SOCIOECONOMIC RESOURCES

Our focus is on gender differences in responses to the statement that “Sometimes politics is too complicated for a person like me to understand.”Footnote ⁴ As Virginia Sapiro (Reference Sapiro1983, 99) notes, this item “captures very precisely traditional notions of women's relationship to politics: politics is man's business, women are simply not capable of understanding it.” Traditionally, it was considered to be a measure of “internal efficacy,” or “individuals’ self-perception that they are capable of understanding politics and competent enough to participate in political acts such as voting” (Miller, Miller, and Schneider Reference Miller, Miller and Schneider1980, 253). However, validity tests have indicated that it correlates poorly with other measures of internal efficacy (Craig, Niemi, and Silver Reference Craig, Niemi and Silver1990; Niemi, Craig, and Mattei Reference Niemi, Craig and Mattei1991). What is interesting about these tests from our perspective is the conclusion that this item “primarily taps the ‘ability’ or ‘skills’ component of internal efficacy, whereas the remaining items are better able to capture the ‘interest’ component that sometimes develops in the absence of formal skills” (Craig, Niemi, and Silver Reference Craig, Niemi and Silver1990, 295). This makes the persistence of the gender gap all the more surprising: Why have women's gains in terms of educational attainment and labor force participation not done more to enhance their confidence in their political skills?

Carol Christy (Reference Christy1985) was one of the first to highlight the disjuncture between the changes in women's education and workforce participation and the continuing gap in women's and men's confidence in their ability to understand politics. Karen Beckwith's (1986) extensive individual-level analysis confirmed that the gender gap persisted, regardless of education, social class, or occupational status.

These findings pose a fundamental challenge to explanations of gender differences in political engagement that focus on women's lack of socioeconomic resources. This type of explanation has proved its worth in accounting for the gender gap in political activity. On the basis of their wide-ranging analysis, Nancy Burns, Kay Lehman Schlozman, and Sidney Verba (2001, 359–60) concluded that “[t]he emphasis upon socioeconomic resources turned out … to be well placed. Women are, on average, disadvantaged with respect to education, income, and occupational status, attributes long known to be associated with political activity. These deficits are important in accounting for the participation gap.”Footnote ⁵ The implication is that once these gender disparities disappear, women and men will be equally active in politics. It is interesting, however, that the same factors performed rather poorly when it came to explaining gender differences in political interest, political knowledge, and external political efficacy (Verba, Burns, and Schlozman Reference Verba, Burns and Schlozman1997). Similarly, Lonna Rae Atkeson and Ronald Rapoport (2003) found that controlling for socioeconomic resources diminished, but did not eliminate, the gender gap in political attitude expression.

There are certainly reasons to expect socioeconomic resources to matter. Education has been hailed as the great equalizer, and enhanced access to education is credited with opening the doors of opportunity, wealth, and power for disadvantaged groups. Education fosters norms of civic duty and political engagement. Most importantly, for our purposes, it develops the cognitive and information-processing skills that are required to deal with the complexities of politics (Verba, Schlozman, and Brady Reference Verba, Schlozman and Brady1995; Brady, Verba, and Schlozman Reference Brady, Verba and Schlozman1995). It is perplexing that women remain less confident than men of their ability to understand politics, given the increase in their level of education. However, we have to bear in mind that not all women have shared equally in these advances. Women make up a disproportionate number of those living in poverty, and that number has been growing: Since the 1970s, female-headed households have accounted for an increasing proportion of those below the poverty line (Fukuda-Parr Reference Fukuda-Parr1999). The feminization of poverty may mean that women living at the margins are too preoccupied with the daily struggle to put a meal on the table and find the rent money to have much energy to devote to keeping abreast of politics.

It is also surprising that increased participation in the workforce has apparently not been accompanied by a narrowing of the gender gap in self-perceived political understanding. Confinement to the domestic sphere has traditionally been seen as isolating women and limiting their opportunities to discuss politics (Kay et al. Reference Kay, Lambert, Brown and Curtis1987). As full-time homemakers, they may be less aware of gender inequalities and thus less likely to develop a feminist consciousness or to question traditional gender roles (Manza and Brooks Reference Manza and Brooks1998; Welch Reference Welch1977). Accordingly, we would expect entry into the paid workforce to enhance women's confidence in their ability to understand a traditionally male-dominated domain. If working outside the home for pay has failed to enhance women's confidence in their ability to understand politics, the explanation may lie in gendered patterns of employment. Gender continues to negotiate, to some degree, the experiences that women have in the paid workforce. Women's underrepresentation in high-status occupations may limit their opportunities to gain a better understanding of politics and government.

