in several settings. Jackson's (Reference Jackson2008) study of constituency email newsletters of British members of Parliament (MPs) showed that regular contact helped MPs to maintain existing supporters and convert those who were undecided or mildly committed to another party. A survey of subscribers to e-newsletters from UK political parties obtained similar results: most recipients were committed supporters seeking ongoing party engagement, while a minority looked to inform their vote choice (Jackson and Lilleker, Reference Jackson and Lilleker2007). In Canada, Marland and Matthews (Reference Marland, Mathews, Marland, Giasson and Esselment2017) explored parties’ use of emails as part of the permanent campaign, finding that the parties send regular emails to build member relationships and particularly to solicit donations. However, rather than blanketing all members with the same messages, parties target such emails based on the members’ demographic characteristics to maximize their interest and contributions.
The campaign's extraordinary length (over a year), combined with candidates’ need to not only recruit supporters but mobilize them to vote and contribute as volunteers or donors, would appear to make email the ideal communications tool for the 2016–17 CPC leadership race. Moreover, the comparatively private nature, greater length and self-selecting nature of email subscriptions mean that candidates may present their positions more openly and in more detail in emails than in the other mediums or forums noted above.
Moral Traditionalism, Religious Accommodation and the CPC
The relationship between faith and party politics in Canada was historically defined by the divide between Liberal-tending Roman Catholics and Progressive Conservative–leaning Protestants (Rayside et al., Reference Rayside, Sabin and Thomas2017). However, after slowly declining in the postwar era, this dynamic shifted in the early 2000s, as religiosity—rather than denomination—became a more consistent predictor of voter preference.
This shift resulted from the Liberals’ increasingly progressive policies on issues such as LGBTQ rights, which alienated religious traditionalists. At the same time, the CPC's creation in 2003 through the merger of the Canadian Alliance and Progressive Conservatives provided social conservatives with a clear home (Farney, Reference Farney2012; Malloy, Reference Malloy2017). The party's first leader, Stephen Harper, was initially sympathetic to socially conservative values. Under his leadership the CPC regularly used nonreligious language to appeal to social and religious conservatives in ways that went largely unrecognized by secular Canadians—what Lydia Bean characterizes as “dog-whistle politics” (Reference Bean2014: 12). The party also introduced policies favoured by those constituencies, such as banning foreign aid spending on abortion services and tax changes supportive of traditional single-income families. Moreover, it devoted considerable resources to wooing ethnic minority voters who traditionally voted Liberal. This outreach to “new Canadians” stressed the CPC's beliefs in socially conservative values and economic freedom (Tolley, Reference Tolley, Lewis and Everitt2017).
The party's initiatives to attract social conservatives and ethnic minority voters peaked in the 2011 election. Fifty-five per cent of religious voters supported the CPC while just 21 per cent voted Liberal—less than half the figure from the 2000 campaign (Rayside et al., Reference Rayside, Sabin and Thomas2017). The CPC also attracted suburban immigrant voters, sweeping it to a majority government (Bricker and Ibbitson, Reference Bricker and Ibbitson2013). Yet these developments proved fleeting, with the 2015 election seeing religious conservatives largely split their vote between the CPC and Liberals. Many immigrant-heavy suburbs also reverted back to Liberal red.
It remains unclear if the 2015 election was an outlier driven by campaign-specific effects (for example, a desire for change after three CPC victories) or if it marked a new voting trend. However, two factors may have harmed the CPC's outreach to these two communities. First, while in office, Harper grew increasingly reluctant to support socially conservative positions that might hinder the party's electoral success (Patten, Reference Patten, Farney and Rayside2013). In particular, following the 2011 election, Harper opposed several initiatives by backbench Conservative MPs that sought to reopen the debate on access to abortion services, leading many social conservatives to question his commitment to the issue (Rayside et al., Reference Rayside, Sabin and Thomas2017).
Second, while not abandoning its immigrant outreach, the CPC increasingly pursued policies to limit RA, especially for Muslims. These measures—which included banning face veils at citizenship ceremonies and the Zero Tolerance for Barbaric Cultural Practices Act that tightened laws on polygamy and other practices—also served to identify religious minorities as targets of public suspicion (Rayside et al., Reference Rayside, Sabin and Thomas2017). During the 2015 election, the party went further, proposing a “Barbaric Cultural Practices Tip Line” for police and musing about banning face coverings for federal public servants. While these initiatives ostensibly were proposed to promote gender equality, critics saw them as covert signals to win support from nativist and anti-Muslim voters (Ellis, Reference Ellis, Dornan and Pammett2016; O'Neill and Thomas, Reference O'Neill, Thomas, Dornan and Pammett2016).
