Introduction
Women represent half of the world's population, making gender parity in legislatures a central measure of equality in representative democracies. In a free and just society, a normative expectation includes a human right for women to participate in public life on an equal basis to men, free of direct or indirect discrimination. Yet women rarely hold equal representation in elected governments across the globe. The representative democracies in this study—the United States, Australia and Canada—are no exceptions. They are selected for study because these three liberal, multicultural societies have much in common: advanced economies, federal systems of government, and all rank poorly in global indices measuring women's representation in parliaments (Inter-Parliamentary Union, 2021). Although some parties in Canada and Australia have established candidate quotas or targets (namely the New Democratic Party in Canada and the Labor Party in Australia), none of these countries have generalized legislated quotas in their parliaments. This is unsurprising given that these three countries are classified as liberal welfare states that emphasize market rather than government interventions (Esping-Anderson, Reference Esping-Andersen1990) suggesting that quota solutions may be less palatable to the general public. Moreover, none of these countries have achieved overall gender parity in their parliaments, and all have significant gender gaps in their lower chambers (UN Women, 2023). Indeed, recent statistics on gender representation across these countries shows they all rank poorly, albeit with substantial variation in their ranking: Australia ranked 33, Canada ranked 61 and the United States ranked 66 for women's political representation (UN Women, 2023). Thus, non-quota measures may be critical to increasing women's representation and garner more support, given these countries' policy landscapes emphasize individual over government interventions.
Increasing women's political participation and representation is shown to bring a host of benefits, including enhancing democratic legitimacy (Schwindt-Bayer and Mishler, Reference Schwindt-Bayer and Mishler2005) and generating better policy outcomes (Norris and Franklin, Reference Norris and Franklin1997). Despite the importance of equal representation, women across the world are far less likely than men to occupy senior positions of influence (Oakley, Reference Oakley2000), including in national legislatures. For example, only about a third of the national lower house chambers in each of our case countries are women. Australia has the highest women's representation in the lower chamber of parliament of the three countries, after an unprecedented 38 per cent were elected in the 2022 federal election, up from 30.5 per cent in 2019 (Hough, Reference Hough2019). This compares to Canada's House of Commons at 30.7 per cent and the US House of Representatives at 29.4 per cent women (UN Women, 2023). Australian studies show that women are generally more electable than men, but fewer women seek election in local government and the federal lower house in the first instance, limiting the probability of achieving gender parity (Carson et al., Reference Carson, Ruppanner and Lewis2019; Carson et al., Reference Carson, Mikolajczak and Ruppanner2021; Martinez i Coma and McDonnell, Reference Martinez i Coma and McDonnell2023).Footnote 1 These findings of fewer women running for office than men are consistent with international studies (Adams and Schreiber, Reference Adams and Schreiber2011). Thus, gender parity in these countries is further limited by fewer women seeking public office.
To narrow the gender gap in elected politics, governments, political parties, policy makers and advocates have paid attention to affirmative action strategies such as gender quotas, which are used in more than half of the world's countries to increase women's political participation (McCann, Reference McCann2013). While the bulk of evidence suggests that legal, voluntary electoral and party gender quotas assist at closing the gap (Beauregard and Sheppard, Reference Beauregard and Sheppard2021: 231; Beauregard, Reference Beauregard2018: 290), it remains that only 25 per cent of the world's national parliamentarians are women, up from 11 per cent in 1995 (Berevoescu and Ballington, Reference Berevoescu and Ballington2021). According to a United Nations’ report, gender parity in national parliaments at the current rate of progress will not be achieved before 2063 (Berevoescu and Ballington, Reference Berevoescu and Ballington2021). A secondary problem for gender quotas is empirical scholarship documenting that support for quotas cleaves along gender and political lines (Beauregard, Reference Beauregard2018; Dubrow, Reference Dubrow2011). This can polarize public and elite opinion about quotas and limit their widespread adoption. Clearly, quotas play a critical role in narrowing the gender gap in electoral politics; but other mechanisms need to be considered to close the gender gap in a timely manner. This has led to a call for gender mainstreaming, or benchmarking gender as a critical lens for parliamentary decision making, to create a gender-sensitive government that includes a variety of policies and resources (Palmieri, Reference Palmieri2011). Gender mainstreaming emphasizes that federal, state and local governments’ policy, resource and employment decisions should be understood through a gender lens to foster greater gender equity. Support for a range of non-quota gender-mainstreaming policies may be less politically polarizing, making them more feasible to implement across party lines to help narrow the gender gap in elected politics. None of the case countries—Australia, Canada or the United States—are leaders in gender mainstreaming. Thus, it is critical to understand the public appetite for these types of measures that have shown to be effective in other nations (Palmieri, Reference Palmieri2011).