The responsibility of raising children is another sex-specific reality that may affect women's self-perceived ability to understand politics. Sapiro (Reference Sapiro1983) first pointed to the inhibiting effect of motherhood on the perceived complexity of politics. Motherhood may well be more of a constraint now than it was 30 or 40 years ago. More women today are working a “double day,” juggling their domestic responsibilities with full-time employment outside the home. The disproportionate amount of responsibility that many of them continue to bear due to caring for home and children may serve to limit the gains in political understanding that might otherwise have accrued from working outside the home. To their surprise, though, Burns, Schlozman, and Verba (Reference Burns, Schlozman and Verba2001) found that having children at home did not significantly affect women's propensity to be politically active, and “[t]ry as we might, we could find no evidence that an absence of free time handicaps women as citizens” (p. 333). Indeed, mothers of school-age children were actually more apt to participate in politics if they held full-time jobs. Leisure time was simply not a significant factor in explaining political activity for women or men. It is interesting to note, though, that there is one important exception to this pattern: The amount of free time did have a significant effect on people's political knowledge (Verba, Burns, and Schlozman Reference Verba, Burns and Schlozman1997). Acquiring political information requires a commitment of time and energy. Indeed, this is why many people are considered to be rationally ignorant about politics (Downs Reference Downs1957). Developing confidence in one's ability to understand politics may require a similar commitment. To the extent that it does, the demands of parenting may help to explain why some women lack that confidence.

DATA AND METHODS

We use data from the 2000 American National Election Study and the 2004 Canadian Election Study mail-back survey to assess these explanations of the gender gap in self-perceived understanding of politics. In comparing Canada and the United States, we are analyzing the gender gap in self-perceived political understanding in settings that are similar in many ways. The key similarities, for our purposes, relate to the emergence of the women's movement and the sweeping changes that have occurred in women's lives in both countries over the past 40 years. According to Lisa Young (Reference Young2000, 7), “Situating the two movements within the broader universe of feminist movements in advanced industrialized countries, we find the two North American movements more similar to one another than to any other national movements.”Footnote ⁶ We also see women making similar gains in both countries. Between 1971 and 2001, the percentage of Canadian women 15 and older with a university degree increased by 12 points (Statistics Canada 2006). In the United States, the percentage of women age 25 and over with a college degree increased by almost 15 points over a similar 30-year span (U.S. Census Bureau 2007). Both countries have also seen significant increases in women's labor force participation: The percentage of Canadian women age 15 and over in the paid workforce increased by 16 points between 1976 and 2004 (Statistics Canada 2006), while the percentage for American women age 16 and over increased by 15 points between 1970 and 2004 (U.S. Department of Labor Bureau of Labor Statistics 2007).

At the same time, though, the two countries differ in potentially consequential ways. Canadian women who have worked for at least 600 hours in the previous 52 weeks are entitled to receive up to $435 a week in maternity benefits for a maximum of 15 weeks. In addition, parental benefits are payable while they are caring for their newborn (or adopted) child, up to a maximum of a further 35 weeks. The parental benefits can be claimed by one parent or shared between the two partners but cannot exceed a combined maximum of 35 weeks. The benefits are funded through the federal Employment Insurance program to which both employers and employees contribute. In the United States, by contrast, only a handful of states provide maternity benefits (in the form of short-term disability benefits for 10 to 12 weeks), and while the Family and Medical Leave Act of 1993 requires employers with 50 or more employees to provide maternity/parental leave, it does not require that the leave be paid.

The two countries also differ when it comes to affordable child care. Parents in the United States incur, on average, 60% of the cost of child-care services, whereas Canadian parents outside Quebec incur slightly less than 50% (OECD Directorate for Education 2006). In the province of Quebec, parents have access to seven-dollar-a-day day care. The United States has one of the poorest records among countries in the Organization for Economic Cooperation and Development when it comes to providing affordable child care to low-income families (OECD Directorate for Education 2006). These differences may explain why employment rates for women with young children differ in the two countries. In 2004, only 57% of mothers with a child under the age of 3 years worked for pay in the United States, compared with 65% of their Canadian counterparts (Statistics Canada 2006; U.S. Department of Labor Bureau of Labor Statistics 2007). The difference disappears once the youngest child enters grade school. If the demands of motherhood are a drag on women's political comprehension, this should be much more evident in the United States than in Canada.