Socially conservative policies and restrictions on religious minorities have limited public support. In 2018, 75 per cent of Angus Reid respondents agreed with the statement: “I think it's a definite plus to have people from different religious backgrounds and beliefs living here in my community” (Angus Reid, 2018). Likewise, Forum Research (2015) found that 70 per cent of Canadians supported same-sex marriage. Canadians’ support for reproductive rights is also high, with Ipsos (2017) reporting that 77 per cent of Canadians believe that abortion should be legally permitted and a majority (53%) saying that it should be permitted “whenever a woman decides.” Socially conservative voters, therefore, compose a minority of Canadians, making the use of more covert messaging around RA and SPP issues a strategic choice for leadership candidates.
CPC Leadership Race 2016–17
CPC leadership elections employ a hybrid voting system. Each member votes for the leader, but each riding has an equal weight in the overall outcome, no matter its membership size. Leadership candidates therefore need to not only attract supporters but also consider their distribution across ridings, and candidates may attempt to win by campaigning on niche issues relevant in less populated areas.
To vote, members needed to pay the $15 membership fee by March 28, 2017 (60 days prior to the election). This membership cut-off divided the campaign into two parts.Footnote 1 In the first phase, candidates not only canvassed existing members but also attempted to recruit new ones in hopes of building a supportive electorate (Cross et al., Reference Cross, Ofer, Scott and Gideon2016). Such recruitment efforts are regularly targeted at religious, ethnic or cultural groups that may be receptive (Cross, Reference Cross2004).Footnote 2 After the membership deadline, candidates switched focus from recruitment to maintaining their supporters and attempting to convert those supporting other candidates. Given the preferential ballot used in the race, such outreach was not necessarily zero-sum, with candidates seeking to establish themselves as voters’ second or greater choices.
The 2016–17 CPC leadership had 14 registered contestants (see Table 1), three of which stood out as single-issue candidates. Kellie Leitch distinguished herself with covert anti-RA appeals focussed on the need to screen immigrants for “anti-Canadian values,” including a lack of support for women's rights (Kingston, Reference Kingston2016). In contrast, Pierre Lemieux and Brad Trost both overtly opposed SPP, identifying as pro-life and advocating protections for free speech and conscience rights (Dickson, Reference Dickson2017). They were also the only leadership contenders to appear at the annual “March for Life” anti-abortion rally at Parliament in May 2017 (Platt, Reference Platt2017). Andrew Scheer did send a greeting that was read to march attendees, positioning him within the social conservative camp, although less visibly. However, Lemieux and Trost were the only candidates endorsed both by the Campaign Life Coalition and by Parents as First Educators (PAFE), which is an organization opposed to the modernization of Ontario's sex-ed curriculum (Allen, Reference Allen2017; Campaign Life Coalition, 2017c).
Table 1 Study Period Messages by Receiving Account

Legislative developments during the campaign also shaped the debates on SPP and RA. Anti-SPP messaging surfaced during the late 2016 debate on Bill C-16, which would protect transgender Canadians from discrimination and hate crimes. Social conservatives strongly opposed the measure, believing that it threatened freedoms of speech and association and would destabilize the traditional sex/gender binary. Trost and Scheer were particularly vocal in their opposition and voted against the legislation.
Anti-RA messaging similarly emerged during the early 2017 debate on M-103, a private member's motion that called for Parliament to condemn Islamophobia and all forms of religious discrimination and asked the House of Commons Standing Committee on Canadian Heritage to study how to eliminate racism and religious discrimination, including Islamophobia (Khalid, Reference Khalid2017). The motion sparked a fierce political mobilization, with opponents claiming that its vague definition of “Islamophobia” could limit free speech (Stone, Reference Stone2017). Others argued it was too focussed on Muslims and should condemn discrimination against all religions equally. Some even claimed M-103 was the first step toward imposing Sharia law in Canada. In the end, all leadership candidates except Chong opposed it.