This motivates our study. We focus on public support for non-quota policies across three similar, established democracies with long-standing gender gaps in their lower house chambers of parliament. Compared to gender quotas, as well as the broader literature on the political representation of women and electoral support for women candidates (for example, Carson et al., Reference Carson, Mikolajczak and Ruppanner2021; Dolan, Reference Dolan2019; O'Neill, Reference O'Neill2015), non-quota measures are an understudied area of research (for exceptions, see Palmieri and Childs, Reference Palmieri and Childs2020; Palmieri, Reference Palmieri2011). The aim of the article is to examine public attitudes to a number of these measures at both local and federal levels in the three case countries. We pair the political theory of gender mainstreaming—the emphasis on government policies and procedures that promote gender sensitivity, inclusivity and representation—with the sociological theory of role strain that underscores, in this case, the conflicting roles of being a parent or spouse with the intense demands of being a public representative. While both men and women face the demands of fulfilling numerous roles, we know from a breadth of research that women politicians or candidates often face difficulty in combining a political career with home life that may impact their decision to run for office, their experiences while in office, and decisions to recontest (Carson et al., Reference Carson, Mikolajczak and Ruppanner2021; Fox and Lawless, Reference Fox and Lawless2005). From this theoretical basis, we select non-quota gender-mainstreaming mechanisms that may alleviate role strain, particularly for women. We expect the public to be more supportive of these non-quota measures that may encourage women to run for and/or stay in office longer. The study's rationale is that before policy makers can implement these gender-mainstreaming measures, it is important to first know if these incentives—a pay raise, childcare subsidies, housework allowances or greater access to virtual meetings—will face public opposition similar to the gender bias or partisan opposition that has polarized public attitudes toward gender quotas (Beauregard, Reference Beauregard2018).
To understand these relationship, we use a survey-based experimental design and large country samples (Australia, N = 8,413; Canada, N = 9,127; and United States, N = 8,454) to test public support for these practical and compensatory non-quota measures aimed at easing competing strains that impede women's (and men's) political careers, as identified in the literature. Our research is guided by three main questions: How does a politician's gender shape public support for gender-mainstreaming policies aimed at lessening work–family role strain? How does public support for such policies vary by levels of government—for example, national versus local? And do women report greater support for policies for women politicians—demonstrating an affinity bias? By addressing these questions, we seek to understand if there is a public appetite to recommend gender-mainstreaming policy changes that may incentivize more women candidates to run and recontest office to help narrow the political gender gap if elected.
This article proceeds as follows: We outline the literature documenting some of the structural barriers that limit women's participation and representation in politics and differences between women's representation in federal and local political spheres (with gender and political spheres as our independent variables). We then draw on gender-mainstreaming and role-strain theories to understand public attitudes about the hypothetical measures in our study (the dependent variables). We provide rationale for our hypotheses and describe our experimental research design before documenting our results. Finally, we discuss the implications of our findings and conclude with key recommendations for policy makers and avenues for further studies. In doing so, our study contributes to the literature on women's representation in established democracies by analyzing public support for non-quota gender-mainstreaming strategies designed to create more gender-sensitive and inclusive governments.
Theory and Hypotheses
The gender gap
Scholarship shows that women face distinct structural barriers to enter elected politics that require interventions if they are to be overcome. These barriers can be broadly categorized as supply-side, which is a lack of women entering politics; or demand-side, which is when women enter politics but are then marginalized or denied positions of influence (Murray, Reference Murray2014: 521–22). On the demand-side, a critical mass of women is often seen as essential to transition from political power derived by a small contingent of women to a considerable minority (Kanter, Reference Kanter1977; Dahlerup, Reference Dahlerup1988). It is argued that only when the number of women in legislative positions is increased will women be able to promote women-friendly and women-oriented policy change, while simultaneously influencing their men colleagues to accept and approve legislation for the benefit of women constituents (Childs and Krook, Reference Childs and Krook2008). While a larger contingent of women representatives is critical, Childs and Krook (Reference Childs and Krook2009) also document the importance of substantive representation by focusing on what women in legislative positions do and how they make a difference. Supply- and demand-side measures are equally valuable, as the more gender-inclusive the internal (party) and external political environments, the more women are expected to run.
Here, we expand upon a breadth of scholarship that largely focuses on national jurisdictions and single out a number of supply-side factors that impede the ability of women to enter politics, enabling men to play an outsized role in elected politics, including the socialization of gender roles in society (Dolan and Lynch, Reference Dolan and Lynch2016; Huddy and Terkildsen, Reference Huddy and Terkildsen1993); the perceived qualifications and traits deemed appropriate for political office (Rosenwasser and Seale, Reference Rosenwasser and Seale1988); access to resources (including free time); “in-group” bias against women, where partisanship can act as a group membership and members are more likely to hold similar viewpoints (Huddy, Reference Huddy2001); negative media coverage of public women (Carlin and Winfrey, Reference Carlin and Winfrey2009; Williams, Reference Williams2017); and scheduling of meetings to suit men's schedules and networks (Beauregard and Taflaga, Reference Beauregard and Taflaga2023: 373; Murray, Reference Murray2014: 521–22; Krook and Norris, Reference Krook and Norris2014: 5).
Further, historical factors such as women's denial of suffrage (and by extension representation) until the twentieth century have reinforced gender imbalances in democratic parliaments (Tremblay, Reference Tremblay2008). A further challenge to gender equality is the role that incumbency plays in elections. Studies find incumbents have a re-election advantage, but if fewer women are incumbents, a smaller proportion than men counterparts can enjoy this benefit (Carroll and Fox, Reference Carroll and Fox2006; Gelman and King, Reference Gelman and King1990). This underscores a historical feature of the composition of many parliaments, including the case countries, whereby men politicians, on average, enjoy longer tenure of representation than women (O'Neill and Stewart, Reference O'Neill and Stewart2009; Glenday, Reference Glenday2021; History, Art & Archives, U.S. House of Representatives, 2023). In addition, Sawer (Reference Sawer2002) finds a preference for “gladiatorial politics” and a lack of mentors for women politicians at the national level that can also act as impediments to gender equality. As these various studies indicate, a range of barriers impact both the entry to politics and the experiences women face once they enter political office.
