Gender role beliefs and attitudes toward abortion: a cross-national exploration.
At the level of elite discourse, one important aspect of the abortion issue is the effect that reproductive freedom (or its absence) has on gender equality. Proponents of legal abortion often argue that control over one's fertility is essential for full gender equality (see, for example, Luker, 1985; McDonagh, 1996, Feree, 2003; Colker, 1989). By contrast, limits on the availability of abortion are considered important in order to maintain the status and importance of traditional gender based divisions of labor (Luker, 1985). However, a number of empirical studies of public opinion have shown that attitudes toward egalitarian or traditional gender roles are weak, inconsistent, or insignificant predictors of abortion attitudes (Cook et al., 1992; Jelen, 2014; Lynxwiler and Gay, 1996; Stricker and Danigelis, 2002; Bolzendahl and Mays, 2004).
This set of findings is important, because there is ample reason to believe that public opinion on abortion may have an effect on public policy. Abortion is an "easy" issue (Carmines and Stimson, 1980) on which mass publics are likely to have coherent opinions (Converse and Markus, 1979). Indeed, Killian and Wilcox (2008) have shown that abortion is among those rare issues which, in the United States, can alter individual partisanship. Rossi and Triunto (2012) have shown that public opinion in Uruguay was related to the liberalization of that country's abortion laws, while Jelen and Bradley (2014) have shown that gender role attitudes tend to be weak predictors of abortion attitudes in nations whose abortion laws are highly restrictive.
The purpose of this exploratory analysis is to investigate the sources of this disjunction between elite-level discourse and mass opinion. Why should a central element of the discourse of scholars, journalists, and political activists appear to have such a peripheral impact on the attitudes of ordinary citizens?
This study will address two general hypotheses:
[H.sub.1]: The relationships between gender role attitudes and attitudes toward abortion will be reduced by a lack of cognitive sophistication on the part of some respondents.
It has long been known (Converse, 1964; Zaller, 1992) that political affairs generally are less salient to ordinary citizens than to political elites, and that the level of factual knowledge, consistency, and sophistication exhibited by mass publics varies enormously. Put simply, the cognitive connection between gender role attitudes and support for or opposition to legal abortion may simply not be apparent to many people to whom the issue does not seem particularly important.
A second hypothesis suggests that the effect of gender attitudes on abortion opinion may be contingent on other considerations:
[H.sub.2]: The relationships between gender role attitudes and attitudes toward abortion will be reduced by respondent characteristics which produce cognitive cross-pressures.
That is, even among respondents who see the relevance of access to abortion to female equality, support for legal abortion among gender role egalitarians may be mitigated or superseded by other variables. For example, Luker (1985) has shown that, among women who engage in overt political activity on the abortion issue, there is an important difference between full-time homemakers and women who participate in the paid labor force. Thus, a housewife who values gender equality may come to oppose abortion because of the status she attaches to her occupation. Similarly, many studies (see Jelen et al., 1994) have shown that, in the aggregate, women are more religious than men. Since religiosity is a well-known predictor of abortion attitudes (Adamczyk, 2013), personal religiosity may reduce the influence of attitudes toward the appropriate social roles of women on attitudes toward abortion.
The approach taken in this study is one of large-N, cross national comparison. This is not to suggest that variations in diverse national politics settings cannot, or do not, have an effect on either abortion attitudes or abortion policy. Indeed, in earlier work, I have shown that variations in abortion attitudes (including variations in the relationships between gender role attitudes and beliefs about abortion) are empirically related to national differences in abortion policy. Specifically, in settings in which gender role attitudes are not significantly related to abortion attitudes (Poland and El Salvador) abortion policy is substantially more restrictive than in nations in which these two sets of attitudes are empirically related (Mexico, Uruguay, and the Czech Republic; see Jelen and Bradley, 2014). However, even in countries in which abortion policies are relatively permissive, and in which abortion and gender role attitudes are significantly related (such as the Czech Republic and Uruguay), the effects of gender role attitudes on attitudes toward abortion is considerably weaker than the effects of other such variables, such as attitudes toward sexual morality or euthanasia. Moreover, the relationships between abortion attitudes and these attitudinal predictors are remarkably stable across religious traditions, and the effects of gender role attitudes are substantially weaker (Jelen, 2014). The main differences among abortion attitudes across diverse religious traditions appear to be based on different marginal distributions of predictor variables, and not on the relationships between abortion attitudes and beliefs about sexual morality, gender role traditionalism, or the sanctity of human life.
