Opinion on Abortion

Although it is just one issue among many in American politics, abortion is frequently debated and highly salient (Arceneaux Reference Arceneaux2002). This issue again made headlines after the U.S. Supreme Court overturned Roe v. Wade in 2022 with its Dobbs v. Jackson Women’s Health Organization decision. In this decision, the Supreme Court upended legal precedent and determined that the U.S. Constitution does not provide the right to abortion. This paved the way for individual states to pass abortion legislation to severely restrict abortion access, and it set off legal wrangling and attempts by health care providers to understand state provisions on conditions that would qualify for exceptions to the abortion bans.

Despite the recent controversy, research demonstrates that public opinion on abortion is relatively crystallized compared with many other political issues (Adams Reference Adams1997; Converse and Markus Reference Converse and Markus1979; Zaller Reference Zaller1992). Most Americans have attitudes toward abortion (Jelen and Wilcox Reference Jelen and Wilcox2003), and these attitudes are more stable and constrained than attitudes on many other political issues. Scholars consider abortion an “easy issue” (Carmines and Stimson Reference Carmines and Stimson1989) or a “morality issue” (Mooney and Schuldt Reference Mooney and Schuldt2008) because conflict centers around values and there is little room for compromise. It is symbolic of broader, deeper debates about worldviews, the role of women in society, and the meaning of parenthood, sexuality, and gender equality (Jelen Reference Jelen1988, 1–9; Luker Reference Luker1985).

Although this issue is divisive, a majority of the public favors abortion access. According to a 2022 Pew Research Center survey, 61% of Americans say that abortion should be legal in all or most cases. Only 37% say that abortion should be illegal in all or most cases. The Pew data demonstrate that women, younger people, and people with a college degree or higher are all more supportive of abortion rights than their counterparts; however, even among men, people over the age of 65, and those with high school or less education, a majority support abortion rights (Pew Research Center 2022).

The two major political parties are polarized on the issue of abortion as well (Carmines, Gerrity, and Wagner Reference Carmines, Gerrity and Wagner2010; Carsey and Layman Reference Carsey and Layman2006; Jelen and Wilcox Reference Jelen and Wilcox2003; Levendusky Reference Levendusky2009; (Osborne et al. Reference Osborne, Huang, Overall, Sutton, Petterson, Douglas, Davies and Sibley2022). The parties have become more differentiated on abortion over time (Sanbonmatsu Reference Sanbonmatsu2002). The Democrats are aligned with a more “pro-choice” stance and Republicans with a more “pro-life” position (Adams Reference Adams1997). Within the electorate, we see that people have increasingly brought their partisan affiliations into congruence with their issue attitudes, either by changing their partisan affiliation to match their beliefs on key issues or by changing their attitudes on policies to match their partisan affiliation (Carsey and Layman Reference Carsey and Layman2006). As a result, attitudes on abortion are now more partisan than in the past.

Even though abortion attitudes have generated a good deal of political and academic attention over the past 50 years, one aspect that has only recently received scholarly attention is the role of sexism in abortion attitudes. Despite the inherent link between biological sex and abortion, as only people with uteruses can become pregnant, little attention has focused on sexism in predicting abortion attitudes.Footnote ¹ In recent years, the literature has begun to explore this connection by analyzing different forms of sexism, particularly benevolent sexism. Even though both men and women can have sexist attitudes, we argue that there is a role for “hostile sexism” in people’s abortion stances. In the next section of this article, we describe two complementary forms of sexism—benevolent and hostile—and the expected association with abortion attitudes.

Forms of Sexism

The literature on sexism in the fields of psychology and sociology is long established. Sexism can be broadly defined as “attitudes, beliefs, or behaviors that support unequal status of women and men” (Swim and Campbell Reference Swim, Campbell, Brown and Gaertner2003, 219).Footnote ² Sexism is not monolithic, and the literature, particularly in the fields of psychology and sociology, distinguishes between different forms of sexism.