Local government is often an overlooked area in “women and political representation” studies (Pini and McDonald, Reference Pini and McDonald2011). Of the limited studies that do focus on women in local government, researchers find it is a critical pathway for providing women with “visibility” in politics, encouraging women into grassroots elected roles (for example, school boards) and providing both experience and a pipeline into higher tiers of elected government (Adams and Schreiber, Reference Adams and Schreiber2011; Pini and McDonald, Reference Pini and McDonald2011: 1; Ransford and Thomson, Reference Ransford, Thomson, Pini and McDonald2011). Cross-national studies find that out of 133 countries, women's representation in local government is overall higher (36 per cent) than in national parliaments (25 per cent) but still far from parity (Berevoescu and Ballington, Reference Berevoescu and Ballington2021).
There are clear benefits to increasing women's representation at the local level. In addition to being a critical site of women's representation and community engagement, local government can be a training ground for representation at state and federal levels, for many, providing a springboard into other tiers of government. Thus, improving women's representation at the local level may help to achieve balanced representation at all levels of government, through increased supply. Because local politics is closer to home and requires some travel but usually less than federal counterparts, it is sometimes erroneously assumed to be less demanding for balancing work and family life (Carson et al., Reference Carson, Mikolajczak and Ruppanner2021; Pini and McDonald, Reference Pini and McDonald2004; Briggs, Reference Briggs2000; Lovenduski, Reference Lovenduski1986). However, existing research on women representatives’ experiences at the local level expose competing work and family pressures. Jakimow's (Reference Jakimow2022) study of local government councillors found some councillors delayed running for office, as “managing a busy household remains a barrier for many women, and some men” (Reference Jakimow2022: 82). Jakimow found that local government representatives were better able to manage different work and family roles at the same time only when measures such as “caring allowances and the ability to remotely attend council meetings” were in place (82). Jakimow found that “motherhood and ideal models of womanhood [were], however, used as a weapon to deter women or frustrate their political ambitions” (83). These work–family challenges mean women politicians often find it difficult to manage the dual pressures from their roles in family and public office and thus may be deterred from public office and seeking (re)election. To mitigate these pressures, research on gender mainstreaming identifies a range of solutions adopted by governments around the world to reduce work–family strain (Palmieri, Reference Palmieri2011). As outlined below, we use some of these proposed solutions as a framework to identify public support for provisions aimed at reducing the unique role strain women often experience in elected office.
Gender mainstreaming and role strain: Theorizing the difficulty in combining political roles and home life
Gender mainstreaming centres gender in decision making across all levels of government to ensure that institutions are sensitive to the issue and that policies are decided and resources allocated in ways that equally incorporate men and women (Palmieri, Reference Palmieri2011). Research shows women politicians disproportionately face competing demands that make combining work and family particularly difficult (Thomas, Reference Thomas2002; Fox and Lawless, Reference Fox and Lawless2005; Carson et al., Reference Carson, Mikolajczak and Ruppanner2021), which activates inter-role conflict (Mitchell, Reference Mitchell1958; Goode, Reference Goode1960). Indeed, the role of the politician is a particularly demanding and stressful public-facing role that when combined with other roles such as parent, daughter or spouse, can foster inter-role conflict, stress and burnout (Mitchell, Reference Mitchell1958; Poms et al., Reference Poms, Fleming and Jacobsen2016). What is more, women politicians are often held to different standards around their public-facing roles as mother, wife and politician than their men counterparts (Klammer and Klenner, Reference Klammer, Klenner, Leitner, Ostner and Schratzenstaller2004: 184; Carson et al., Reference Carson, Ruppanner and Lewis2019) and are more likely to not enter or to leave politics in response to these challenges (Stalsburg and Kleinberg, Reference Stalsburg and Kleinberg2016; Ransford and Thomson, Reference Ransford, Thomson, Pini and McDonald2011). Women politicians in local government were more likely than their men counterparts to sacrifice full-time work because the demands of their family and political career were in conflict, thus intensifying economic inequality in politics between the genders (Carson et al., Reference Carson, Mikolajczak and Ruppanner2021).
Reducing these gendered disparities is a central tenet of gender mainstreaming, with the goal to enact policies and resources to support the caregiving demands of women (and men) politicians (Palmieri, Reference Palmieri2011). Jakimow (Reference Jakimow2022) shows policies such as access to childcare resources and remote participation in council meetings were effective resources to support women politicians. Palmieri and Childs (Reference Palmieri and Childs2020) identify equal access to digital literacy, resources and participation as critical to women representatives’ inclusion during the COVID-19 pandemic. And Palmieri (Reference Palmieri2011) documents the diverse resources adopted by parliaments around the world to support caregivers, including on-site and/or financial assistance for childcare, schedules realigned to school calendars, fewer night votes and greater flexibility in work location. These resources form a critical dimension of gender mainstreaming to create gender-sensitive parliaments, in part by reducing work–family role strain, particularly for women.