After more than a year of campaigning, the final voting occurred on May 27, 2017. Leitch, the most vocal anti-RA candidate, finished seventh on the first preference ballot, with 7 per cent of votes, and was dropped following the tenth ballot. Anti-SPP candidates performed better: Scheer came second on the first preference ballot with 22 per cent support, Trost came fourth with 8.35 per cent and Lemieux sixth with 7.38 per cent. When Lemieux was eliminated after the eighth vote, most of his supporters moved to Trost. Those supporters then mostly switched to Scheer after Trost's elimination on the eleventh ballot. Despite trailing in all previous ballots, Scheer won on the thirteenth ballot, beating Bernier with 50.95 per cent to 49.05 per cent.
Many analysts attributed Scheer's win to unexpectedly strong social conservative support (see, for example, Rana et al., Reference Rana, Ryckewaert and Nash2017; Smith, Reference Smith2017). While Scheer was a leading candidate, Lemieux and particularly Trost lagged in fundraising, endorsements and opinion polls and so were expected to be peripheral to the outcome (Grenier, Reference Grenier2017). However, anti-SPP advocacy groups, including Campaign Life Coalition, PAFE and RightNow, claimed to have recruited many party members on their behalf (Campaign Life Coalition, 2017a; Paling, Reference Paling2017). These groups also took credit for the result. As the Campaign Life Coalition's post-election press release stated:
The results … demonstrate the strength of the social conservative movement and importance of pro-life and pro-family voters. The principled, bold social conservative candidates Brad Trost and Pierre Lemieux finished forth and seventh respectively, and Andrew Scheer, who has a pro-life voting record, won. CLC sold more than 11,000 memberships … and ran a strong get-out-the-vote campaign. (Campaign Life Coalition, 2017b)
By comparison, Scheer and his supporters stressed that his campaign had presented a message of unity that attracted members from different issue publics (Rana et al., Reference Rana, Ryckewaert and Nash2017; Smith, Reference Smith2017). As MP Tom Lukiwski, a Scheer supporter, described it: “They [social conservatives] played a huge role in Andrew's victory … Andrew has always said that he's always willing to listen and speak … with all factions within our party, including fiscal conservatives, social conservatives, libertarians, you name it” (quoted in Rana et al., Reference Rana, Ryckewaert and Nash2017).
Method
To examine candidate messaging on RA and SPP during the 2016–17 CPC leadership race, we subscribed to campaign emails from all registered leadership contestants. Data collection began in November 2016, with new candidates added as they registered. Subscriptions were made via candidates’ campaign websites using a Gmail address in the name of one the authors, who lived in the riding of Ottawa Centre. However, several challenges were encountered with the subscriptions. In early 2017, discussions of candidate emails on social media made it apparent that some candidates had sent more messages than we had received. Assuming that candidates may distinguish between party members and non-members in their communications, we began a second sign-up round in mid-January 2017 using the email of a party member residing in the rural riding of Huron-Bruce.
A further challenge was that despite offering email subscription tools on their websites, some candidates sent messages infrequently or not at all. For instance, five candidates sent at least one email to the non-member account in late 2016 but then stopped messaging it prior to January 2017. Of these, three (Leitch, Obhrai and O'Toole) resumed emailing the non-member account in February or March but not before they had begun emailing the member account. The other two (Chong and Saxton) stopped emailing the non-member account entirely. Chong did send messages to the member account, but the only email received from Saxton was a Christmas greeting to the non-member account. Lemieux and Peterson sent no messages to either account.
Our subscription difficulties may have resulted from candidates’ efforts to target different messages to different members. The O'Leary, O'Toole and Scheer campaigns tracked subscribers’ locations, with each account receiving different invitations to events in their respective communities. However, our research design does not allow us to explore what further microtargeting took place, since variations in message content could result from differences in the recipients’ location, membership status or other factors. Marland and Mathews (Reference Marland, Mathews, Marland, Giasson and Esselment2017) faced a similar challenge in their study of emails from Canadian parties, which employed a single receiving email account. While accepting that microtargeting may have shaped the messages received, they nonetheless contend that they received “the full universe of messages sent to a general user” (93). To offer insight into potential microtargeting based on geography and membership status, we disaggregate our results based on the receiving account whenever possible.
