Introduction
Democracies thrive when their citizens are knowledgeable about politics. As Delli Carpini and Keeter (Reference Carpini, Michael and Keeter1996: 219) argue, “all things being equal, the more informed people are, the better able they are to perform as citizens.” Political knowledge is crucial to developing instrumental rationality and ideological consistency, which contributes to political efficacy (Galston, Reference Galston2001; Willeck and Mendelberg, Reference Willeck and Mendelberg2022). It provides citizens with political power by encouraging government responsiveness and electoral accountability (Fraile and Gomez, Reference Fraile and Gomez2017: 91). When citizens lack political knowledge, democracy suffers. Citizens disengage from politics, which allows elites to govern with minimal oversight and little regard for citizens’ preferences. The result is the gradual erosion of democracy. Amy Lerman’s (Reference Lerman2019: 4) book neatly illustrates some of the consequences of what happens when citizens are uninformed about government responsibilities. Most Americans today, she argues, “associate ‘public’ with ineffective, inefficient, and low-quality services—and conversely, to connect ‘private’ with effective, efficient, and higher-quality provision.” Because these attitudes are highly resistant to change, they have become a “self-fulfilling prophecy,” with many Americans turning away from public services regardless of how effective they actually might be, which in turn homogenizes and weakens the public system, such as when wealthy parents pull their kids out of public schools (Lerman, Reference Lerman2019: 15–16).
A particularly concerning feature of many modern democracies is the presence of a stubbornly persistent political knowledge gap between men and women (Barabas et al., Reference Barabas2014; Dolan and Hansen, Reference Dolan and Hansen2020; Jerit and Barabas, Reference Jerit and Barabas2017). This gap is troubling in that knowledge is an important predictor of political participation and efficacy, and unequal engagement with politics can in turn produce public policies that ignore or are harmful to women. Some studies suggest that this knowledge gap is rooted in individual-level differences with respect to education and resources, with women historically having less education and more family commitments relative to men (Burns et al., Reference Burns, Kay and Verba2001; Dow, Reference Dow2009). Others point to the importance of political and economic contexts, arguing that gendered accessibility (opportunity) and gender-bias signalling (role models) can influence the gender gap (Fraile and Gomez, Reference Fraile and Gomez2017). Finally, some studies argue that political knowledge differences between men and women may simply be a function of how familiar they are with different government programs and services. Stolle and Gidengil (Reference Stolle and Gidengil2010) find that the political knowledge gap shrinks across almost all groups of women when they are asked about their knowledge of government services and benefits.
We build on the existing literature by considering the issue from another perspective: how differences in political knowledge across men and women might be shaped by political institutions, in this case, federalism. By dividing powers across orders of government, federalism demands a great deal of citizens wishing to assign responsibility for government actions (Cutler, Reference Cutler2008; Kennedy et al., Reference Kennedy, Sayers and Alcantara2022). During the early days of the COVID-19 pandemic, for instance, governments across Canada and at all levels imposed significant restrictions to combat the spread of the virus. As time wore on, some Canadians objected to these restrictions and placed the blame solely on Prime Minister Justin Trudeau, ignoring the fact that provincial and municipal governments also implemented significant economic and mobility restrictions (Sayers et al., Reference Sayers, Alcantara and Armstrong2022). One result of this misunderstanding, among others, has been an increase in political polarization (Pennycook et al., Reference Pennycook, McPhetres, Bago and Rand2022).
In the case of men and women and the division of powers, if women are less able to correctly ascribe which level of government does what, then they are more likely to become disaffected and disengaged from politics. It can also erode the quality and substance of political participation, even if there is no gap in the quantity of participation. In a recent paper, Armstrong et al. (Reference Armstrong, Stephenson and Alcantara2023: 7) find that gender, age and education have a significant effect on the ability of citizens to place the parties ideologically from left to right in the same way as experts. Women, for instance, place the Green Party to the right of the Liberal Party, whereas men and experts place the Greens to the left of the Liberals. This incongruence could result in voters choosing parties that are not well-placed to deliver what they expect in terms of ideologically-driven public goods (although see Dassonneville et al., Reference Dassonneville, Nugent, Hooghe and Lau2020).