What remains unclear is whether publics are supportive of allocating resources to create more gender-sensitive governments to help women politicians remain in office. Understanding the public's support for smaller incremental resources, such as those noted in Jakimow's (Reference Jakimow2022) study (that is, caring allowances and virtual meeting options) is critical, given the political fissures over quotas. Here, we focus on four non-quota mechanisms capturing gender mainstreaming: a pay raise, housework allowance, childcare allowance and making available online meetings.
From the literature outlined above, we expect support for particular policies to differ when the target/beneficiary is a woman (versus a man) politician. Specifically, we expect the public in all three countries to be more supportive of these resources for women than men politicians, in part because of women's disproportionate responsibility for home life and, for some, a desire to create more gender-equal government that elevates the status of women representatives. Thus, we expect the public to view these resources as more urgently needed for women politicians and thus exhibit greater support. Our first hypothesis is:
H1: Public support will be greater for women politicians than men politicians to receive additional non-quota gender mainstreaming benefits to support their political careers.
We also expect public support to be greater for those in local than federal government. Understanding public support for non-quota gender-mainstreaming measures in local elections is critical, as local governments are decentralized, making it more problematic to introduce quotas. Quota laws are also more difficult to adopt in systems with decentralized candidate nominations, as well as potential co-ordination problems inherent in directing a national candidate slate (Baldez, Reference Baldez2006: 107). Gender quotas work by bringing more women into the political system, but they do not promise to change the dynamics of political party processes and so the status quo may endure (Baldez, Reference Baldez2006: 106). In some instances, such as some local governments in Australia and Canada, political parties are absent. Support for quota measures dictated at the party or national level is therefore not feasible. But non-quota measures, aimed at providing more resources for local leaders, may be. The public often feels more connected to and trusting of local politicians (Fitzgerald and Wolak, Reference Fitzgerald and Wolak2016), which may bolster support for non-quota gender-mainstreaming resources for local representatives. Local politicians may be compensated for local representation on a part-time basis and thus juggle this with other work, earning much less than their state (provincial) or federal counterparts for their representative role (Victorian Independent Remuneration Tribunal, 2023). This may foster greater sympathy from the public about additional supports for these politicians’ elected roles. Thus, there are good reasons to increase women's political representation beyond a reliance on gender quotas. Here, we assess the extent to which the public reports greater support for non-quota gender-mainstreaming resources. Given public perceptions that local government is considered more family friendly than federal politics and is closer to home and that politics at the federal (or national) level is deemed more “gladiatorial,” we would expect the public to be more supportive of these policies when directed toward local government compared to federal politics. Therefore, our second and final hypothesis is:
H2: Public support for offering non-quota gender-mainstreaming incentives for politicians will be greater for local than federal officeholders.
Gender affinity
While it is often theorized that women are more likely to vote for women candidates due to an affinity bias, existing research is mixed on whether this actually occurs. In the United States, several studies reported that women are more likely to support and vote for women than men candidates (Brians, Reference Brians2005; Dolan, Reference Dolan2010; Plutzer and Zipp, Reference Plutzer and Zipp1996). In some instances, it was found that women were even willing to shift political party affiliation in order to vote for a woman candidate (Fox, Reference Fox1997). Other studies show women were no more likely to vote for women candidates than men (McDermott, Reference McDermott2009; King and Matland, Reference King and Matland2003). More recent literature suggests that a shared gender identity between voter and candidate can shape vote choice but only under certain contexts (Badas and Stauffer, Reference Badas and Stauffer2019). In the United States, gender identity can be an important driver of political behaviour, but the relationship is also impacted by partisan affiliation, ideological stances, race and even an evaluation of the capacities and capabilities of a candidate (Marien et al., Reference Marien, Schouteden and Wauters2017). Similarly, it appears that there are differences in affinity bias among women voters at national-level and state-based-level elections (Badas and Stauffer, Reference Badas and Stauffer2019).
In Canada, research into gender affinity in voting is similarly mixed and appears to be weaker than in the United States. Cutler (Reference Cutler2002) found voters were more likely to support a federal leader of the same gender amid other conditions. Yet Cutler and Matthews (Reference Cutler and Matthews2005) found no gender-affinity effect among voting preferences between women and men candidates for the Vancouver mayor, even when controlling for ideological and party-gender overlap. Similarly, when researching the behaviour of “flexible voters,” or those without a strong attachment to a certain political party, Goodyear-Grant and Croskill (Reference Goodyear-Grant and Croskill2011) found that women voters in Canada were no more likely to vote for a woman than man candidate. The authors also highlight how in Westminster-style parliamentary systems, including Australia, New Zealand and the United Kingdom, women are less incentivized to cast votes for women candidates, as their political institutions and practices tend to discourage candidate-based voting in favour of party-based support.
Research documents that focus exclusively on gender can oversimplify complex relations (Dolan, Reference Dolan2019). And the fact that women voters have political and policy preferences that align more closely with positions being taken by women candidates, it becomes difficult to distinguish whether the voting choice is based on gender affinity itself or more ideological and political alignment (Bird et al., Reference Bird, Jackson, Michael McGregor, Moore and Stephenson2016). Despite this nuance in the literature, there is some evidence that in some contexts women may be more inclined to support women candidates over their men counterparts. In these instances, affinity bias can be explained as a social identity process or simply a heuristic that voters use to make easy distinctions between candidates (Badas and Stauffer, Reference Badas and Stauffer2019; Bird et al., Reference Bird, Jackson, Michael McGregor, Moore and Stephenson2016). Thus a candidate's gender may serve as an information cue that draws voters to candidates who are demographically similar to themselves, thereby ensuring descriptive representation.