The type of sexism that has most recently been connected to abortion attitudes is “benevolent sexism.” Benevolent sexism is “a subjectively favorable, chivalrous ideology that offers protection and affection to women who embrace conventional roles” (Glick and Fiske Reference Glick and Fiske2001, 109). This type of sexism “is comprised of warm but ultimately patronizing attitudes that revere women as deserving of men’s affection and protection, and as more moral, sensitive and sympathetic than men” (Murphy et al. Reference Murphy, Sutton, Douglas and McClellan2011, 814). It taps into notions that women are the “better half” or the “nobler sex,” so they cannot have roles in politics or business because they are too good, pure, or weak to serve in these roles. As such, men should be protectors and providers for women, who should occupy a different space than men because of their gentler constitutions (Glick and Fiske Reference Glick and Fiske1996). Benevolent sexism views motherhood as a natural role for women, so abortion would not fit with that role.

The other complementary form of sexism identified by Glick and Fiske (Reference Glick and Fiske1996) is “hostile sexism,” which can be defined as “hostility toward women who oppose traditional roles” (Travaglia, Overall, and Sibley Reference Travaglia, Overall and Sibley2009, 599). Traditional roles for women have historically included the roles of wife, mother, and homemaker. “Hostile sexism is a negatively valenced cluster of attitudes toward women that reviles them as manipulative, devious, complaining, and competitive with men” (Murphy et al. Reference Murphy, Sutton, Douglas and McClellan2011, 814). It taps into feelings that women take pleasure in creating problems for men. It is a “combative ideology that is hostile toward women who challenge traditional patterns and are seen as seeking to control men, either by using their sexuality or feminist ideology” (Thomae and Houston Reference Thomae and Houston2016, 152). Those reporting hostile sexist views believe women tease men regarding sex, they complain in the workplace, and they are critical of men (Glick and Fiske Reference Glick and Fiske1996).

Additionally, other types of sexism are identified in the literature, including overt sexism and covert sexism, modern sexism, and neosexism.Footnote ³ These forms of sexism may be connected to abortion attitudes as well. We focus on the role of hostile sexism in particular, for three reasons. First, hostile sexism is the complement to benevolent sexism in the overarching theory of “ambivalent sexism” identified by Glick and Fiske (Reference Glick and Fiske1996). As Glick and Fiske describe, benevolent sexism is chivalrous or paternalistic behavior toward women who conform to traditional roles; hostile sexism is antipathy toward women who are seen as taking power. We argue that both have a role in attitudes toward abortion.

Second, publicly available data include measures of hostile sexism but not measures of covert sexism, implicit sexism, and so on.Footnote ⁴ The data for complementary hypothesis testing are not available in other publicly available sources such as the General Social Survey, ANES, and other surveys. Finally, despite attempts to frame abortion restrictions as “pro-woman,” we argue that hostile sexism is still an important predictor of abortion attitudes. Although laws limiting abortion rights are often couched in terms of protecting women, an important component of anti-abortion legislation is controlling behavior. It is the last point on abortion framing versus attitudes that we focus on in the next section.Footnote ⁵

Sexism and Framing of Abortion

The framing of the abortion debate in recent years demonstrates a strategic movement toward a “pro-woman” (Roberti Reference Roberti2021) or “woman-protective” anti-abortion frame (Donley and Lens Reference Donley and Lens2021; Siegel Reference Siegel2007) among abortion opponents. To advance legislation against abortion, abortion opponents have moved from a pro-fetus or morality focus to a pro-woman frame. The pro-woman frame “situates abortion as inherently harmful for women—both psychologically and physically—and insists that women deserve better than abortion” (Roberti Reference Roberti2021, 207). The laws treat “women as an uninformed group in need of protection from their own ignorance and the practices of abortion providers” (Doan and Schwarz Reference Doan and Schwarz2020, 6). By promoting “regulatory abortion bills that ‘educate’ women about abortion or protect them from potential harm of abortion, anti-abortion lawmakers may claim they are advocates for women” (Roberti Reference Roberti2021, 207). Laws that require ultrasounds, waiting periods, mandatory “counseling” sessions, and “informed consent” for pregnant people before they can obtain an abortion all incorporate this type of woman-protective frame when trying to limit access to abortion services. The laws are “presented as protecting women from coercion and from making poor decisions” (Denbow Reference Denbow2015, 97). We would also mention that these types of frames are consistent with the views of conservative women’s organizations, such as Concerned Women for America and the Independent Women’s Forum (Schreiber Reference Schreiber2002). These organizations, and women who identify as conservative, oppose abortion specifically because they view abortions to be harmful to women (Schreiber Reference Schreiber2002, Reference Schreiber2008).