In short, this article asks: Does federalism exacerbate gender-based differences in political knowledge? We analyze data from the 2020, 2021 and 2022 Democracy Checkup surveys (Harell, Reference Harell2022a; Reference Harell2022b; Reference Harell2023) to explore whether a gap in knowledge of federalism exists in Canada. Our findings suggest that a gap does exist, but also that there is heterogeneity with respect to knowledge about specific levels of government, which conforms partly with existing understandings of how women and men differentially engage with each government (Dolan et al., Reference Dolan, Deckman and Swers2016; Mahon and Collier, Reference Mahon and Collier2016; Mehravar et al., Reference Mehravar, Alcantara and Roy2023).
Political Knowledge and Gender
Political knowledge is typically defined as the extent to which citizens know important facts and bits of information about the political world in which they are embedded (Lupia, Reference Lupia2015). This factual knowledge can be subdivided into three types: textbook knowledge of how governments are structured and operate, surveillance knowledge of current events, and knowledge of national and world histories (Barabas et al., Reference Barabas2014; Jennings, Reference Jennings1996). In many national election studies, such as in Canada and in the United States, surveys ask respondents to correctly identify some combination of national and world leaders, political parties, political institutions and branches of government (Delli Carpini and Keeter, Reference Carpini, Michael and Keeter1996; Lupia, Reference Lupia2015; Stephenson et al., Reference Stephenson2021). Those who correctly answer these questions are thought to possess more political knowledge than those that do not. Most studies have found that “the public’s knowledge of institutions and processes is significantly higher than its knowledge of people and policies, perhaps because the former are more stable over time and require less monitoring” (Galston. Reference Galston2001: 221).
Political knowledge, whether measured directly through factual questions or less commonly through proxies such as education, has been found to have a strong effect on political behaviour. Singh and Roy (Reference Singh and Roy2014: 97–98), for instance, find that “more knowledgeable individuals” (measured in terms of their answers to “trivia-type political questions”) are more likely to access “a higher quantity of information and higher quality information,” and that these voters “are more likely to cast a vote for the most ideologically proximate candidate.” Studies using education have found similar, although mixed, results, especially when education is measured in terms of quantity. The effect of education is much clearer when the focus is on the quality of education received (for example, pedagogical approach), pre-adult socialization experiences, and the centrality of social networks (Persson, Reference Persson2015; Willeck and Mendelberg, Reference Willeck and Mendelberg2022).
It is important to recognize that knowledge of politics is unevenly distributed across the electorate (Armstrong et al., Reference Armstrong, Stephenson and Alcantara2023). As mentioned earlier, a large body of research has found a persistent gender gap in terms of political knowledge in democratic societies. Initial research suggested much of this gap can be explained by the gendered and problematic character of question wording and topic selection (Lizotte and Sidman, Reference Lizotte and Sidman2009; Mondak and Anderson, Reference Mondak and Anderson2004). In a similar vein, others have found that men are more likely to guess rather than choose “don’t know,” calling into question studies that coded non-responses as incorrect; it can be problematic to compare groups that have identical levels of knowledge without considering the possibility that each group may have different propensities to guess (Dolan and Hansen, Reference Dolan and Hansen2020; Fortin-Rittberger, Reference Fortin-Rittberger2020). Addressing this concern effectively is extremely challenging with survey data but recognizing it as an issue is fundamental given we cannot know, for certain, how accurately these data estimate the knowledge gap.
Moving beyond question wording and coding decisions, most explanations for the gendered knowledge gap tend to focus on three elements: ability (for example, the skills to acquire knowledge), opportunity/resources (for example, the availability of knowledge) and motivation (for example, political interest) (Barabas et al., Reference Barabas2014; Delli Carpini and Keeter, Reference Carpini, Michael and Keeter1996; Jerit and Barabas, Reference Jerit and Barabas2017). Common indicators used to measure these individual-level factors include attention paid to media, age, education, income, religiosity, race, children at home, standards of living, urban/rural place of residence, and political interest (Fortin-Rittberger, Reference Fortin-Rittberger2016; Fraile and Gomez, Reference Fraile and Gomez2017: 102). One recent study of political knowledge and the gender gap in Latin America found that among these factors, only education, urbanization, and race had a strong effect on reducing the knowledge gap between men and women (Fraile and Gomez, Reference Fraile and Gomez2017: 102).