Despite (or indeed because of) the mixed evidence in support of a gender-affinity bias in candidate support, we explore this phenomenon here without advancing specific theoretical expectations.
Data and Methods
The selection of our cases was motivated by Australia, Canada and the United States’ shared attributes, such as their historical legacies, individualistic tendencies and gender inequality in representation at all levels. Given these countries are market oriented, rather than government oriented, in their policy landscapes, we don't expect significant cross-national differences. The United States, which is the most individualistic of these three countries, may exhibit greater support for these non-quota measures. But Canada and Australia also exhibit strong individualistic norms and policies. Our non-quota gender-mainstreaming strategies capture this sentiment—individuals turning to the market to solve work–family problems (for example, pay raise, buying childcare or housework support). Thus, we note significant between-country differences in our discussion of the findings but do not explicitly hypothesize between-country differences here. Given our country cases were strategically selected to capture similarity across these nations, we expect country differences to be minimal.
To test our hypotheses, we use data from a set of survey-based experiments (Druckman, Reference Druckman2022; Mutz, Reference Mutz2011) conducted in parallel in Australia (N = 8,413), Canada (N = 9,127) and the United States (N = 8,454). Survey data collection was conducted between August 25 and October 12, 2021, by SurveyMonkey using their End Page methodology. Study participants were recruited from the millions of survey-takers completing one of the thousands of user-created surveys on the SurveyMonkey platform every day. After completing a survey on an unrelated topic, samples of respondents from the targeted countries (ascertained by their internet protocol [IP] addresses) were invited to complete another (voluntary, uncompensated) research survey. SurveyMonkey's methodology yields large, demographically representative samples.Footnote 2 Samples obtained using this methodology have been used extensively to study voting behaviour and social and political attitudes (Carson et al., Reference Carson, Gravelle, Phillips, Meese and Ruppanner2023; Chen et al., Reference Chen, Valliant and Elliott2019; Gravelle et al., Reference Gravelle, Phillips, Reifler and Scotto2022; Medeiros and Gravelle, Reference Medeiros and Gravelle2023; Williams et al., Reference Williams, Gravelle and Klar2022). Descriptive statistics for our samples appear in Table A1 in the online appendix.
These online river samples stem from nonprobability sampling procedures. Still, our samples are fit for purpose (compare Baker et al., Reference Baker, Michael Brick, Bates, Battaglia, Couper, Dever, Gile and Tourangeau2013), as our primary inferences relate to the causal effects of our experimental treatments and not the descriptive characteristics of our samples (Druckman, Reference Druckman2022). Experimental political science research has repeatedly found that online nonprobability samples and probability samples yield very similar estimates of treatment effects (Coppock et al., Reference Coppock, Leeper and Mullinix2018; Krupnikov and Levine, Reference Krupnikov and Levine2014; Mullinix et al., Reference Mullinix, Leeper, Druckman and Freese2015). Experimental designs are particularly well suited to studying the effects of candidates’ or politicians’ gender identities, as the experimental manipulation of gender avoids the problems of candidate-selection effects and the intersection of gender and partisanship (among other factors) that arise with observational studies (Dolan and Sanbonmatsu, Reference Dolan, Sanbonmatsu, Lupia, Greene, Kuklinski and Druckman2011).
Our vignette-based experiment (described in Figure 1) is structured around a short, realistic, hypothetical case of a politician's career, highlighting the personal and job challenges faced by the politician and how she or he coped with those challenges. The experiment uses a 2 × 2 factorial design in which the gender of the politician (named Jane or John) and the level of government (either local or federal) are randomized. Immediately following the randomly manipulated vignette, respondents were asked about their support for (or opposition to) providing various financial or practical resources to mitigate role strain: (1) John/Jane should receive a pay raise, (2) John/Jane should get a housekeeping allowance to pay someone else to help him/her with the housework, (3) John/Jane should get a childcare allowance to pay someone else to help with the children, and (4) Congress/the House of Commons/the Federal Parliament (or city council) should hold more meetings online so John/Jane does not have to travel as often to attend committee meetings. These four measures serve as our dependent variables and are scaled from strongly agree (scored +0.5) to strongly disagree (scored −0.5).
Structure of the experimentally manipulated vignette.
Source: Authors
We test our hypotheses by pooling our three country samples and estimating a set of linear models. Our models include indicators for the hypothetical woman politician, Jane, local level of government, and two of the countries, Canada and Australia (the United States serves as the reference category). We also include a set of demographic and attitudinal covariates: gender, age (years logged), educational attainment and political ideology (using a liberal–conservative scale in the United States and left–right in Canada and Australia). A more parsimonious model specification that excludes these covariates (reproduced in Table A2 in the online appendix) yields nearly identical results. An alternative model parameterization using ordinal logit (reproduced in Table A3 in the online appendix) yields results that align closely with the linear models, as do country-specific models (reproduced in Table A4). We elect to display our model results graphically in a coefficient plot and a set of effect plots, though the full model results are again reproduced in Tables A2–A4.