Data and Method
Data for this study were taken from the World Values Study for 1999-2007. (2) The dependent variable is a 10 point scale, in which respondents are asked to place themselves on a continuum, on which 1 represents a belief that abortion is "never justified," and 10 indicates a stance in which abortion is "always justified." This is a scale, on which respondents are ask to place themselves. No cues as to the meaning of intermediate responses were provided.
The most important independent variables are measures of gender role attitudes. To my knowledge, there is no consensus on the correct measurement of attitudes toward appropriate gender roles (or, more colloquially, "feminism"). The analyses presented here are based on four Likert items from the WVS:
Spouse Income: Both the husband and wife should contribute to household income (coding reversed).
Working Mother: A working mother can establish just as warm and secure a relationship with her children as a mother who does not work (coding reversed).
University Gender: A university education is more important for a boy than for a girl.
Men Better Leaders: On the whole, men make better political leaders than women do.
These items would seem to exhibit a certain face validity, and all are coded so that low values indicate more "traditionalist" responses.
A factor analysis of these items reveals a two-factor solution, in which the "spouse income" and working mother" items load most heavily on one factor, and the "university gender" and "men better leaders" items have the strongest loading on the second factor. This two-factor solution is quite robust, and emerges when separate analyses are estimated for Roman Catholics Protestants, Muslims, Hindus, Buddhists, and adherents of Eastern Orthodoxy (see Jelen, 2014).
Based on these results, two indices of gender role attitudes were computed: A measure of attitudes toward "public" gender roles was measured by taking the mean of the "university gender" and "men better leaders" items (gamma = .507), while a measure of "private" gender role attitudes was computed by calculating the mean of the "working mother" and "spouse income" Likert items (gamma = .332). On their face, these two measures would seem to tap aspects of expectations of appropriate gender roles in the public spheres of business and industry, and comparable expectations about the optimal role of females in family settings, respectively. For purposes of comparison, a combined gender role index was based on the mean of all four items (Cronbach's alpha = .448).
While these indices would seem to have a certain face validity, these results indicate that the four gender role items exhibit limited construct validity. Whether the four gender role items included in the WVS are considered as a unidimensional scale, or as two relatively independent measures, the items are empirically related, but do not form particular neat scales. These somewhat messy results may indicate poor measurement quality from a methodological standpoint, or, more substantively, may reflect weakly organized mass attitudes.
As hypothesized, the analysis includes two sets of conditional variables, which may affect the relationships between measures of attitudes toward appropriate gender roles and abortion. One set of such variables deals with the level of cognitive sophistication or crystallization of gender role attitudes. For some respondents, gender role expectations may be highly salient and central to individual belief systems, while others might regard such matters as highly peripheral to their more immediate concerns (Zaller, 1992). The first of these is the simply the respondent's level of formal education. The second is a measure of constraint among the four gender role items, which is calculated by computing the standard deviation of these items, after recoding each Likert scale to a common direction. The sample is divided into thirds, with "high constraint" respondents coded as those in the lowest third of the distribution of the standard deviation variable, "medium" respondents as those in the middle third of the distribution, and "low" constraint respondents defined as respondents in the upper third of the standard deviation scale.