There is also evidence that a gender gap in political interest exists with respect to motivation: women are likely to be more politically informed about specific issues (Verba et al., Reference Verba, Burns and Lehman Schlozman1997; Stolle and Gidengil, Reference Stolle and Gidengil2010), perhaps due to their propensity to access and use related government services and benefits. This is relevant to our analysis because in federal systems, different orders of government are assigned discrete responsibilities and resources (Sayers et al., Reference Sayers, Alcantara and Armstrong2022). Few studies (Fraile and Gomez, Reference Fraile and Gomez2017: 102) consider how the pattern of policy jurisdiction embedded in political institutions might gender political knowledge. We tackle this topic by focusing on knowledge of the distribution of responsibilities across the multiple governments in a federal system. If interest in specific policy areas varies and different governments are responsible for different policies, does it follow that interest in and therefore knowledge of politics is a function of the institutional division of responsibilities? In the Canadian case, if political interest is sensitive to the distribution of responsibilities across municipal, provincial and federal governments, it could be a source of variation in knowledge of politics between men and women. If so, this is prima facie an argument for how political institutions may shape gendered understandings of politics.
Towards a Theory of Gendered Political Knowledge: Proximity and Government Usage
Research has found that women are much more interested in social issues whereas men are preoccupied with the economy (Gidengil, Reference Gidengil1995). These preferences manifest in how men and women participate in politics: Dolan et al. (Reference Dolan, Deckman and Swers2016) find that women are more engaged with proximate forms of politics such as schoolboards and city councils. Verba et al. (Reference Verba, Burns and Lehman Schlozman1997) also find that of 10 political knowledge items, men are more knowledgeable than women on all but one, the head of the local public school system, and that despite there being a noticeable gender gap for interest in national politics, the gap disappears in relation to local politics. Similarly, Mehravar et al. (Reference Mehravar, Alcantara and Roy2023) show that women are more likely than men to donate to Canadian provincial parties relative to federal parties, perhaps because federal and provincial governments have different policy responsibilities and priorities. There is also evidence to suggest that women may exploit features of a federal system to their benefit. For example, Mahon and Collier (Reference Mahon and Collier2016) find that women-led activism for childcare support is typically directed at provincial and municipal governments rather than the federal government due to the perception that the federal government is less engaged in this area, and that provincial and municipal governments are more accessible. Finally, regarding the gender gap in information, Stolle and Gidengil (Reference Stolle and Gidengil2010) find that the gender gap disappears or reverses for questions about government services and programs in data from a sample of Canadians in Toronto and Montreal. Their results suggest that women tend to know more about the politics that matter to their everyday lives to the extent that they access programs and services related to those areas.
Translating these findings into the institutional arena requires understanding the specifics of the federal division of authority. In Canada, the provinces are largely responsible for social policies whereas the federal government is significantly more engaged with the overall economy. This division is complicated by the transfer of monies from the federal government, which in general raises more money than it needs to cover its core responsibilities, to the provinces, which usually confront the opposite situation (Sayers et al., Reference Sayers, Alcantara and Armstrong2022). This federal transfer process provides the federal government with some limited opportunities to influence provincial policymaking. The federal government is also broadly dominant in macro-economic policy, although provinces have key roles in such areas as tax rates and securities regulation. Provincial governments remain overwhelmingly dominant across the central social policy areas we are interested in here, namely health care, education and social services. Together these policy areas account for approximately 75 per cent of provincial budgets (Prince, Reference Prince and Dunn2016: 475). In sum, the provinces deliver most social services, often considered more interesting to women, while the federal government plays a central role in economic policy, seen by some as of more interest to men (Gidengil, Reference Gidengil1995).
Although the specific powers of municipal governments are not constitutionally enumerated, city councils are responsible for many policies that are essentially subsets of the major provincial responsibilities. Municipal responsibilities shape the daily lives of voters, such as health authorities, transit, sewage, water, police services and so forth. Research has shown that women participate more at the municipal level, partly due to an interest in community issues (Dolan et al., Reference Dolan, Deckman and Swers2016), and that there is no gender gap in local-level political interest (Verba et al., Reference Verba, Burns and Lehman Schlozman1997).