This “frame extension” has been adopted by those pursuing anti-abortion policies in the United States in “order to attract new adherents to their cause. While the dominant rhetoric of the movement today focuses on the fetus and the immorality of abortion, pro-woman rhetoric has reemerged that leverages both the language of women’s rights as well as science in order to propose that abortion is bad for women” (Rose Reference Rose2011, 1). State laws now regularly incorporate the pro-woman frame (Roberti Reference Roberti2021). As a result of this strategic framing of the abortion issue, it makes sense that benevolent sexism is related to abortion attitudes in the United States. Political activists have worked to create a palatable anti-abortion message that sounds supportive of women rather than hostile toward women. The strategic messaging focuses on women as potential victims of bad influences who need protection and education. This aligns well with gender portrayals under benevolent sexism, which emphasize women’s traditional roles as wives and mothers (Huang et al. Reference Huang, Davies, Sibley and Osborne2016; Osborne et al. Reference Osborne, Huang, Overall, Sutton, Petterson, Douglas, Davies and Sibley2022) and women as the weaker sex who need protection. Under this view, it is not necessarily misogyny that makes people less pro-choice, but rather the beliefs that women are natural caretakers and should embrace their roles as mothers. Sexism plays role in abortion beliefs through attitudes toward motherhood (Huang et al. Reference Huang, Davies, Sibley and Osborne2016). To go against this “natural” role would be harmful to women.

Creating a “pro-woman” frame is a calculated attempt to repackage the abortion debate. “This frame transformation represents a strategic tactic of anti-abortion legislators to soften political behavior and beliefs that are seen as hostile toward women, especially during a time where there is an unprecedented amount of anti-abortion bills introduced in the states” (Roberti Reference Roberti2021, 207). Yet, despite the pro-woman framing efforts in the abortion debate, there is an underlying power conflict surrounding abortion. Who gets to make decisions about abortion—when, where, and under what circumstances one can be obtained—is also about power and control.

Women who can make reproductive choices independently may be seen as challenging traditional power structures and threatening the power of men. This piece connects to hostile sexism, as “hostile sexism is central to the perception that it is ok for men to impose these restrictions” (Petterson and Sutton Reference Petterson and Sutton2018, 244). As Petterson and Sutton (Reference Petterson and Sutton2018, 235) show, “hostile sexism is associated with perceptions that men have the right to constrain women’s reproductive choices.” Hostile sexism, and the ability of men to control women, can be a predictor of abortion attitudes, regardless of pro-woman frames. Since “hostile sexism uncovers antipathy toward women,” Gothreau, Arceneaux and Friesen (Reference Gothreau, Arceneaux and Friesen2022, 2) also find that “hostile sexism predicts less support for abortion and birth control access, as well as funding for Planned Parenthood.” Petterson and Sutton (Reference Petterson and Sutton2018, 236) note that despite the role of hostile sexism in abortion attitudes, those opposed to abortion “seldom describe their position as being motivated to preserve male control over women. More typically, they couch their position as a desire to protect the fetus, and often the pregnant woman herself.” Petterson and Sutton provide important theoretical contributions to the study of sexism’s relationship to abortion attitudes using a sample of undergraduate psychology students in the United Kingdom and Amazon Mechanical Turk workers. Our article expands on this work in the U.S. context, examines a longer timeline, and uses a nationally representative sample.