Results
The results from the linear models largely support our theoretical expectations. We consider first the main effects of our politician gender experimental manipulation (H1) and our federal-local manipulation (H2) (see Figure 2). Three of the four non-quota policies receive greater support when the beneficiary is described as a woman (Jane) versus a man (John), specifically a pay raise (b = 0.047, s.e. = 0.005, p < 0.001), housekeeping allowances (b = 0.062, s.e. = 0.005, p < 0.001) and childcare subsidies (b = 0.080, s.e. = 0.005, p < 0.001). These results support our first hypothesis: public support for non-quota gender-mainstreaming policies is higher when the recipients are women compared to men. Still, these effects are modest. Assuming a Canadian woman of average age, education and ideological placement evaluating a federal politician, the expected value for agreement with giving a pay raise to John is −0.106 (on the −0.5 to +0.5 scale), which rises to −0.059 for Jane. Agreement with providing John with a housekeeping allowance has an expected value of −0.140 compared to −0.078 for Jane. Agreement with providing John with a childcare allowance has an expected value of −0.071 while it is 0.010 for Jane. By contrast, there is no significant difference in support for more online meetings between the Jane and John treatments (b = −0.002, s.e. = 0.005, p = 0.652). This translates to expected values of 0.215 for John and 0.213 for Jane—a negligible difference. We can only speculate that the ubiquity of online meetings for many professional occupations brought about by the COVID-19 pandemic has induced widespread acceptance of (or resignation to) online meetings among the mass public.
(Panel A). Coefficient plot: Agreement with non-quota policies.
Notes: OLS point estimates and 95 per cent (α = 0.05) confidence intervals are plotted. Dark grey, circle-shaped point estimates are from the main effects only model specification testing H1 and H2. Light grey, diamond-shaped point estimates are from the interactive (Jane × Woman) model specification testing gender affinity bias.
(Panel B). Coefficient plot: Agreement with non-quota policies.
Notes: OLS point estimates and 95 per cent (α = 0.05) confidence intervals are plotted. Dark grey, circle-shaped point estimates are from the main effects only model specification testing H1 and H2. Light grey, diamond-shaped point estimates are from the interactive (Jane × Woman) model specification testing gender affinity bias.
The results also support our second and final hypothesis, namely that public support for non-quota gender-mainstreaming policies is greater for local (as opposed to national or federal) officeholders. When our hypothetical politician is described as a local politician, as opposed to holding national office, support is higher for a pay raise (b = 0.039, s.e. = 0.005, p < 0.001), housekeeping allowances (b = 0.024, s.e. = 0.005, p < 0.001), childcare subsidies (b = 0.025, s.e. = 0.005, p < 0.001) and more online meetings (b = 0.028, s.e. = 0.005, p < 0.001) (see Figure 2). Again, these effects are modest in size. Assuming a Canadian woman of average age, education and ideological placement evaluating Jane, the expected value for agreement with giving a pay raise to a federal politician is −0.059 (on the −0.5 to +0.5 scale); this rises to −0.019 for a local politician. Agreement with providing a housekeeping allowance to a federal politician has an expected value of −0.078, while it is −0.053 for a local politician. Agreement with providing a childcare allowance to a federal politician has an expected value of 0.010, which rises to 0.035 for a local politician. Endorsement of more online meetings increases from 0.213 when the vignette describes Jane as a federal politician to 0.241 when she is described as a local politician.
We next consider the literature that provides mixed findings on gender-affinity bias. In our analysis, we explore this issue by interacting respondents’ gender identity with our politician's gender in the experimental manipulation (see Figure 2; see also Table A2). In short, we find evidence in support of past research that has found a gender-affinity bias. In our study, women are more likely to endorse non-quota policies when the recipient is Jane than when John is the recipient. The Jane × woman interaction is positive and significant for each policy: a pay raise (b = 0.046, s.e. = 0.010, p < 0.001), a housekeeping allowance (b = 0.036, s.e. = 0.011, p < 0.001), a childcare allowance (b = 0.035, s.e. = 0.011, p < 0.01) and more online meetings (b = 0.019, s.e. = 0.010, p < 0.05).
Understanding these interactions is simpler when translated into expected scores on the −0.5 to +0.5 scale of our dependent variables. Considering, first, agreement with giving a pay raise, and focusing on the estimates for Jane and John when depicted as a federal politician, women participants express stronger agreement with giving Jane a pay raise (−0.047 on the −0.5 to +0.5 scale) than John (−0.117). By contrast, men participants are about as likely to agree with giving a pay raise to Jane (−0.106) as John (−0.117). We obtain similar (though marginally higher) estimates when Jane and John are depicted as local politicians. Again, women participants express stronger agreement with giving Jane a pay raise (−0.008) than John (−0.078), while men are about equally likely to agree with giving a pay raise to Jane (−0.067) and John (−0.078) (see Figure 3).
Agreement with politician receiving a pay raise.
Notes: Predicted values are calculated from the OLS estimates for the interactive model specification (Model 1.2 in the online appendix). Error bars correspond to 83 per cent confidence intervals; non-overlapping confidence intervals thus correspond to a significant difference in predicted values with α = 0.05.
When we look at agreement with providing a housekeeping allowance to a federal politician, women participants again express stronger agreement with providing Jane with such an allowance (−0.069) than John (−0.149). A similar (albeit less pronounced) pattern holds among men participants, with stronger agreement with providing such an allowance to Jane (−0.098) compared to John (−0.142). We again find marginally higher estimates for a housekeeping allowance for a local politician. Women are more in agreement with providing Jane with such an allowance (−0.044) than John (−0.125), as are men (−0.074 for Jane compared to −0.118), though these expected scores present a smaller difference (see Figure 4).
