Abortion is among the most controversial and frequently debated issues in Congress. Conflicts over abortion policy riders snarled appropriations bills throughout the 1990s (Dodson Reference Dodson2006; Swers Reference Swers2002). During Barack Obama's presidency, efforts to defund Planned Parenthood contributed to budget brinksmanship and government shutdowns (Swers Reference Swers, Shames and Och2018). Despite the issue's prominence, there are few analyses of the predictors of roll call voting on abortion issues beyond the late 1990s and no attention to differences in behavior by abortion issue type.
We conduct a comprehensive analysis of abortion voting between the 103rd (1993–94) and 115th (2017–18) Congresses, focusing on the evolving relationship between gender and partisanship. We find that gender is a significant predictor of Republicans’ voting behavior. On average, Republican women are less likely to support the pro-life position than Republican men, although this difference attenuates over time and Republican women largely fall in line with their fellow partisans by the 110th Congress (2007–08). Our analyses of specific abortion issues during this period demonstrate that Republican women were particularly likely to defect from their party's position on bills concerning women's autonomy, such as those related to contraception and family planning funding. By contrast, district characteristics drive the apparent gender difference in Democratic voting on abortion issues, except for votes related to late-term abortion bans. On these votes, Democratic men are more likely than Democratic women to support the pro-life position. Thus, our analyses suggest that the type of abortion issue debated and the public's responsiveness to partisan issue framing may bring forward gender considerations that mediate the influence of party and independently impact members’ voting decisions.
WHY STUDY VOTING ON ABORTION?
Abortion has played a central role in the polarization of the parties. After the Roe v. Wade decision, there was no strong partisan divide on the issue. However, by the 1980s, Republicans had become firmly pro-life, while Democrats were increasingly committed to reproductive choice as a pillar of women's rights (Adams Reference Adams1997; Sanbonmatsu Reference Sanbonmatsu2002; Wolbrecht Reference Wolbrecht2000). Partisan realignment among the public followed, with pro-life voters aligning with the Republican Party and pro-choice voters identifying with the Democratic Party (Adams Reference Adams1997; Ainsworth and Hall Reference Ainsworth and Hall2011; Cook, Jelen, and Wilcox Reference Cook, Jelen and Wilcox1992). Assuming that partisanship and polarization largely explain reproductive policy voting, congressional scholars have focused their attention on explaining policy development, examining bill sponsorship and action in committee and on the floor, as well as conducting case studies of particular issues (Ainsworth and Hall Reference Ainsworth and Hall2011; Dodson Reference Dodson2006; Swers Reference Swers2002, Reference Swers2013, Reference Swers, Shames and Och2018). Additional work at the state level analyzes the diffusion of restrictive abortion policies (Bentele, Sager, and Aykanaian Reference Bentele, Sager and Aykanaian2018; Berkman and O'Connor Reference Berkman and O'Connor1993; Kreitzer Reference Kreitzer2015).
Existing congressional vote analyses focus on a single Congress (Swers Reference Swers1998) or a small set of Congresses (Frederick Reference Frederick2009; Oldmixon Reference Oldmixon2002, Reference Oldmixon2005) and do not reach past the passage of the Partial-Birth Abortion Act of 2003. However, many controversial proposals have emerged since the early 2000s that have required members to take difficult votes, such as those restricting funding for abortion on the Obamacare insurance marketplaces, 20-week abortion bans, and efforts to defund Planned Parenthood. Moreover, no existing studies delve into differences across types of abortion policy issues even though public opinion varies on specific reproductive issues. While policies related to contraception enjoy widespread public support and seemingly favor pro-choice groups’ issue framing regarding women's rights (Vansickle-Ward and Wallsten Reference VanSickle-Ward and Wallsten2019), the public is ambivalent about late-term abortions and more responsive to pro-life arguments focused on the life of the unborn child on these issues (Sellers Reference Sellers2010; Swers Reference Swers2013). It is well known that parties and interest groups work to define and highlight issues in ways that garner public support for their policy stances and divide the opposition party (Petrocik Reference Petrocik1996; Sellers Reference Sellers2010; Sides Reference Sides2006). Thus, it is possible that gender matters more for some types of votes than for others, leading, for example, Republican women or Democratic men to vote differently than their copartisans. Our research seeks to disentangle the relationship between partisanship and gender in voting on abortion and to understand how this relationship varies over time and by policy issue.