It would be ideal if the available data also included measures for benevolent sexism so that both could be included in the model. Unfortunately, that is not the case. The available ANES measures tap into hostile sexism. Recent research on other aspects of electoral politics have explored the importance of hostile sexism and gender stereotypes, particularly in vote choice. Although both types of sexism are related to people’s votes in the 2016 U.S. presidential election, it is hostile sexism in particular that “is the basis for polarization, with those high in hostile sexism (men and women both) resisting female leadership” (Winter Reference Winter2022, 22). Cassese and Barnes (Reference Cassese and Barnes2018) find that many White women espouse sexist opinions and that a woman’s level of sexism is a strong predictor of voting for Donald Trump in 2016. Similarly, hostile sexism is related to vote choice in Australia’s 2019 federal election (Beauregard Reference Beauregard2021). Hostile sexism has been shown to create “an attitudinal barrier—especially for conservatives—to supporting equal pay for women” (Connor and Fiske Reference Connor and Fiske2019, 22). It is a predictor of tolerance toward sexual harassment (Russell and Trigg Reference Russell and Trigg2004) and opposition to breastfeeding in public (Huang, Sibley, and Osborne Reference Huang, Sibley and Osborne2020). Other recent work demonstrates the importance of attitudes toward women and the role of sexism and gender stereotypes in contemporary American public opinion (Cassese and Barnes Reference Cassese and Barnes2018; Kam and Archer Reference Kam and Allison2021; McThomas and Tesler Reference McThomas and Tesler2016; Valentino, Wayne, and Oceno Reference Valentino, Wayne and Oceno2018).

Although the framing of the abortion debate might tap into benevolent sexism, we argue that at the attitudinal level, it might manifest as hostile sexism as well. Benevolent sexism is more patronizing and paternalistic—having babies is good for women because it fits with their nature, according to this view. Women should have this explained to them so that they know not to have abortions. Hostile sexism would extend this opposition to abortion because women making reproductive choices independently potentially takes control away from men. Under this view, abortion opponents may not only believe that abortion is bad for women because it goes against their natural roles as mothers, but also oppose abortion because of their antipathy toward women and fear of women seeking power. Opposition to abortion is also related to their dislike of women and ability to take control. Our article attempts to capture the hostile dimension of sexism in its association with abortion attitudes in public opinion while acknowledging the role of benevolent sexism and its framing.

Other Predictors of Abortion Attitudes

This study focuses on the role of hostile sexism and attitudes toward women. However, it is worth noting that research has identified other key predictors of abortion attitudes. One of the focal points of earlier abortion attitudes analysis is religion. Research on abortion attitudes demonstrates that religious considerations are important predictors of abortion beliefs. Both religious tradition and level of religiosity are related to people’s abortion attitudes (Combs and Welch Reference Combs and Welch1982; Jelen Reference Jelen1988). This is a dominant frame when discussing abortion attitudes both within the literature and within public discourse.

We control for level of religiosity in our models. The variable is constructed by combining worship attendance frequency, views of the Bible, and importance of religion (Layman Reference Layman2001) (Cronbach’s α ranges between 0.76 and 0.82). We also separately control for religious tradition by including dichotomous variables for people who are Catholic, evangelical Protestant, or Jewish. Past research has shown that people who identify as Catholic or evangelical are more pro-life in their abortion attitudes than mainline Protestants. According to a recent YouGov analysis, Jewish people are the most pro-choice religious group, after atheists and agnostics (Ballard Reference Ballard2022).

As discussed in the first section of this article, another important predictor of abortion attitudes is party affiliation. Research on the role of party identification on abortion attitudes has demonstrated a clear link, with Democrats identifying predominantly as pro-choice and Republicans identifying predominantly as pro-life (Adams Reference Adams1997). The parties are clearly differentiated in their abortion stances (Carmines, Gerrity, and Wagner Reference Carmines, Gerrity and Wagner2010; Carsey and Layman Reference Carsey and Layman2006; Jelen and Wilcox Reference Jelen and Wilcox2003; Levendusky Reference Levendusky2009; Osborne et al. Reference Osborne, Huang, Overall, Sutton, Petterson, Douglas, Davies and Sibley2022; Sanbonmatsu Reference Sanbonmatsu2002). Over time, “the cumulative effect of abortion attitudes led pro-life Democrats and pro-choice Republicans to switch parties” so that people’s abortion attitudes align with party identification (Killian and Wilcox Reference Killian and Wilcox2008, 561). Our models also control for party identification and ideology.