Although data collection began earlier, we limit the study to emails received from January 17, 2017 (the day we subscribed using the member account) until May 27, 2017 (the voting date). This approach concentrates the analysis on the most intense period of campaigning and minimizes variations in message volume between candidates that only contacted one account. Even though Kevin O'Leary launched his campaign after the study period began (January 18) and withdrew before it closed (April 26), we include him in the analysis, as he attracted considerable support, with over 33,000 members recruited (Canadian Press, 2017). O'Leary also continued emailing his supporters after resigning, in order to encourage them to vote for his own preferred candidate, Maxime Bernier.
Table 1 summarizes the emails that each account received during the study period. Of the seven candidates that contacted both the member and non-member, none sent an identical set of messages to both accounts, and no consistent pattern was observed in the distribution of the messages sent to one account, the other, or both. Therefore while some leadership contestants appeared to target different messages to different accounts, the basis of that targeting is not evident.
To make the volume of emails more manageable, we identified a relevant sample of messages by importing the emails into NVivo and conducting a content search for terms pertaining to SPP and RA.Footnote 3 To find SPP-related messages we used terms pertaining to the policy issues themselves (for example, abortion, sexual diversity, assisted dying, gender equality, education) as well as Christian terminology (for example, bible, Catholic, Easter). To identify RA-related messages we searched for both the names of minority religious groups (for example, Buddhist, Muslim, Sikh) and the social challenges such groups are accused of creating (for example, terrorism, radicalization, honour crimes). To capture both overt and covert appeals, we included not only those terms that dealt directly with the relevant issues (for example, assisted dying or Sharia), but also those that could covertly signal policy sympathy to those concerned with the issues (for example, conscience rights, values and family).Footnote 4 The supplementary online materials list the terms employed.
The search identified a sample of 469 messages. Each was coded for whether it dealt with one or both of the target themes (SPP and RA), the stance signalled (supportive or opposed) and the directness of the message (overt versus covert). The coding for each issue was recorded separately in recognition that a message could express overt support on one issue and covert opposition on the other.
We coded messages as pertaining to RA if they signalled (1) support or opposition for multiculturalism and diversity; (2) support or opposition for immigration; (3) raised concerns about religious extremism, either in Canada or abroad; or (4) raised concerns about the erosion of Canadian values or identity. We coded messages for SPP based on whether they signalled (1) support or opposition for the adoption of progressive policies on relevant social and legal issues (for example, abortion, assisted dying, LGBTQ rights, support for traditional single-earner households, sexual education), or (2) a willingness to allow opponents of SPP to advocate on those issues. Given the politicization of Christian religious language within Canadian politics,Footnote 5 we coded references to personal Christian faith or the use of religious language as signalling opposition to SPP unless the message content explicitly negated that understanding. While references to free speech could potentially demonstrate opposition to either SPP or RA, the longer history of free speech concerns among social conservatives meant that they were coded to SPP except where references to RA were also present.
For openness, we coded messages as covert if recipients would be unlikely to identify the policy stance signalled without prior knowledge of advocacy on the issue. For instance, committed social and religious conservatives would likely be aware that Prime Minister Harper sought to restrict backbench MPs’ efforts to reopen the abortion debate (Rayside et al., Reference Rayside, Sabin and Thomas2017). As such, they would presumably interpret candidates’ promises to give backbenchers more independence as a sign of support (or at least tolerance) for greater abortion advocacy. Similarly, those seeking to restrict RA would likely be familiar with M-103, and so could interpret references to the motion without further context. The presence of multiple words or phrases that could signal a candidate's views on SPP or RA increased the likelihood that a message would be coded to one of those issues.
For example, this excerpt from Andrew Scheer contains covert messaging on both SPP and RA:
I'm a real conservative who stands for getting rid of the GST/HST on home energy, making life more affordable for families and standing up for freedom of speech and against M-103. (Scheer email, May 11, 2017)
Many social conservatives would interpret the combined references to making “life more affordable for families” and “standing up for freedom of speech” as signals that Scheer supported both traditional single-earner households and open debates on contentious social issues, such as abortion. As such, they would likely conclude that Scheer was covertly signalling moral traditionalism and opposition to SPP. At the same time, the statement that he is “standing up … against M-103” would be interpreted as covert opposition to RA.