These findings lead to our main expectation: that we will observe differences in the size of the knowledge gap between men and women by level of government. We expect this gap to be smaller for knowledge of lower levels of government relative to higher ones because women are more likely to take a stronger interest in the governments that have jurisdiction over the social goods and services that they are most likely to interact with on a daily basis (Dolan et al., Reference Dolan, Deckman and Swers2016; Gidengil, Reference Gidengil1995; Mahon and Collier, Reference Mahon and Collier2016; Mehravar et al., Reference Mehravar, Alcantara and Roy2023; Verba et al., Reference Verba, Burns and Lehman Schlozman1997). We would expect, for instance, that women are more likely to be able to correctly identify the level of government that has jurisdiction over public transit because women are more likely to own and use a bus pass relative to men (Vance and Peistrup, Reference Vance and Peistrup2012). Similarly, research has found that women tend to spend increases in their personal income on household living expenses, relative to men, which suggests they may be more attuned to water and wastewater policy, which directly impacts household costs and quality of life (Pahl, Reference Pahl1995). As well, Lupia (Reference Lupia2015) suggests that political knowledge levels are likely tied to whether they are relevant to an important objective or task. Women are more likely to be knowledgeable about provincial and municipal governments because this knowledge is crucial to successfully accessing important government services and influencing government policy related to those services. Formally, we expect:
H1: The gap between men and women will be smaller for knowledge of provincial government responsibilities compared to federal government responsibilities.
H2: The gap between men and women will be smaller for knowledge of municipal government responsibilities compared to federal government responsibilities.
Data and Methods
We test our expectations using Democracy Checkup survey data from 2020, 2021 and 2022 (Harell, Reference Harell2022a; Reference Harell2022b; Reference Harell2023). These national online surveys were conducted using samples from the Leger Opinion Panel and delivered through Qualtrics. Survey weights are used in our analyses to improve the representativeness of the data. Our dependent variables are measures of political information drawn from questions that asked about the responsibility of different levels of government for specific policies: employment insurance (E.I.) (a federal responsibility), defence policy (federal), healthcare (provincial), primary and secondary education (provincial), public transit (municipal) and sewage and water (municipal).Footnote 1 Respondents were asked: “which level of government is primarily responsible for the following policy areas?” (see Table A1 in the online appendix).
We initially attempted to measure knowledge using composite scores for each level of government, aggregating the six knowledge items into three separate variables. However, factor analysis and reliability tests revealed that the items did not load well onto distinct factors, indicating they do not constitute reliable scales for political knowledge even within specific government levels (Figure A1 and A2). This finding aligns with our theoretical expectation that knowledge varies not just by government level but also by policy area. As such, aggregating knowledge into level-specific scales would obscure meaningful differences in the nature and salience of the individual policy areas. Focusing on individual items also allows us to provide a more nuanced understanding of how gender differences manifest across these domains while ensuring greater empirical reliability.
For each knowledge item, correct responses are coded as 1. For incorrect responses, we employ two coding approaches: a conservative approach, where “don’t know” responses (more likely to be given by women) are excluded, and a more liberal approach, where “don’t know” is treated as uninformed and coded as 0. This latter coding is likely to expand any gender gap because more women will be categorized as incorrect than men when the “don’t know” responses are re-coded. Given the gendered challenge of understanding the motivation behind answering “don’t know,” and the empirical reality of its distribution, we contend that the conservative approach provides a more realistic look at the actual gender gap since incorrect responses are not over-inflated (see Dolan and Hansen, Reference Dolan and Hansen2020; Fortin-Rittberger, Reference Fortin-Rittberger2020). This dual coding strategy helps to account for varying levels of political awareness and allows us to assess the robustness of our findings with different assumptions about “don’t know” responses, but we acknowledge that it is not perfect, and the existence of “don’t know” responses is a limitation of our study.Footnote 2
Our main independent variable is a dummy variable that indicates whether a respondent identifies as a man (reference: woman).Footnote 3 This variable provides a measure of the gender gap in each type of political information. Our control variables draw upon the existing literature and are sociodemographic measures of ability and resources: age, education, community size, ethnicity, whether one is married or cohabitating, and employment status.Footnote 4 Although including measures of motivation such as political interest and news attention has been done in other research (Fraile and Gomez, Reference Fraile and Gomez2017), such variables are likely also to be influenced by gender, our key variable of interest. We therefore limit our control variables to sociodemographics only, to avoid underestimating or biasing our results.Footnote 5
We use logistic regression models to estimate the effect of gender while controlling for relevant sociodemographic and attitudinal factors. To account for unobserved heterogeneity across the different survey waves used in this analysis, we included fixed effects for survey year (2020, 2021, 2022) in all models.