Previous scholarship finds significant differences in abortion attitudes based on sociodemographic characteristics. Swank (Reference Swank2021) finds that gender, marital status, and social class are related to people’s opinion on abortion. Research also demonstrates that sexual identity or orientation can be linked to abortion attitudes (Worthen Reference Worthen2020), with lesbian, gay, and bisexual people more likely to favor access to abortion (Grollman Reference Grollman2017, Reference Grollman2019).

It is important to note that the research regarding gender’s link to abortion attitudes has presented less unified findings. Some scholars find that gender and attitudes about gender roles are not strong predictors of attitudes on abortion, or at least do not exert a direct influence on abortion attitudes (Jelen and Wilcox Reference Jelen and Wilcox2003). More recently, however, scholars have argued that when religiosity is appropriately controlled for within models, women are more supportive of abortion rights than men (Lizotte Reference Lizotte2015). Religiosity can be a “suppressor variable” that masks the gender difference in abortion attitudes between men and women (Barkan Reference Barkan2014). Social identities can interact with each other to predict abortion attitudes in different ways.

Earlier studies find that White Americans, on average, provide more permissive responses on abortion than Black Americans (Combs and Welch Reference Combs and Welch1982; Hall and Ferree Reference Hall and Ferree1986). However, recent survey findings suggest that Black Americans and Asian Americans are more pro-choice than White Americans (Pew Research Center 2022). Other scholars also find complex interactions between gender, race/ethnicity, and religiosity. For example, among Latino men, evangelicalism and worship attendance are associated with less support for abortion (Holman, Podrazik, and Mohamed Reference Holman, Podrazik and Mohamed2020).

Education has also been demonstrated to be a predictor of abortion attitudes. People who are less educated are more likely to take the pro-life stance (Blake Reference Blake1971; Tedrow and Mahoney Reference Tedrow and Mahoney1979) than those with more education. The reasons people take pro-life stances might vary based on education, too; Jelen (Reference Jelen1988) demonstrates that people who are less educated are more pro-life because they believe that abortion will foster sexual promiscuity, whereas more highly educated people take pro-life stances because of concerns over respect for life.

Because of the established significance of covariates including religiosity, religious tradition, party identification, race, age, education, sexual orientation, and marital status, we include controls for all of these factors in the models presented in the next section, along with the measure of hostile sexism.Footnote ⁶ All variables are coded to range from 0 to 1, unless otherwise noted.

Data on Hostile Sexism and Abortion Attitudes

We use 2012, 2016, and 2020 ANES data to test our hypothesis. ANES conducted between 4,270 and 15,729 interviews (pre- and post-election) across the years. Table 1 shows that the surveys were conducted both online and face-to-face across these years. The American Association for Public Opinion Research (AAPOR) Response Rate 1 ranges between 36.7% and 50% depending on the year and modality. The AAPOR Response Rate 3 was 49% in 2012 and 40.9% in 2020.

Table 1. American National Election Studies information

Notes: We would like to thank one of the reviewers for encouraging us to report on these essential features of the ANES.

Source: http://www.electionstudies.org.

These publicly available surveys provide an opportunity to test the relationship between sexism and contemporary abortion attitudes. Previously, surveys were inconsistent in including measures tapping into sexist attitudes (Schaffner Reference Schaffner2022). The 2016 ANES has the most comprehensive battery of questions tapping into hostile sexism. These include questions asking respondents whether women interpret innocent remarks as sexist, whether women fail to appreciate what men do for them, whether women seek power to control men, whether women try to keep men on a tight leash, whether women who ask for equality are seeking special favors, and whether women complaining about discrimination create more problems than they solve. These are all components of the hostile sexism index identified by Glick and Fiske (Reference Glick and Fiske1996). The specific question wording can be found in the online appendix. We created an index out these variables to measure hostile sexism in 2016 (Cronbach’s α = 0.79)