In contrast, we coded messages as overt if the stance expressed would be evident even to those unfamiliar with the issue. As these excerpts demonstrate, coding such messages was more straightforward:
Muslim radicals want us to follow Sharia Law. They insidiously want to rollback rights which women fought very hard for. Not under my watch. Equal rights of men and women is too important. (Blaney email, March 16, 2017)
I have advocated family-friendly, child-friendly and seniors-friendly policy including: a ban on gender-selection abortion and protection for pre-born victims of crime; to resist and to repeal Bill C-16 that wants to allow biological men in women's and girls’ washrooms. (Trost email, April 27, 2017)
To simplify the process, one author coded all messages for SPP while the other coded for RA. To ensure consistency, a random set of 10 per cent of the messages from the text-search sample was recoded by the opposite author. This exercise produced an average kappa value of 0.8559 across the eight potential coding possibilities (for example, SPP support-covert; SPP support-overt; SPP oppose-covert, etc.), indicating strong inter-coder agreement.
Results
Table 2 presents an overview of the emails coded as relating to RA or SPP, broken down into early and late periods. Throughout the entire study period slightly more messages touched on RA (154 or 21.7%) than SPP (142 or 20.0%). However, this result masks variation over time. While message volumes were relatively stable between periods (326 vs. 383), the proportion dealing with at least one issue rose from roughly one-quarter in the early period to over one-third in the late campaign. SPP-related messages had the highest growth, with those touching on both issues and SPP alone doubling in absolute and proportionate terms. In contrast, messages focussed solely on RA declined.
Table 2 Study Period Emails by Issue

As Table 2 shows, the change in issue volumes between periods is statistically significant, suggesting that the candidates altered their message strategies over time. However, the reason for the change is less clear. The decline in RA-only messages suggests that candidates moved on to other issues after M-103 passed. The late-period spike in SPP messages also partly resulted from candidates’ Easter greetings, although the change in issue focus remains significant without them.Footnote 6 Instead, as described further below, the rise in SPP messaging appears to reflect increased competition for socially conservative voters after the membership deadline.
Looking beyond differences between periods, Tables 3 and 4 display how message content varied between the two email accounts. While receiving similar message volumes on SPP and RA, the member account received fewer messages overall. As shown in Table 4, RA- and SPP-focussed emails composed a marginally greater proportion of messages to the member (34%) than to the non-member (27%), potentially suggesting that leadership contestants targeted such messages at rural recipients, who often are assumed to be more morally conservative than their urban counterparts. A closer examination of Table 3, however, reveals that in absolute terms, the non-member account received more messages opposing RA and SPP, while the member account received more supporting progressive positions on both issues. The breakdown of message sentiment and overtness in Table 4 further demonstrates that anti-RA and anti-SPP emails sent to the non-member were more likely to employ overt language than those sent to the member. Therefore, while the rural-based member account received more messages concerning SPP and RA, those received by the urban-based non-member account were more likely to express overt social conservatism.
Table 3 Sentiments in Study Period Emails by Receiving Account

*Messages may express sentiments for both RA and SPP. As such, sub-sentiment counts exceed the total number of messages with religious sentiments.
Table 4 Sentiments in Study Period Emails as Percentage of Messages to Each Account

*Messages may express sentiments for both RA and SPP. As such, sub-sentiment percentages exceed the total percentage of messages with religious sentiments.
This unexpected focusing of overt socially conservative content on the non-member account largely results from the fact that the two candidates most overtly opposed to RA and SPP, Blaney and Trost, only communicated with that account. Of candidates who messaged both accounts, only Leitch sent more emails with socially conservative sentiments to the member account. However, the content of Leitch's messages did not vary between accounts—the member just received more.
Tables 5 and 6 present the coding of candidates’ emails for messaging on RA and SPP. To examine messaging strategies, we combined the counts of supportive and opposed messages to produce a summary statistic for each candidate on each issue in each period. To facilitate comparison, we standardized these summary statistics as a proportion of each candidate's messages during each period. We repeated the process to produce standardized summary statistics for message openness as well.
Table 5 Coding of Emails for Signalling on RA

Table 6 Coding of Emails for Signalling on SPP

Leitch was the most opposed to RA, a view evident in nearly 80 per cent of her emails. Blaney, Scheer and Trost followed, with roughly one-quarter of their messages opposed. Bernier also consistently opposed RA in a handful of messages in each period. O'Toole and Alexander both sent a mixture of supportive and opposing messages but were opposed on aggregate. On the other side, Chong and Obhrai showed clear support for RA in 30 and 39 per cent of their messages, respectively. Raitt had a single message supporting RA, while O'Leary did not address the issue. Overall, candidate messaging on RA was more opposed in the late campaign, which is surprising given that M-103 was debated in the early phase. The major exception to this trend was Scheer, who sent far fewer anti-RA messages in the second phase in both absolute and proportionate terms than he had in the first. Although the reason for Scheer's change is unclear, the decline in anti-RA messaging brought him closer to the RA positioning adopted by Bernier, his strongest competitor.