Results
We begin by providing some descriptive information about the initial distribution of all responses for each political knowledge question (including “don’t knows”), disaggregated by gender. These distributions are presented in Table A2 in the online appendix. Across every individual question, men are more likely to provide correct answers, while women are more likely to provide either incorrect answers or “don’t know” responses, confirming the expectations of the literature. The magnitude of the difference in “don’t knows” is substantial—at least double the percentage of women than men responded with uncertainty. This distribution confirms that when “don’t know” responses are coded as incorrect, it artificially inflates the proportion of uninformed women; we are therefore more comfortable interpreting the results using our more conservative approach to coding that excludes “don’t knows.”Footnote 6
Of particular importance for our research question here, however, is that the extent of the difference in responses varies depending on the policy area. Because there is no reason to expect that the tendency for men to guess would vary by policy, we interpret the pattern of differences as reflecting something about the policy areas themselves, and hence our interest in the impact of federalism on gendered political knowledge. Figure 1 provides a visual representation of the gender gap in reported correct responses for each policy area, with differences between men and women plotted alongside their confidence intervals. The figure highlights that the largest gender gap is for E.I., where men are 11.68 percentage points more likely to provide correct answers than women. This gap is followed by healthcare, where the gap is 7.94 percentage points. In contrast, smaller gaps are observed for sewage and water (5.52 percentage points), primary and secondary education (5.36 percentage points), and defence policy (5.36 percentage points). The smallest gap is for public transit, where the difference is just 4.2 percentage points.
Gender Gap in Political Knowledge Across Policy Areas.
These results provide initial support for our expectations, suggesting that the gender gap in political knowledge decreases as we move from federal to provincial, or federal to municipal, levels of government. However, it is also evident that there is substantial variation within each order of government, depending on the policy area. For instance, the gap in knowledge of E.I. is significantly larger compared to defence, even though both are federal responsibilities. Similarly, within the provincial and municipal levels, the healthcare and sewage and water knowledge gaps are larger than those for education and transit, respectively. These results suggest that while the overall pattern supports our hypotheses, the nature of the policy area itself—its salience, complexity and relevance to daily life—also influences the magnitude of the gender gap.
Tables A3 and A4 in the online appendix present the results from logistic regression models estimating the gender gap for each political knowledge item individually. These models assess whether men are more likely than women to correctly identify the government responsible for each policy area, controlling for sociodemographic factors.Footnote 7 We present a summary of our findings in Figure 2, which illustrates the average marginal effect of gender (men vs. women) on the likelihood of correctly identifying the responsible government for each policy area, with confidence intervals to indicate the reliability of these estimates. Figure 2 is based on the model that excludes “don’t know” responses (the corresponding figure for the model where “don’t know” is treated as uninformed is available as Figure A3 in the appendixFootnote 8). The results depicted in Figure 2 highlight the presence of a gender gap across different knowledge items, with men consistently showing a higher likelihood of providing correct answers compared to women. The magnitude of the gap, however, varies by policy area and the corresponding government. For federal responsibilities, the gender gap is particularly pronounced for E.I., with men over 6 percentage points more likely to answer correctly than women. This result may reflect the fact that men were more likely to access E.I. in Canada relative to women in 2022–2023 (Canada 2024: Ch. 2). The gap for defence policy is much smaller at 1.2 percentage points, but it is still statistically significant, suggesting a relatively modest but consistent advantage for men.
Average Gender Gap Across Knowledge Items by Level of Government.
When we compare these results to political knowledge of provincial responsibilities, we find a gender gap of 4.24 percentage points for healthcare and 2.1 percentage points for education, both of which are smaller relative to E.I. but larger than the gap observed for national defence. For municipalities, the gender gap is less pronounced, and the confidence intervals provide insight into the uncertainty of these estimates. The gender gap for public transit is small, at 1.4 percentage points. For sewage and water, the gap is 2.15 percentage points, indicating a modest difference between men and women.