The 2020 ANES includes several of these items, including whether women interpret innocent remarks as sexist, whether women seek power to control men, whether women who ask for equality are seeking special favors, and whether women complaining about discrimination create problems. These are all components of the hostile sexism index. We created an index of hostile sexism using these variables in 2020 (Cronbach’s α = 0.72)

The 2012 ANES also includes a few of the same measures for hostile sexism, although there are only two measures available in 2012. Despite this limitation, we also include analysis of the 2012 ANES here. The two items do not assess the same breadth and depth of hostile sexism as the six measures in 2016, but we believe it is important to include all available years to demonstrate that hostile sexism is important for abortion attitudes not just during the Trump presidential campaigns. This is a more conservative test of our hypothesis, in that we analyze using a weaker measure, in a year when gender dynamics were not as pronounced. The 2012 measures are whether women who ask for equality are seeking special favors and whether women complaining about discrimination create more problems than they solve. Similarly, we created an index out of these two variables to measure hostile sexism in 2012 (Cronbach’s α = 0.57).

For all three years of data, the hostile sexism indices are coded so that higher values indicate the more sexist view.Footnote ⁷ The measures available for hostile sexism are somewhat different across the three years, but findings demonstrate support for our hypothesis that people who espouse more hostile sexist views are less supportive of abortion rights. Please refer to the online appendix for the specific question wording of hostile sexism items across years.Footnote ⁸

In addition to the measures of hostile sexism, the ANES also includes attitudes toward discrimination faced by women in the United States in all three surveys. As part of the battery about groups that might be discriminated against, the ANES asks how much discrimination women face. Following the modeling strategy of Cassese and Barnes (Reference Cassese and Barnes2018), who also examine the role of hostile sexism using 2012 and 2016 ANES data, we include the discrimination against women variable as a control covariate in our models. We coded the variable so that the higher responses indicate an increasing belief in discrimination against women in the United States. The specific question wording is available in the online appendix.

The abortion attitudes question is asked in 2012, 2016, and 2020 in the format that has been used by the ANES for decades. It is a four-category scale ranging from purely pro-life attitudes to purely pro-choice attitudes—abortion should never be permitted or should always be permitted, regardless of the circumstances surrounding the abortion. The middle two categories acknowledge that abortion should be permissible in a given set of circumstances. One of the middle categories includes allowance for abortion in cases of rape, incest, or danger to the woman’s life. The other middle category, referred to as the “clear need” option, indicates that a woman should be able to obtain an abortion for other reasons, too, so long as a clear need has been established.Footnote ⁹

We reversed the ordering of the response options in our analyses so that higher response options refer to more pro-life attitudes. The specific question wording and response options are available in the online appendix.

Analysis of Abortion Attitudes

Based on our theoretical framework, we have one main hypothesis: higher values of hostile sexism will be associated with a higher probability of pro-life abortion attitudes as opposed to pro-choice abortion attitudes. Our major focus is the comparison of pure pro-choice and pure pro-life response options in abortion attitudes. We also interpret the findings for other response options. We expect the relationship to hold after controlling for multiple theoretically relevant variables.

Bivariate analyses between hostile sexism and abortion attitudes between 2012 and 2020 provide preliminary support for our hypothesis. We computed pairwise correlation coefficient estimates between hostile sexism and abortion attitudes for each year. The plot of the estimates can be seen in Figure 1. The trend shows a growing association between hostile sexism and abortion attitudes over time. Those who score high on the hostile sexism scale tend to be more pro-life, with the trend increasing between 2012 and 2020. We also computed 95% confidence intervals around these estimates by using bootstrapped standard errors. We find that the strength of association between hostile sexism and abortion attitudes has grown since 2012 at statistically significant levels. After 2016, the association stays around the same level in 2020.Footnote ¹⁰

Figure 1. Bivariate relationship between hostile sexism and abortion attitudes over time.