Agreement with politician receiving a housekeeping allowance.
Notes: Predicted values are calculated from the OLS estimates for the interactive model specification (Model 2.2 in the online appendix). Error bars correspond to 83 per cent confidence intervals; non-overlapping confidence intervals thus correspond to a significant difference in predicted values with α = 0.05.
As for agreement with providing a childcare allowance to a federal politician, women participants agree more strongly with providing such an allowance to Jane (0.019) than to John (−0.079). This also holds among men participants, with greater agreement with providing such an allowance to Jane (−0.025) compared to John (−0.089), though the difference is again less pronounced. Agreement with providing a childcare allowance to a local politician is again slightly higher. Women more strongly agree with providing Jane with childcare allowance (0.044) than with providing one to John (−0.054). Men participants also more strongly agree with providing Jane with such an allowance (0.000) than to John (−0.063), though again the difference is less pronounced than what holds among women (see Figure 5).
Agreement with politician receiving a childcare allowance.
Notes: Predicted values are calculated from the OLS estimates for the interactive model specification (Model 3.2 in the online appendix). Error bars correspond to 83 per cent confidence intervals; non-overlapping confidence intervals thus correspond to a significant difference in predicted values with α = 0.05.
Lastly, women participants express marginally stronger agreement with holding more online meetings to accommodate a federal politician when the hypothetical politician is Jane (0.217) compared to John (0.211). By contrast, men participants express slightly less agreement with holding more online meetings to accommodate Jane (0.151) than John (0.163). The same patterns are seen with local politicians. Women participants express slightly stronger agreement with holding more online meetings to accommodate Jane (0.245) than John (0.238); men are slightly less amenable to more online meetings to accommodate Jane (0.178) than John (0.191) (see Figure 6).
Agreement with holding more online meetings.
Notes: Predicted values are calculated from the OLS estimates for the interactive model specification (Model 4.2 in the online appendix). Error bars correspond to 83 per cent confidence intervals; non-overlapping confidence intervals thus correspond to a significant difference in predicted values with α = 0.05.
Considering the covariates in our models for support for the different non-quota policies, Canadians and Australians are less supportive of pay raises than US respondents while not being appreciably different in their support for housekeeping and childcare allowances; they are also somewhat more supportive of the increased use of online meetings. The association between educational attainment and support for non-quota policies varies across the different policy proposals. Higher levels of educational attainment are associated with lower support for pay raises and housekeeping allowances but higher support for more online meetings. Education has no effect on support for childcare allowances. By contrast, older respondents and those who identify as a conservative (or on the political right) are consistently less supportive of such policies (see Figure 2).
Our country-specific models also allow us to examine the relationship between party identification and support for non-quota policies. In the American case, Democrats express substantially stronger support than Republicans for the range of non-quota policies. In Canada, those who identify with the Conservative Party or the People's Party express less support for non-quota policies than Liberal Party identifiers (the reference category). Interestingly, those who identify with the New Democratic Party also express less support for providing pay raises and housekeeping allowances. These results likely indicate that spending on such non-quota policies—that is, spending on politicians—do not align with the public spending priorities of supporters of either left- or right-leaning parties. In the Australian case, there are no consistent differences between those who identify with the Labor Party (the reference category), the Liberal-National Coalition, or the Greens.
The country-specific models do not indicate that there are substantively different effects of our experimental manipulations, nor do we find evidence of heterogeneous treatment effects across countries for either of our experimental manipulations with the pooled data. Further, we had no strong a priori reasons to expect any. There is no evidence that the Jane versus John manipulation exerts a different effect in the United States, Canada, or Australia with respect to support for pay raises (F = 0.124, d.f. = 2, 25980, p = 0.939), housekeeping allowances (F = 3.408, d.f. = 2, 25980, p = 0.182), childcare allowances (F = 4.339, d.f. = 2, 25980, p = 0.115) or online meetings (F = 5.485, d.f. = 2, 25980, p = 0.065). Similarly, the local versus federal manipulation does not differ in its effect across the three countries with respect to support for pay raises (F = 1.850, d.f. = 2, 25980, p = 0.397), housekeeping allowances (F = 0.019, d.f. = 2, 25980, p = 0.990), childcare allowances (F = 4.433, d.f. = 2, 25980, p = 0.109) or online meetings (F = 0.764, d.f. = 2, 25980, p = 0.682). Thus, we are not identifying significant between-country differences for our three selected cases.