In contrast, candidates’ signalling on SPP was markedly different between the early and late campaigns. While opposing views dominated in each period, the volume of messages on both sides of the issue increased sharply in the second phase. O'Leary sent the sole pro-SPP message in the early campaign, while Obhrai emerged as the progressive champion in the late campaign. Chong was also supportive on balance, although his score was reduced by two messages about his plan to give backbench MPs more freedom, which were coded as covert appeals to social conservatives.
On the opposing side, the proportion of anti-SPP messages rose by more than 10 percentage points in the late period for Bernier, O'Toole and Scheer, and by more than 20 percentage points for Leitch and Trost. This anti-SPP spike appeared to reflect increased competition for socially conservative voters. For instance, Trost sent 12 late campaign messages asking voters to rank him first and Lemieux second, as the only candidates with traditional views on abortion and conscience rights. Three of the emails also asked members to not support Scheer, alleging that he would prioritize electoral victory over conservative principles. Scheer himself sent 11 late campaign messages that signalled opposition to SPP by highlighting his support for free speech. Of these, two messages justified free-speech protections by pointing to restrictions on pro-life and pro-Israel events at universities.
As noted above, we also produced summary statistics capturing each candidate's overall message covertness on each issue in each period. To examine the relationship between these sets of coding, Figure 1 plots the summary statistics for candidates’ RA policy positions versus those for their message covertness on the issue, while Figure 2 plots policy stance versus covertness for SPP. The figures suggest that policy stance and covertness are correlated: candidates opposed to both issues overwhelmingly employed covert language, while supportive positions were expressed more overtly. The main exceptions were Blaney, who openly opposed RA, and Trost, who overtly opposed SPP throughout the campaign and RA in the first phase.

Figure 1 Support for RA by Message Openness

Figure 2 Support for SPP by Message Openness
Trost and Blaney were peripheral candidates, finishing fourth and ninth respectively. Rather than reaching out to different issue publics, their campaigns focussed on certain subjects. Indeed, Trost described joining the race specifically to raise issues such as “Life, including a prohibition of gender-selection abortion, and for protection of pre-born victims of crime” (Trost email, May 6, 2017). He further urged his supporters not to consider candidates’ potential for general election success but to “vote your principles.” However, the other peripheral candidates generally couched their anti-RA and anti-SPP appeals in more coded language.
In terms of campaign phases, Figure 1 shows that more candidates moved into the opposed/covert quadrant between the early and late periods, suggesting that candidates became less willing to openly oppose RA over time. Figure 2 reveals a similar pattern for SPP: candidates were initially closely clustered in their policy positions and openness but became more dispersed in the late campaign, with several becoming more opposed. Except for Leitch, SPP messaging also became increasingly overt, including several messages from Obhrai openly supporting LGBTQ rights.
To test the significance of these trends, we constructed a dataset that captured each email's coding for covertness, policy position and campaign period as binary variables.Footnote 7 The dataset was weighted for each issue in each period to ensure that the analysis was not skewed by variations in message volumes between candidates.Footnote 8 We then conducted binary logistic regression analyses to examine if message overtness is indeed shaped by candidates’ policy positions and the campaign phase.
The results in Table 7 confirm that pro-SPP or pro-RA messages were significantly more likely to be overt, while those opposing the two policies tended to be covert. Candidates also appeared to adjust their messaging strategy as the campaign went on: they were significantly more likely to use covert language to signal their views on RA in the late phase of the campaign, although there was no such change in messaging on SPP. This reality may reflect the negative media attention on the RA issue.
Table 7 Binary Logistic Regression Results for Relationship between Message Openness and Policy Stance

*p = .1, ** p = .05, *** p = .01, **** p = .001
The results further reveal the balance between advocacy on RA and SPP within the CPC. Despite Harper's efforts to restrict anti-abortion advocacy, our findings suggest that it remained prominent among CPC candidates seeking support from morally traditional issue publics. Moreover, rather than competing, advocacy on RA and SPP largely ran in parallel. The number of messages on each issue was similar, and candidate positions on the issues were closely linked. This trend would support our past contention that opposition to RA functions as a new vehicle for mobilizing social and religious conservatives, but it casts doubt on our suggestion that it may replace advocacy on SPP (Rayside et al., Reference Rayside, Sabin and Thomas2017).