For each country, we begin with a baseline model that only includes sex. In order to assess whether differences in women's and men's socioeconomic resources explain the gender gap, we add education, income, and employment-related variables to the model, along with controls for age cohort, race, marital status, and children in the home. If socioeconomic and employment-related differences are contributing to the gender gap in self-perceived ability to understand politics, the sex coefficients should shrink when these variables are included. However, a complete assessment of the role of socioeconomic resources requires that we also test whether there are gender differences in the way that these resources work to enhance people's confidence in their capacity for understanding politics (see Burns, Schlozman, and Verba Reference Burns, Schlozman and Verba2001). To do so, we estimate a model that interacts each of the variables with sex. This also enables us to test whether the presence of children in the home has sex-specific effects. Parenting is expected to have much more of a constraining effect on women, especially in the United States; having children may make little difference in men's confidence in their ability to understand politics.

The dependent variable is based on responses to the following agree/disagree items: “Sometimes politics is too complicated for a person like me to understand” (ANES) and “Sometimes politics and government seem so complicated that a person like me can't really understand what's going on” (CES). The response categories (see Appendix) have been recoded 1 for those who disagreed with the statement and 0 for those who did not disagree (including those who neither agreed nor disagreed in the ANES and those who were unsure in the CES)Footnote ⁷. Since the dependent variable is dichotomous, all of the models are estimated using binary logistic regression.

With the exception of the number of children, the independent variables are dummy coded: sex (coded 1 for female), education (two dummy variables, one coded 1 for less than high school and the second coded 1 for university graduates), employment status (coded 1 for working for pay), occupational status (two dummy variables, one coded 1 for managerial and professional occupations and the second coded 1 for lower-status occupations), family income (two dummy variables, one coded 1 for low incomes and the second coded 1 for high incomes)Footnote ⁸, marital status (coded 1 for legally married), age cohort (two dummy variables, one coded 1 for those born between 1943 and 1957 and the second coded 1 for those born between 1958 and 1986)Footnote ⁹, and racial minority (coded 1 for minority).Footnote ¹⁰ The number of children under 18 years of age at home is coded from 0 (no children) to 3 (three or more children).

FINDINGS

Figure 1 tracks the percentage of women and men who reject the statement that politics is sometimes too complicated for a person like themselves to understand. This item was first asked in the 1952 ANES, giving us a lengthy time series that ends in 2000 (when the item ceased to be considered a “core” question). The item has been included in every CES (save for 1980 and 1988) since the first study was conducted in 1965.Footnote ¹¹ The similarity in both the size and the persistence of the gender gaps in self-perceived understanding of politics in the United States and Canada is striking. This similarity is all the more striking in that the overall trends diverge: Canadians appear to have gained greater confidence in their ability to understand politics, whereas Americans show little net change.Footnote ¹² In 1952, the gender gap in the United States was 15 percentage points; almost half a century later, the gap was 16 points. In the intervening years, the gap was never less than 11 points. In Canada, the gap was 15 points in 1965 and 13 points almost 40 years later and, again, the gap was never less than 11 points in all those years. Also striking in the case of the United States is the similarity to the equally persistent gender gap in expressing political attitudes (see Atkeson and Rapoport Reference Atkeson and Rapoport2003).

Figure 1. The gender gap in self-perceived political understanding. Percent confident is the percentage who reject the statement that politics is sometimes too complicated for a person like themselves to understand. (Source: American National Election Studies, 1952 to 2000, and Canadian Election Studies, 1965 to 2004.)

The persistence of the gaps in the face of the profound changes in women's lives during these same periods suggests that a socioeconomic resource model is not going to be very helpful in explaining why women feel less confident than men of their ability to comprehend politics. This does, indeed, prove to be the case (see Table 1). Simply focusing on differences in women's and men's socioeconomic resources does little, if anything, to explain the gender gap in self-perceived political understanding in either country. Adding such key resources as education, income, and labor force participation (along with suitable controls) to the regression model for the United States leaves the coefficient for sex virtually unchanged, as a comparison of the first two columns of Table 1 indicates. The same is true if we substitute measures of occupational status for labor force participation (column 3). In Canada, too, these variables do little to diminish the impact of sex.

Table 1. The impact of socioeconomic resources and parenting on self-perceived ability to understand politics

Note: Estimation is by logistic regression. The column entries are regression coefficients. Robust standard errors are shown in parentheses.

**** p < .001, ***p < .01, **p < .05, *p < .10

Yet socioeconomic resources are clearly important to people's self-perceived ability to understand politics. In both countries, education and income prove to be the critical resources in this regard. People with less than a high school education and/or a low income are significantly less likely to reject the notion that politics is too complicated for them to understand. Conversely, college graduates in the United States are much more likely to disagree with this statement, and so are college graduates and those with higher incomes in Canada. Surprisingly, though, employment-related variables do not have a significant effect in either country. This is the case whether we look simply at the effect of being employed or also take occupational status into account.