The bivariate analyses show a growing association between hostile sexism and abortion attitudes between 2012 and 2020. However, they do not control for theoretically relevant variables. In the next sections of the article, we explain our modeling rationale and present multivariate findings that show further empirical support for our main hypothesis.

Although the abortion attitudes variable is somewhat ordered in terms of permissiveness toward abortion, we argue that it is a nominal variable. The response options do not monotonically increase or decrease in one direction. There is a clear difference between the pure pro-choice and pure pro-life options—and our hypothesis mainly concerns these responses. However, the ordering might be vague in the case of the clear need and rape/incest/danger options. Hence, we treat the abortion attitudes variable as an unordered covariate and run multinomial logit models for 2012, 2016, and 2020.

It is important to note that a multinomial logit model is essentially “equivalent to a series of pairwise logit models,” and its coefficient estimates are “interpreted in the same way as binary logit model parameters are interpreted, with comparison being with the base category” (Cameron and Trivedi Reference Cameron and Trivedi2022, 910). Although interpretations are identical, running a “series of binary logits is not optimal” due to changes in the sample sizes (Long and Freese Reference Long and Freese2014, 387). Therefore, we run multinomial logit models in which picking the pure pro-choice option served as the comparison category, unless otherwise indicated. The multinomial logit coefficient estimates in each model show the probability of picking a particular response option as opposed to the pure pro-choice option. Our focus is the comparison of the pure pro-choice and pure pro-life response options to test our main hypothesis.

The models use full samples (online and face-to-face) in all years, and the estimates are computed by using complex survey weights.Footnote ¹¹ The multinomial logit model results from 2012, 2016, and 2020 are presented in Figures 2, 3, and 4, respectively.Footnote ¹² The comparison category for the top-left and two bottom figures (pure pro-life, clear need, and rape/incest/danger options) is the pure pro-choice option. The comparison category for the top-right figure (pure pro-choice) is the pure pro-life option. The circles refer to the multinomial logit coefficient estimates, and the spikes are the 95% confidence intervals. If the confidence interval spike does not overlap 0, the coefficient estimate is statistically significant.Footnote ¹³

Figure 2. Abortion attitudes and hostile sexism in 2012.

Figure 3. Abortion attitudes and hostile sexism in 2016.

Figure 4. Abortion attitudes and hostile sexism in 2020.

Findings

Figure 2 shows that hostile sexism predicts abortion attitudes in the 2012 ANES at statistically significant levels for almost all response options. We find strong support for our hypothesis in 2012: people who score higher on hostile sexism are more likely to have a pure pro-life stance on abortion as opposed to a pure pro-choice stance. They are also more likely to pick the rape/incest/danger option as opposed to the pure pro-choice option. We find no statistically significant difference in the probability of picking the clear need option over the pure pro-choice option in 2012. Overall, hostile sexist attitudes are significantly related to attitudes on abortion, with higher levels of hostile sexism related to more support for pro-life attitudes as opposed to pro-choice attitudes. The finding stands even controlling for opinions on discrimination against women, level of religiosity, religious denomination, partisanship, ideology, sexual orientation, marital status, and other variables found to be significant in past research.

Figure 3 presents findings from the 2016 multinomial logit model. Similarly, the results provide a strong empirical support for our main hypothesis. People who score higher on hostile sexism are more likely to be pure pro-life as opposed to be pure pro-choice on abortion attitudes, controlling for other variables. They are also more likely to express support for the two middle categories (clear need and rape/incest/danger) compared with the pure pro-choice category, on average. Hostile sexist attitudes are significantly related to attitudes on abortion in 2016.

The significant findings in the clear need and rape/incest/danger categories may be attributable to two things. First, the measure in 2016 includes six items (instead of two), and it is a more comprehensive measure covering more of the depth and breadth of hostile sexism. Second, the dynamics of the 2016 election highlighted attitudes about women in ways not seen in 2012. Particularly, Hillary Clinton was running as the Democratic nominee, the first woman ever to obtain a major-party nomination for the presidency. Simultaneously, Donald Trump was running as the Republican nominee and faced questions about his attitudes toward and treatment of women for his Access Hollywood tape comments.