Finally, it should be noted that despite the media's focus on candidates’ positions on RA and SPP, emails touching on those issues accounted for less than a third of all study period messages. Moreover, many RA- and SPP-focussed messages also discussed other issues. Consequently, religious issues were not the primary focus of the campaign.
Conclusion and Discussion
Our results demonstrate that the level of public support for the positions held by leadership candidates shapes the overtness with which they signal those positions to supporters. Candidates advancing positions popular with some party members but not the general public tend to employ covert language that only those familiar with the issue will likely identify. Conversely, those supporting positions with more public support tend to employ more overt language. In the 2016-17 CPC leadership race, contestants who opposed the accommodation of religious minorities or socially progressive policies on issues such as sexual diversity and abortion were significantly more likely to signal their views covertly. By comparison, candidates who supported these policies utilized overt language that would be evident to all members. The two candidates who consistently defied this relationship between content and covertness, Blaney and Trost, appeared unconcerned with how their overtly socially conservative messages could affect the party's broader electoral success.
We further find that leadership candidates may vary the content and directness of their appeals between campaign stages. Prior to the membership sales deadline, candidates’ primary objective is to recruit new members who will support the candidate with their first-preference votes. Once sales close, however, candidates compete for the second, third or greater preferences from members recruited by other contestants. This dynamic is evident in the post-cut-off surge in SPP messaging. All late campaign messages concerning RA were also significantly more likely than those in the early phase to employ covert language.
Further research is needed to explore if similar trends exist on other issues, in other parties or in other jurisdictions. For instance, did CPC candidates with more extreme positions on healthcare privatization signal these views covertly? Does message overtness change with candidates’ perceptions of a party's chances of winning the next election? And does the membership cut-off have the same impact on leadership campaigns for parties that do not weight votes between constituencies?
This research also enhances our understanding of issue ownership within leadership campaigns. Candidates pursue different strategies to mobilize issue publics within their party, while recognizing the risks that such appeals may pose in subsequent national elections. Importantly, these findings reinforce our understanding of the moderating effect of national parties on the embrace of ideological and policy extremes by political actors. This is not to say that Canadian politics no longer sees calls to restrict ethnic, religious or sexual diversity. Indeed, recent elections in Ontario, Quebec and New Brunswick may signal a shift in the electorate's acceptance of anti-RA or -SPP messaging, especially in Quebec where the Coalition Avenir Québec government remains focussed on banning the wearing of overt religious symbols by public servants and reducing the province's intake of immigrants (Shingler, Reference Shingler2019).
The findings also appear to contradict our past suggestion that opposition to RA is replacing anti-SPP advocacy as the primary focus of political contention for Canada's social and religious conservatives (Rayside et al., Reference Rayside, Sabin and Thomas2017). While the former dominated the CPC's 2015 federal election messaging, many leadership candidates actively opposed both issues. Indeed the candidates’ total RA and SPP support scores in Tables 5 and 6 were highly correlated, suggesting that the messages appealed to the same audiences.Footnote 9 As such, contention on both subjects will be likely within the CPC for the foreseeable future.
Finally, the data collection challenges confirm that some candidates varied their communications to different recipients. Several candidates only contacted the member account, suggesting a focus on committed partisans. Others only messaged the non-member, potentially indicating a desire to attract new members. Still others emailed both accounts, though none sent an identical set of messages to each recipient. Our research design did not allow us to determine the basis for the variation observed, although the non-member account received more overtly socially conservative messages than the member account. Further research with more members and non-members who vary in location and demographics is required to investigate how parties and leadership contestants target their emails across recipients. We also encourage researchers studying future campaigns to monitor social media during the race to detect any potential gaps in the messages received.
Supplementary material
To view supplementary material for this article, please visit https://doi.org/10.1017/S0008423919000246
Acknowledgments
We dedicate this article to David Rayside, who mentored us in the study of religion in Canadian politics. We also thank Jeff Sabin, Chris Alcantara and Jonathan Malloy for their support and encouragement and the anonymous reviewers for their helpful comments. We are also grateful for the support we each received from the Social Sciences and Humanities Research Council Postdoctoral Fellowship Program while conducting this research.