The second hypothesis posits the possibility of psychological cross-pressures as inhibitors of cognitive connections between attitudes about gender roles and abortion. Although it is possible to conceive of an indefinite number of such variables, two are considered here: Personal religiosity (operationalized as frequency of private prayer) (3) and female labor force participation. The latter variable is operationalized by a dummy variable, in which respondents can self-identify as "housewives." Self-identified homemakers are contrasted with respondents who participate in the paid labor force (either full or part time) or are retired. Several analysts have suggested that there are important differences between homemakers and women who are part of the paid labor force (see especially Cook et al., 1992; Luker, 1985).
For purposes of comparison, the effects of attitudes toward sexual morality and the sanctity of human life are also considered. The sexual morality measure is the mean of two 10 point scales (1 = never justified; 10 = always justified) for homosexuality and prostitution (gamma = .738). (4) The assumption here is that these items tap a more general attitude toward sexual activity outside of marriage (Jelen and Bradley, 2014; Jelen, 2014; Cook et al., 1992; Tuman et al., 2014). "Respect for life" is operationalized as attitudes toward euthanasia, again measured on a 10 point scale. While this measure is controversial, some analysts have argued that attitude toward euthanasis represents a more general respect for the sanctity of human life (see especially Cook et al., 1992; Jelen and Bradley, 2014; Bolzendahl and Mays, 2004).
To what extent are attitudes toward appropriate gender roles related to mass attitudes on abortion? A preliminary set of findings is presented in Table 2, which simply reports the zero-order product moment correlations between responses to the abortion scale and other predictor variables.
Of primary interest are the correlations between the gender role attitudes and the abortion scale. This table contains all three gender role indices, as well as the individual items which comprise the gender role scales. As these data show, all measures of gender role attitudes are statistically significant, and have the expected sign. Given the large N in the WVS, the level of statistical significance is not surprising. Perhaps surprisingly (at least to me), the "public" gender role scale, and its component Likert items, are slightly stronger predictors of abortion opinion than are the corresponding "private" gender role scale and the items on which this scale is based.
Consistent with the results of studies of opinion in the United States, the relationships between these variables and abortion attitudes are, at best, of moderate magnitude. To provide a sense of the relatively low magnitude of these relationships, all of the correlations between measures of gender role attitudes and abortion attitudes are much weaker than the relationships between the abortion scale and the measures of sexual morality attitudes and attitudes toward euthanasia. The pattern of these relationships is not altered appreciably when multivariate models of abortion attitudes are included (Tuman et al., 2014; Jelen and Bradley, 2014; Jelen, 2014).
[H.sub.1] suggests that these weak relationships may be attributable to a lack of cognitive sophistication on the part of some respondents. That is, it is possible (indeed likely) that many respondents simply do not perceive a connection between egalitarian attitudes about the appropriate political, economic, or familial roles of women and support for legal abortion. This possibility is addressed in Table 3, which contains separate correlations between public and private gender role scales and the abortion measure for differing levels of attitudinal consistency across the individual gender role items. (5) Following Converse (1964), high levels of attitudinal consistency are considered evidence of relatively high levels of belief sophistication.
The results presented in Table 3 suggest that [H.sub.1] is confirmed, but to a very limited extent. Both the public and private gender role scales exhibit stronger correlations for respondents with higher levels of constraint. However, both the absolute magnitudes of the relationships, as well as the differences between respondents at differing levels of attitudinal consistency, are quite small. Substantively, this pattern suggests that, even among many respondents who are consistent gender role "egalitarians" or "traditionalists," the cognitive connection between gender role attitudes and beliefs about the acceptability of legal abortion are elusive.
The effects of education as an alternative measure of sophistication, and those associated with possible sources of cross-pressures, are considered in Table 4. [H.sub.2] suggests that respondents with egalitarian gender role attitudes may hold relatively restrictive attitudes about abortion, due to the effects of cross-pressures, such as economic status or religiosity, and this hypothesis is addressed in Table 4.
This table presents the results of three Ordinary Least Squares (OLS) regression models. The first simply contains estimates of the effects of the private and public gender role scales on abortion attitudes. The second is a traditional OLS model, in which the effects of education, homemaker status, and religiosity are "controlled," while the third contains interaction terms between the gender role scales and each exogenous variable. All three models contain controls for respondent religious affiliation. (6) The standardized regression coefficients (beta) are also included, in order to provide a clearer sense of the (limited) magnitude of the relationships involved.