Finally, the 2020 model provides more recent empirical support for our hypothesis regarding the relationship between hostile sexism and abortion attitudes. Figure 4 demonstrates that those who score high on hostile sexism are more likely to pick the pure pro-life position as opposed to the pure pro-choice position, controlling for other theoretically relevant variables. Similarly, they are more likely to choose the rape/incest/danger option rather than the pure pro-choice option. As in 2012, the only category that fails to show a statistically significant difference is the clear need clause. This might indicate that respondents have a hard time distinguishing their attitudes between the pure pro-choice and clear need options. Nevertheless, the model findings demonstrate remarkable empirical support for our main hypothesis.

To provide a more substantive interpretation of the multinomial logit results, we computed average marginal effects. The average marginal effect refers to the change in the probability of picking each response option rather than others when there is a one-unit instantaneous change in the hostile sexism variable (for details on the interpretation of average marginal effects, see Cameron and Trivedi Reference Cameron and Trivedi2022; Long and Freese Reference Long and Freese2014). We computed the average marginal effects by holding all other variables at their observed values (Hanmer and Kalkan Reference Hanmer and Kalkan2013). The findings are presented in Figure 5. The hollow circles refer to the change in the probability of picking a particular abortion attitude response option rather than the other options as hostile sexism increases by one unit, on average. The spikes refer to the 95% confidence intervals computed by the delta method. If a spike does not overlap 0, it refers to a statistically significant change.

Figure 5. Average marginal effects of hostile sexism in 2012, 2016, and 2020.

Figure 5 shows that as hostile sexism increases by one unit, on average, the probability of picking the pure pro-choice option as opposed to the other options declines in 2012, 2016, and 2020. The decline is statistically significant in all years. We observe that the probability of picking the pure pro-life response option as opposed to other options increases as hostile sexism also increases by one unit, on average. The change is statistically significant in 2016 and 2020 and barely misses the threshold in 2012. A similar relationship is also found in picking the rape/incest/danger clause as opposed to the others. Growing hostile sexism is associated with an increase in the probability of picking this option rather than the others, and the change is statistically significant in 2012 and 2016. The figure corroborates what the coefficient estimates display in regard to the absence of a systematic link between hostile sexism attitudes and the probability of picking the clear need response option.

It is important to note that there is also a strong relationship between religiosity and people’s attitudes on abortion. People who are more religious are more likely to provide pure pro-life, rape/incest/danger, or clear need responses compared with pure pro-choice responses in 2012, 2016, and 2020. People who are highly religious are less likely to provide pure pro-choice responses than pro-life responses. This fits with the findings in the literature. Religious traditions provide a series of mixed findings, except for Jewish people, who express more pro-choice attitudes compared with other traditions in the model in some years.Footnote ¹⁴

We do not find a statistically significant relationship between people’s opinions on the level of discrimination against women and picking the pro-life option rather than the pro-choice option, except for 2020. In all years, Republicans and conservatives are less likely to report pro-choice attitudes than pro-life attitudes, on average. Non-Latinx White Americans do not differ from other respondents in their abortion attitudes in most cases. People who are more highly educated are more likely to provide pure pro-choice attitudes on abortion rather than pure pro-life attitudes. In general, younger people are also more likely to choose the pure pro-choice stance compared with the pure pro-life stance. Neither the marital status nor the sexual orientation variables reach consistent levels of statistical significance in our models predicting pure pro-life attitudes as opposed to pure pro-choice attitudes.

In sum, the findings provide strong empirical support for our main hypothesis. Hostile sexism predicts pure pro-life abortion attitudes at statistically significant levels in 2012, 2016, and 2020. Those who score high on the hostile sexism scale tend to consider themselves more pro-life as opposed to pro-choice. The link stays robust when controlling for theoretically relevant variables. The findings empirically support the expectation that contemporary abortion attitudes in the United States have empirically undeniable sexist roots.