The key point is that none of these factors explains the gender gap: If women have less confidence than men in their ability to understand politics, it is not because they are poorer on average or because they have lower levels of educational attainment or occupational status or because they are less likely to be working for pay outside the home. This begs the question of whether women derive the same political benefits as their male counterparts from political resources like education and income and employment. It also begs the question of whether there are other factors that are working to offset the effects of the advances that women have made in higher education and the workplace.

We are particularly interested here in the impact of parenting. Do the demands of caring for children work to diminish women's confidence in their understanding of politics? If they do, we would expect this effect to be much more evident in the United States where maternity leave is typically much less generous and affordable day care is harder to find. It turns out that the number of children under the age of 18 living at home has a significant negative effect on parents’ self-perceived ability to understand politics only in the United States (Table 1). However, we need to determine whether this effect is specific to women, as our argument implies.

In order to determine whether there are sex-specific effects, we added multiplicative interaction terms to the models estimated in Table 1 (see Table 2). As Thomas Brambor, William Roberts Clark, and Matt Golder (2006; see also Braumoeller Reference Braumoeller2004) emphasize, with this type of interaction model, the focus is on the marginal effect of each independent variable for the relevant values of the conditioning variable and not just on the significance or insignificance of the coefficients themselves.Footnote ¹³ If we want to compare the effects of, say, number of children for men and women, the coefficient for number of children gives us the marginal effect for men (that is, when female = 0) and the female*number of children term indicates whether the effect is significantly different for women, but it does not tell us whether the marginal effect of having children is statistically significant for women. It is quite possible for the interaction term to be significant in the absence of a significant marginal effect and vice versa. Accordingly, a proper assessment requires that we calculate the marginal effects for women (that is, when female = 1), along with the corresponding standard errors. These are presented in the lower portion of Table 2 under “marginal effects.”Footnote ¹⁴

Table 2. Sex and the impact of socioeconomic resources and parenting on self-perceived ability to understand politics

Note: Estimation is by logistic regression. The column entries are regression coefficients. The marginal effects for women show the effects of socioeconomic resources and parenting when female takes the value of one. Robust standard errors are shown in parentheses.

**** p < .001, ***p < .01, **p < .05, *p < .10

It is clear that the dampening effect of parenting in the United States is sex-specific. Having children under the age of 18 living at home does not affect men's self-perceived ability to understand politics. For women, on the other hand, the number of children has a significant negative effect: Other things being equal, the probability that an American woman will reject the notion that politics is too complicated to understand is 12 points lower if she has two children at home under the age of 18, compared with a woman who has none.Footnote ¹⁵ In this respect, little seems to have changed for American women over a quarter century: Sapiro (Reference Sapiro1983) found that women who were more tied to the private sphere by the demands of motherhood felt less confident than other women of their ability to understand politics. While parenting clearly has a constraining effect when it comes to American women's perception that they can understand politics, Canadian women are not similarly constrained: The marginal effect falls far below conventional levels of statistical significance.Footnote ¹⁶

The second key finding relates to the resource model. We find no evidence of gender differences in the way that socioeconomic resources work to enhance confidence in people's ability to understand politics. The effects of a college education are essentially the same for women and men in both Canada and the United States. The same is true of affluence in Canada (a high income is not a significant factor for women or men in the United States). It is a lack of resources that has differential effects, at least in the United States. Having less than a high school education or having a low income does not significantly reduce men's confidence in their ability to understand politics; the effects are only significant for women. This is not to imply that having a college education closes the gap between women and men. In the United States, the probability that a female college graduate will reject the suggestion that politics is too complicated to understand is fully 19 points lower than it is for a male college graduate. In Canada, the gap is 14 points. In both countries, the gaps are statistically significant (p < .05).

Finally, the results for race and age cohorts deserve to be highlighted. Controls were included for racial background because race has been linked to a diminished sense of political self-efficacy (Woodly Reference Woodly2006). We might have expected the effects of belonging to a racial minority to be even more consequential for women, given that minority women are doubly marginalized in politics as both women and minorities. However, it turns out that the effects are confined to men in both Canada and the United States.Footnote ¹⁷ The implication is clear: The gender gap in self-perceived ability to understand politics would be wider still if it were not for the negative effects of being a black man in the United States or a minority man in Canada.