Not surprisingly, Model 1 simply reiterates the results of the bivariate analyses reported earlier. Both gender role scales are significantly, if weakly, related to abortion attitudes, and again, counter intuitively, the "public" index is a stronger predictor of abortion attitudes than is the private gender role scale. Model 2 shows that these patterns are essentially unchanged by controlling for homemaker status, education, and religiosity, although the magnitude of the coefficient associated with attitudes toward public gender roles is reduced somewhat by the imposition of multivariate controls.
By contrast, the inclusion of interactive terms, as presented in Model 3, has a substantial effect on the estimates of the consequences of gender role attitudes on responses to the abortion scale. The estimation of these more complex models reduces the effect of the private gender role index to statistical insignificance. Further, while the effects of the public index are significant, the sign of the coefficient is reversed. While it is difficult to offer a substantive explanation for this result, the standardized coefficient suggests that this relationship is quite weak.
The effects of education on abortion attitudes are simply additive, and are increased somewhat by the inclusion of the interactive terms in Model 3. That is, more educated respondents are more likely to regard abortion as justified than their less educated counterparts, and this relationship does not appear to mediate the effects of either gender role index.
By contrast, the effects of the interaction term between homemaker status and the private gender roles index are statistically significant. While the effects of the simple homemaker variable are significant in Model 3, the sign is reversed, although, again, the magnitude of this relationship is quite weak. Finally, the effects of religiosity appear to enhance the effects of the public gender roles index on attitudes toward abortion. These results offer some limited support for [H.sup.2], although the effects of cross-pressures are (again) quite moderate and complex. Gender role attitudes appear to be at least potentially multidimensional, and alternative dimensions of gender role egalitarianism/traditionalism are affected by different cross-pressures. (7)
This study is motivated by a deceptively simple puzzle, namely, why is the empirical connection between attitudes toward appropriate gender roles and approval or disapproval of abortion so weak at the level of mass opinion, when the connection between these two values seems so compelling in elite-level discourse? The results of this study have perhaps contributed to a partial resolution of this problem.
Readers who value parsimony in explanations of social or political phenomena will be disappointed by the results of this study. The study is based on two hypotheses, relating the weak relationships which pose the problem to two general sources: A lack of cognitive sophistication, and the existence of cross-pressures. Both general hypotheses, and the more specific tests based on these expectations, have received limited support. The relationship between gender role attitudes and beliefs about abortion are (slightly) stronger for respondents who exhibit relatively high levels of internal attitudinal consistency, and more education respondents are significantly (albeit moderately) more likely to report "pro-choice" perspectives on abortion. Similarly, the effects of egalitarian gender role attitudes are limited among female respondents who do not participate in the paid labor force, and among highly religious persons. Support for the latter hypothesis is limited by the fact that different cross-pressures interact with different aspects of gender role attitudes.
Of course, this research is quite preliminary and exploratory. The measures of cognitive sophistication are fairly imprecise, and there are clearly many more variables which could reinforce the effects of egalitarian gender role attitudes (or limit the effects of such attitudes). Nevertheless, the results presented here suggest that there may not exist a simple, parsimonious explanation for the weak relationship between attitudes toward abortion and beliefs about the appropriate public and private roles of women. Identifying respondents for whom abortion is indeed a "woman's issue" involving gender equality may involve specifying a number of different variables on which the relationship between abortion and conceptions of women's proper social roles is contingent. "Itemizing the deductions" is a tedious, and perhaps unsatisfying task, but one which may be necessary to understand the elusive limitations of this aspect of the abortion controversy.
Received 28 July 2014 * Received in revised form 2 September 2014
Accepted 12 September 2014 * Available online 1 April 2015
(1.) A version of this paper was presented at the Annual Scientific Meeting of the International Society for Political Psychology, Rome, Italy, July 7, 2014.