Taken together, these results offer mixed support for our expectations. If we exclude national defence, the results for the other knowledge items follow expectations. The gender gap is larger for E.I. (federal) than for provincial policies (healthcare and education), and municipal policies (transit and sewage/water). These findings also align with prior research, suggesting that women tend to be more informed and engaged with local government (Dolan et al., Reference Dolan, Deckman and Swers2016; Mehravar et al., Reference Mehravar, Alcantara and Roy2023; Stolle and Gidengil, Reference Stolle and Gidengil2010; Verba et al., Reference Verba, Burns and Lehman Schlozman1997). Yet the results for national defence do not fit this pattern.
Given these mixed results, we conducted a series of formal pairwise comparisons, the results of which are presented in Table 1. Of relevance for our hypotheses are the comparisons between defence and E.I. and the provincial and municipal items, but the other comparisons provide insight into the nature of policy-related political knowledge. The significant pairwise comparisons indicate that employment insurance stands out the most from all the other items, with gender gaps significantly different from both provincial and municipal knowledge items as well as federal defense policy. However, in keeping with the results above, no significant differences are found between defence policy and several other provincial or municipal responsibilities (except healthcare, which was better known). Interestingly, no significant differences are found between the gaps for most provincial and municipal items, which further underscores the similarity in the magnitude of the gender gaps across these governments.
Pairwise Comparisons of Gender Gaps Across Knowledge Items
Note: Comparisons based on Figure 1.
Overall, these results provide mixed support for our expectations, highlighting a more pronounced gender gap for one federal responsibility—E.I.—compared to some provincial and municipal ones—healthcare, education and transit. This nuanced pattern complicates a straightforward conclusion, indicating that gender disparities in political knowledge are not uniformly distributed across the three types of government. We nonetheless think these results offer some preliminary support for the idea that women are more likely to be knowledgeable about the programs, services and infrastructure that they use daily, and so future research in this area is warranted. We feel additionally confident in this conclusion given the results of our alternative dependent variable and model specifications (Table A4 in the online appendix). When “don’t know” responses are coded as incorrect, we know that this disadvantages women who are less likely to guess their answer; the gender coefficients tend to get larger in those models. While we cannot overcome the inherent challenges of using survey data to assess political knowledge, the consistency of the results suggests that the gap, and its patterns across government types—municipal, provincial, and national—is an important political phenomenon to be studied.
Regarding the curious findings for national defence, it is possible that the small gap could be the result of men also being subject to the “use it, know it” mechanism that we attribute to women. The small gender gap on this item could reflect the fact that neither gender interacts with that policy area on a regular basis. In light of the lack of significant differences between the gaps for most provincial and municipal items, Guppy et al. (Reference Guppy, Sakumoto and Wilkes2019) have found that the “gendered division of household labor in Canada” has become much smaller over time, and so perhaps the political knowledge advantage that women may have had in the past due to their primary responsibilities over the household has adjusted in response to the rebalancing of the division of household labour. These results would be broadly consistent with Lupia (Reference Lupia2015) who argues that knowledge is more prevalent when it is personally relevant.
Conclusion
Our case study of Canada suggests that the distribution of policymaking power across governments in federations is associated with variations in the gender gap in political knowledge. By assigning distinct baskets of functions to different governments, federalism exposes the responsibilities of each and incentivizes them to compete for and explain their jurisdictional authority to voters. Given the documented variation in political interest between men and women, governments in a federation present distinctive appeals to each gender (Haussman et al., Reference Haussman, Sawer and Vickers2016; Mehravar et al. Reference Mehravar, Alcantara and Roy2023; Vickers, Reference Vickers2010). This phenomenon shapes the willingness of men and women to commit resources to the acquisition of political knowledge, resulting in a federalized gender gap. In other words, the decomposition of political authority in federations reveals the interaction of institutional and behavioural drivers of political interest and knowledge acquisition, which can help disentangle the supply of and the demand for political knowledge, and in turn the gendered knowledge gap. While our findings are supportive of this perspective, they are not definitive; further research, including survey design and methodological techniques that could overcome gendered patterns of survey responses, is needed to identify the causal mechanisms underlying this gap.
Supplementary material
To view supplementary material for this article, please visit https://doi.org/10.1017/S0008423925000137
Open access