LITERATURE REVIEW
Current research identifies several important influences on abortion policy voting, including partisanship, ideology, religion, and gender. It is well established that partisanship and ideology are the dominant predictors of voting behavior (Poole and Rosenthal Reference Poole and Rosenthal2007). As members of partisan teams, legislators are expected to vote with their party both on substantive policy issues and on procedural votes that set the terms for House debate (Lee Reference Lee2016). As a so-called easy issue that provokes strong opinions, abortion remains on the agenda year after year and can produce lasting shifts in the party system (Adams Reference Adams1997; Carmines and Stimson Reference Carmines and Stimson1989; Oldmixon Reference Oldmixon2005). Indeed, Adams (Reference Adams1997) finds that the parties realigned around abortion between the 1970s and 1990s, with more than 80% of Democrats voting pro-choice and Republicans pro-life by 1994 (see also Wolbrecht Reference Wolbrecht2000).Footnote 1 Similarly, Karol (Reference Karol2009) notes that state-level public opinion in the 1970s was a stronger predictor of senators’ abortion votes than partisanship, but by the late 1980s, the relationship had reversed, indicating that senators were more responsive to intense factions within their parties than to opinion in their states. Throughout the 1990s and early 2000s, taking a pro-choice position often defined what it meant to be a moderate Republican, and being pro-life was an indicator of conservativism among Democrats (Oldmixon Reference Oldmixon2002, Reference Oldmixon2005; Swers Reference Swers1998, Reference Swers2002).
The dominance of partisanship as a predictor of legislators’ votes generally seems to leave little room for the influence of other factors. However, in line with the idea of morality politics, votes on abortion are often characterized as “votes of conscience” on which members draw on their own principles, particularly their religious beliefs (Oldmixon Reference Oldmixon2005; Tatalovich and Schier Reference Tatalovich and Schier1993). Scholars note that members who are evangelical Christians are particularly active on abortion and Catholic members have more pro-life voting records. Indeed, being Catholic was an important predictor of pro-life voting among Democrats through the early 2000s (Oldmixon Reference Oldmixon2002, Reference Oldmixon2005; Swers Reference Swers1998, Reference Swers2002; Tatalovich and Schier Reference Tatalovich and Schier1993). Furthermore, members representing districts with higher proportions of religious conservatives and Catholics vote more pro-life (Oldmixon Reference Oldmixon2002, Reference Oldmixon2005).
Finally, research demonstrates that women legislators across both parties draw on their life experiences as women and a desire to represent women when crafting their agendas (Dittmar, Sanbonmatsu, and Carroll Reference Dittmar, Sanbonmatsu and Carroll2018; Osborn Reference Osborn2012; Reingold Reference Reingold2000; Swers Reference Swers2002, Reference Swers2013). Because abortion is the most heavily debated women's rights issue in Congress, the issue is central to the women's organizations that are an important element of the Democratic Party and a key priority of social conservatives who are a pivotal constituency within the Republican coalition (Sanbonmatsu Reference Sanbonmatsu2002; Swers Reference Swers2002, Reference Swers2013; Wolbrecht Reference Wolbrecht2000). Whether stemming from life experiences or a desire to represent women constituents, Democratic women are especially likely to advance pro-choice policies and to oppose efforts to limit women's access to abortion or family planning services (Dittmar, Sanbonmatsu, and Carroll Reference Dittmar, Sanbonmatsu and Carroll2018; Dodson Reference Dodson2006; Swers Reference Swers2002, Reference Swers2013, Reference Swers, Shames and Och2018). At the state level, the presence of a critical mass of Democratic women in the legislature or on particular committees is associated with fewer adoptions of certain pro-life policies (Bentele, Sager, and Aykanaian Reference Bentele, Sager and Aykanaian2018; Berkman and O'Connor Reference Berkman and O'Connor1993; Kreitzer Reference Kreitzer2015).
Meanwhile, analyses of the impact of gender on abortion voting find that Republican women often drive apparent gender differences because they are more likely to support pro-choice positions than Republican men (Oldmixon Reference Oldmixon2005; Swers Reference Swers1998, Reference Swers2002; Tatalovich and Schier Reference Tatalovich and Schier1993). However, increased partisan polarization in Congress has likely reduced this gender gap. Indeed, Frederick (Reference Frederick2009) finds that Republican women in the House are more liberal than Republican men, but these differences disappear by the 108th (2003–04) and 109th (2005–06) Congresses. Swers (Reference Swers, Shames and Och2018) also demonstrates that since the Tea Party revolution in 2010, some conservative Republican women have taken lead roles in efforts to restrict abortion, particularly in initiatives to investigate and defund Planned Parenthood.