(2.) Waves 4 and 5.
(3.) Elsewhere (Jelen, 2014) I have shown that frequency of prayer is generally a stronger predictor of abortion attitudes across diverse religious traditions than either frequency of attendance at religious services or subjective religiosity.
(4.) In separate analyses of different religious tradition, the gamma coefficient between these two variables ranges from .893 for Muslims to .610 for Roman Catholic respondents.
(5.) I am indebted to David Sears for this suggestion.
(6.) Roman Catholicism, as the tradition at the aggregate median of abortion attitudes, is the comparison category.
(7.) The effects of education seem somewhat sensitive to variations in model specification. By contrast, the models relating interactions between gender role attitudes, religiosity, and homemaker status are quite robust, and are quite similar whether the effects of different sets of independent variables (including interactions) are considered separately or in conjunction with one another.
Adamczyk, Amy (2013), "The Effect of Personal Religiosity on Attitudes Toward Abortion, Divorce, and Gender Equality," EurAmerica 43: 213-253.
Bolzendahl, Catherine I., and Daniel J. Mays (2004), "Feminist Attitudes and Support for Gender Equality: Opinion Change in Women and Men, 1974-1988," Social Forces 83: 759-789.
Carmines, Edward G., and James A. Stimson (1980), "The Two Faces of Issue Voting," American Political Science Review 74: 78-91.
Colker, Ruth (1989), Feminism, Theology, and Abortion: Toward Love, Compassion, and Wisdom," California Law Review 77: 1011-1075.
Converse, Phillip E. (1964), "The Nature of Belief Systems in Mass Publics," in David Apter (ed.), Ideology and Discontent. New York: Free Press, 206-264.
Converse, Philip E., and Gregory Markus (1979), "'Plus ca Change...." The New CPS Election Study Panel," American Political Science Review 73: 32-49.
Cook, Elizabeth Adell, Ted G. Jelen, and Clyde Wilcox (1992), Between Two Absolutes: Public Opinion and the Politics of Abortion Boulder, CO: Westview.
Farere, Myra Marx (2003), "Resonance and Radicalism: Feminist Framing in the Abortion Debates of the United States and Germany," American Journal of Sociology 109: 304-344.
Jelen, Ted G. (2014), "The Subjective Bases of Abortion Attitudes: A Cross-National Comparison of Religious Traditions," Politics and Religion 7: 550-567.
Jelen, Ted G., and Jonathan Doc Bradley (2014), "Abortion Opinion in Emerging Democracies: Central Europe and Latin America," Politics, Groups, and Identities 2: 51-65.
Jelen, Ted G., Sue Thomas, and Clyde Wilcox (1994), "The Gender Gap in Comparative Perspective," European Journal of Political Research 25: 171-186.
Killian, Michael, and Clyde Wilcox (2008), "Do Abortion Attitudes Lead to Party Switching? Political Research Quarterly 61: 561-573.
Luker, Kristin (1985), Abortion and the Politics of Motherhood. Berkeley, CA: University of California Press.
Lynxwiler, John, and David Gay (1996), "The Abortion Attitudes of Black Women, 1978-1991," Journal of Black Studies 2: 260-277.
McDonagh, Eileen (1996), Breaking the Abortion Deadlock: From Choice to Consent. New York: Oxford University Press.
Rossi, Maximo, and Patricia Triunto (2012), "Abortion en Uruguay y America Latina: El Posiciomamiento de las Ciudadanos," Equidad Desarrollo 18: 9-21.
Staggenborg, Suzanne (1994), The Pro-Choice Movement: Organization and Activism in the Abortion Conflict. New York: Oxford University Press.
Strickler, Jennifer, and Nicholas L Danigelis (2002), "Changing Frameworks in Attitudes Toward Abortion," Sociological Forum 17: 187-201.
Tuman, John P., Danielle Roth-Johnson, and Ted G. Jelen (2013), "Context and Conscience: Attitudes Toward Abortion in Mexico," Social Science Quarterly 94: 100-112.