Discussion and Conclusion
The gender gap in political representation has endured in many countries, particularly in the three democracies considered here. Despite the implementation of several mechanisms to increase women's representation in national and local governments over time, women still face significant barriers to engage and to continue in a political career. Quotas and direct mentorship are critical policy mechanisms to encourage women to run and be elected to office. Unfortunately, these measures alone are not sufficient to close gender gaps in representation in a timely manner, as evidenced by Berevoescu and Ballington's (Reference Berevoescu and Ballington2021) analysis. Nor are quotas (or gender targets) implemented systematically across all parties in our three country cases—the United States, Canada and Australia. In part, the gender gap remains because women are balancing different work and family demands than men, often with less support (Ryan et al., Reference Ryan, Pini and Brown2005; Carson et al., Reference Carson, Mikolajczak and Ruppanner2021). The roles of mother, wife and elected official are intense, with competing demands in combining all three. Alleviating some of this role strain by implementing resources grounded in gender mainstreaming is another avenue toward attracting and retaining women in local and federal government. Here, we follow gender-mainstreaming literature (Palmieri and Childs, Reference Palmieri and Childs2020; Palmieri, Reference Palmieri2011) and build on others’ calls to investigate how practical and financial supports, identified as critical gender-mainstreaming resources, might incentivize women's political participation (see Jakimow, Reference Jakimow2022), such as pay raises, childcare support and more flexible participation in public office. One answer in local government to the enduring problem of women's underrepresentation is to increase councillor compensation so that women (and men) are not compelled to balance political life with other employment. In the federal sphere, pay increases can enable greater agency to hire in extra resources to deal with competing work and family demands.
Our research design involved two key experimental manipulations: the gender of the politician and their jurisdiction (local or federal government). Our models included a range of covariates, including political orientation, education and age, and our sample sizes were large, providing added robustness to test our two hypotheses. This design provides a novel approach to understanding these questions, especially in the context of three countries that have historically failed to engage women equally, compared to men, in elected politics.
Our results suggest there is generally public goodwill for additional financial and practical solutions that would benefit women's representation in office, especially at the local level. Specifically, we find the greatest public support comes for online meetings, childcare allowances and pay raises. Support is greater for women politicians than their men counterparts, and for those in local government. We also explored gender-affinity bias and find that women are more supportive of women politicians to access these resources, demonstrating an affinity effect. This last finding adds weight to support for gender-affinity effects in a contested literature and is an area for further research. Together, our findings offer pragmatic solutions that are meaningful to improving equal gender representation, given that local government is recognized as an important training ground and pipeline into other political tiers. Thus, in addition to providing a critical service to local communities, increasing women's access to local government assists with increasing women's visibility and name recognition, which can be a successful strategy for running for office at state or federal level. This pipeline should be fortified. Our respondents are not contentious to these gender-mainstreaming resources, and thus our study offers a roadmap for political parties and stakeholders committed to equal gender representation to consider.
A drawback, however, is that the public is only moderately supportive of some measures. When we graph levels of agreement, we see that online meetings overwhelmingly receive significant support. The others—pay raises, childcare allowances and housework support—have less enthusiastic support yet still demonstrate greater support for women than men and, in particular, women in local government. It may be that these measures will cost the taxpayers money and thus their support is lower than for online meetings, which should, theoretically, be free. But it also indicates that there remains a public openness about gender-mainstreaming resources, which have been successfully implemented in other countries (Palmieri, Reference Palmieri2011). This question of varying levels of enthusiasm for non-quota policies is a call for additional research, specifically investigating the reasons for supressed support and possible pathways forward. We also acknowledge other factors beyond role strain, such as structural barriers and discrimination against women politicians and candidates, that may still persist even if efforts to reduce role strain are adopted. That said, a first step forward toward gender parity may be trialling policies to provide resources to reduce role strain—notably childcare supports, mandatory online meetings and pay raises—in local government, in order to build an evidence base to integrate effective gender-mainstreaming solutions across levels of government. Thus, our findings warrant additional research that might include intervention trialling.
This study offers some optimism for women politicians, in that publics are prepared to consider non-quota measures to make gender equality in politics possible. Yet our approach is not without limitation. While our respondents are supportive of offering women more resources than men to manage their work and family demands, irrespective of political orientation, we did not determine who should pay for these resources. We show that, with the exception of Australia, those who identify with right-of-centre political parties in the United States and Canada were less supportive of these policies. It may be that these policies are seen as less palatable if they are at taxpayers’ expense. A second limitation is that we sample from countries with similar cultural, historical and political profiles. These findings may not generalize readily to other countries with different democratic cultures or welfare state regimes. This points to the need for extensions of this research to other political and economic contexts. Further research should test these interventions and public sentiment congruently to see how they impact politician's experiences across these countries and public support for the policies to build a clear evidence base. Finally, subgroup analysis across these countries—notably additional analysis of those with children themselves—would be useful to underscore which segments of the electorate are most supportive of gender-mainstreaming resources.
The key contribution from this study is that the gender gap in politics requires innovative solutions beyond gender quotas if gender parity is to be achieved in a timely manner. We show that across three democracies, the public is not hostile, in principle, to paying women more money or offering non-quota supports such as childcare to help them balance work and family to have a political career; and the public is very supportive of practical measures such as online meetings. Given that women are often penalized for failing to publicly demonstrate they can balance work and family sufficiently in terms of electability, this illustrates why these resources are essential in the first place.
Supplementary material
The supplementary material for this article can be found at https://doi.org/10.1017/S0008423923000720
Acknowledgments
We would like to sincerely thank the anonymous reviewers and CJPS editors for their helpful input and guidance that has improved this article. We would also like to thank the judges with the Representation and Electoral Systems Section of the American Political Science Association who awarded an earlier version of this article the 2023 Leon Weaver Award. We were deeply honoured to receive this award. Finally, we would like to acknowledge the Women's Leadership Institute Australia (WLIA) that provided Professor Andrea Carson with a research fellowship that funded this experimental study.
Open access