The results for age cohorts are also surprising. The shared experience of growing to adulthood since the advent of second-wave feminism could be expected to boost women's political self-confidence. However, age cohort does not have a significant effect for women in either country. Indeed, women in the United States who came of age under the influence of second-wave feminism are, if anything, less likely than older women to feel that they can understand politics (the effect approaches statistical significance p = .11). This is in striking contrast to their male counterparts.

Two of our findings warrant further examination. First, we need to determine whether the sex-specific effect of children in the United States is a reflection of the double day. If it is, the effect of having children under 18 living at home should be greater for women who are in the paid workforce. In order to determine whether this is, in fact, the case, we estimated separate models for women that include multiplicative interaction terms for number of children and employment-related variables.Footnote ¹⁸ The results are shown in Table 3, along with the marginal effects of the number of children at home for women who are employed.

Table 3. The impact of parenting and employment on women's self-perceived ability to understand politics

Note: Estimation is by logistic regression. The column entries are regression coefficients. The marginal effects show the effects of having children when the occupational variables take the value of one. Robust standard errors are shown in parentheses.

**** p < .001, ***p < .01, **p < .05, *p < .10

The results are surprising. The effects of having children at home are actually greatest for women in the United States who are not working for pay outside the home. These full-time homemakers are significantly less likely to feel that they can understand politics. Employment outside the home clearly helps to counter the negative effects of raising children: The impact of children is significantly smaller for women in the paid workforce than it is for their stay-at-home counterparts. In fact, if we classify women according to occupational status, the marginal effect of children is almost zero for women in lower-status occupations; it is only women in professional and managerial occupations who are significantly disadvantaged by having children at home. We cannot determine from the data at hand whether this is because children represent a drain on the time or on the mental energy that might otherwise be available for following politics. Clearly, though, raising children compromises their self-perceived ability to understand politics. If there is a double-day effect in the United States, it is confined to women in professional and managerial positions. There is no evidence of any effect in Canada.

Professional and managerial women should be the most likely to have transcended traditional sex-role socialization. The same should be true of women who have come of age since the emergence of the second-wave feminist movement. Yet, as we saw in Table 2, age cohorts do not have the anticipated effects. Clearly, we need to take a closer look, especially in light of Susan Carroll's (1988) argument that women require psychological independence from traditional sex-role socialization in order to be autonomous political actors. Accordingly, we added measures of gender role conceptionsFootnote ¹⁹ and attitudes about feminismFootnote ²⁰ to the models estimated in Table 3. Since employment-related variables do not have significant effects in Canada, we estimate a single model for Canadian women. In both countries, it turns out that women who strongly reject traditional conceptions of gender roles are also significantly more likely to reject the notion that politics is too complicated to understand (see Table 4). The probability of feeling confident of their ability to understand politics is 9 points lower in the United States and 7 points lower in Canada for women who adhere to a traditional view of gender roles. However, positive attitudes toward feminism do not have a significant effect on women's confidence in their ability to understand politics in either country.

Table 4. The impact of beliefs about gender roles and feminism on women's self-perceived ability to understand politics

Note: Estimation is by logistic regression. The column entries are regression coefficients. Robust standard errors are shown in parentheses.

**** p < .001, ***p < .01, **p < .05, *p < .10

Early exposure to politics is clearly a more important factor. This is evident when we add a measure of the frequency with which politics was discussed at home during the women's childhood.Footnote ²¹ This measure is only available for Canada. As the final column of Table 4 shows, the more frequently politics was discussed, the more likely Canadian women are to be confident of their ability to understand politics. Jeremy Mayer and Heather Schmidt (Reference Mayer and Schmidt2004; see also Rosenthal, Rosenthal, and Jones Reference Rosenthal, Rosenthal and Jones2001) argue that gender is the largest influence on the political engagement of children in junior high school because political socialization teaches boys to be more interested in politics while teaching girls to be more passive.Footnote ²² Boys feel politics belong to them, whereas girls are more likely to see politics as something that boys ought to be more interested in. Our results suggest that early political exposure in the home may help to counteract the effects of female socialization. The probability that a Canadian woman will reject the notion that politics is too complicated to understand is fully 28 points higher if her family often discussed politics when she was growing up, compared with a woman whose family never did, and the gender gap shrinks from 18 points to eight points.

An interesting question to pursue would be whether having a mother who was politically engaged matters more. Psychologists have emphasized the importance of vicariousness as a source of self-efficacy (Bandura Reference Bandura1977): If an individual sees others performing difficult or challenging tasks successfully, that individual may be more inclined to believe that he or she could also perform the tasks successfully. Indeed, Atkeson and Rapoport's (2003) work on political attitude expression suggests that a young girl's mother may be a particularly salient role model.