Wetstein, Matthew (1996), Abortion Rates in the United States: The Influence of Opinion and Policy. Albany, NY: State University of New York Press.
Zaller, John R. (1992), The Nature and Origins of Mass Opinion. New York: Cambridge University Press.
TED G. JELEN
Department of Political Science, University of Nevada, Las Vegas
Table 1 Factor Analysis of Gender Role Attitude Items (Varimax Rotation) Spouse Income .425 .679 Working Mother .492 .600 University Gender .745 -.375 Men Better Leaders .733 -.415 Eigenvalue 1.515 1.134 Spouse Income: Both the husband and wife should contribute to household income (coding reversed). Working Mother: A working mother can establish just as warm and secure a relationship with her children as a mother who does not work. (coding reversed) University Gender: A university education is more important for a boy than for a girl. Men Better Leaders: On the whole, men make better political leaders than women do. All items are Likert scales Source: World Values Survey, 1999-2007. Table 2 Bivariate Correlations of Selected Variables with Abortion Attitudes Abortion scale Sexual Morality Index .623 *** Euthanasia justified .550 *** Men Better Leaders .214 *** University Gender .150 ** Public gender roles index (a) .211 *** Working Mother .095 *** Spouse Income .008 * Private gender roles index (b) .068 *** Combined gender roles index (c) .194 *** (a) Mean value of Men Better Leaders and University Gender items: Gamma = .507 (b) Mean value of Working Mother and Spouse income items: Gamma = .332 (c) Mean value of all four gender role items: Alpha = .448 * significant at .05 ** significant at .01 *** significant at .001 Source: World Values Survey, 1999-2007. Table 3 Correlation of Public and Private Gender Role Indices with Abortion Scale, by Gender Role Constraint High Medium Low constraint Constraint Constraint Public gender roles .205 *** .174 *** .136 *** Private gender roles .128 *** .121 ** -.002 ** significant at .01 *** significant at .001 Source: World Values Survey, 1999-2007. Table 4 Regression Models of Abortion Attitudes (OLS) Model 1 Model 2 b beta b beta Private .085 *** .022 .095 *** .024 Public .399 *** .131 .252 *** .082 Homemaker -- -- .174 *** .028 Homemaker*private -- -- -- -- Homemaker*public -- -- -- -- Education -- -- .129 *** .118 Education*private -- -- -- -- Education*public -- -- -- -- Prayer -- -- .250 *** .208 Prayer*private -- -- -- -- Prayer*public -- -- -- -- E. Orthodox .848 *** .076 .717 *** .080 Hindu -.146 -.009 .061 .004 Protestant -.335 *** -.044 -.348 *** -.046 Buddhist .573 *** .035 .250 ** .016 Muslim -1.063 *** -.191 -.602 *** -.100 Constant 1.759*** .374 *** Adjusted [R.sup.2] .082 .126 N 61175 30047 Model 3 b beta Private -.001 -.003 Public -.192 * -.063 Homemaker -.381 * -.061 Homemaker*private .151 ** .102 Homemaker*public .085 .029 Education .182 *** .157 Education*private -.047 -.148 Education*public .034 .106 Prayer -.035 -.046 Prayer*private .012 .034 Prayer*public .093 *** .247 E. Orthodox .742 *** .081 Hindu .060 .004 Protestant -.340 *** -.045 Buddhist .281 ** .018 Muslim -.663 *** -.105 Constant 1.894 *** Adjusted [R.sup.2] .172 N 30047 ** significant at .01 ** significant at .001 Source: World Values Survey, 1999-2007.
|Printer friendly Cite/link Email Feedback|
|Author:||Jelen, Ted G.|
|Publication:||Journal of Research in Gender Studies|
|Date:||Jan 1, 2015|
|Previous Article:||The story of the fairy-exile: Domnica Radulescu's Train to Trieste.|
|Next Article:||Sexual orientation and poverty in the UK: a review and top-line findings from the UK household longitudinal study.|