Studies of state legislatures also note growing pro-life advocacy among Republican women. Gunnar Bentele et al. (Reference Bentele, Sager and Aykanaian2018) find that between 2008 and 2014, states that had Republican majorities with more Republican women in their caucuses were more likely to pass restrictive abortion legislation. Focusing on bill sponsorship, Reingold et al. (Reference Reingold, Kreitzer, Osborn and Swers2020) demonstrate that conservative Republican women are especially active proponents of pro-life polices framed as helping women rather than fetal-centric bills or proposals with a religious or moral frame. More research is clearly needed to understand whether the type of issue under consideration elevates gender considerations for particular members and how these potential gender differences are impacted by increasing party polarization around abortion.
THEORY AND HYPOTHESES
This research seeks to disentangle the relationship between partisanship and gender in voting on bills related to reproductive rights and to understand how this relationship varies over time and by issue type. This project brings together existing research that finds partisanship, gender, and interest group activity to be key drivers of legislative behavior (Bawn et al. Reference Bawn, Cohen, Karol, Masket, Noel and Zaller2012; Dittmar, Sanbonmatsu, and Carroll Reference Dittmar, Sanbonmatsu and Carroll2018; Grossman and Hopkins Reference Grossmann and Hopkins2016; Poole and Rosenthal Reference Poole and Rosenthal2007; Reingold Reference Reingold2000; Swers Reference Swers2002, Reference Swers2013). Because of the Republican Party's commitment to social conservatism, we would expect both Republican men and women to be more pro-life than Democrats (Grossmann and Hopkins Reference Grossmann and Hopkins2016; Schreiber Reference Schreiber2008). Given the centrality of women's groups to the Democratic coalition, support for women's rights is a key tenet of Democratic politics for both women and men. The literature on women legislators, however, suggests that Democratic women's commitment to issues perceived as safeguarding women's rights may be stronger than Democratic men's, as it is reinforced by their life experiences and desire to represent their women constituents. Therefore, we also expect Republican women to be more supportive of efforts to defend women's rights.
H1: The parties’ stances on women's rights and gender roles create different challenges and incentive structures for their members, meaning that gender will differentially affect members’ voting behavior on abortion-related issues by party.
H1a: The dual influences of gender and partisanship will pull in different directions for Republican women, creating cross-pressures for some Republican women on policies related to abortion. Therefore, Republican women will be less likely than Republican men to support pro-life positions on abortion-related legislation.
H1b: The dual influences of gender and partisanship will reinforce each other for Democratic women. Therefore, Democratic women will vote more strongly pro-choice than Democratic men.
We also expect that the relationship between partisanship and gender has changed over time as the party coalitions have evolved. The increasing centrality of women's groups and abortion rights organizations such as NARAL and Planned Parenthood to the Democratic coalition means that support for women's rights and reproductive rights in particular has become an increasingly important litmus test for Democratic candidates (Crowder Meyer and Cooperman Reference Crowder-Meyer and Cooperman2018; Sanbonmatsu Reference Sanbonmatsu2002). The fundraising influence of EMILY's List, a group devoted to electing only pro-choice women candidates, has also grown over time (Burrell Reference Burrell2014; Crowder Meyer and Cooperman Reference Crowder-Meyer and Cooperman2018). Additionally, individual donors, particularly Democratic women, are more committed to expanding women's representation in office—likely with the expectation that this will lead to greater attention to women's interests (Thomsen and Swers Reference Thomsen and Swers2017). Democratic candidates have also become increasingly reliant on the mobilization of women, whom they appeal to with campaigns focused on preserving Roe and combating threats to access to reproductive health services (Dittmar Reference Dittmar2016; Frasure-Yokley Reference Frasure-Yokley2018). If pro-choice stances are increasingly a prerequisite for raising the funds and building the electoral support necessary to win a Democratic nomination, we expect that the Democratic gender gap in voting on abortion-related bills will attenuate over time.
Similarly, the influence of social conservatives in the Republican Party has grown since Ronald Reagan's presidency. Christian conservative groups such as Focus on the Family, Concerned Women for America, and the Family Research Council, as well as pro-life organizations such as the National Right to Life Committee and the Susan B. Anthony List, are increasingly pivotal players in Republican politics (Sanbonmatsu Reference Sanbonmatsu2002; Schreiber Reference Schreiber2008; Wolbrecht Reference Wolbrecht2000). In their analysis of donors in 2010 and 2012, Thomsen and Swers (Reference Thomsen and Swers2017) find that Republican donors focus most on the candidate's conservativism, with women preferring more conservative candidates than men. Moreover, religious conservatives are among the most loyal Republican voters (Abramowitz Reference Abramowitz2010; Grossman and Hopkins Reference Grossmann and Hopkins2016). If candidates must increasingly respond to the expanding influence of social conservatives by taking pro-life positions, the Republican gender gap in voting on abortion-related bills will attenuate over time.
H2: The differential effects of gender on representatives’ partisan voting behavior on abortion-related legislation will decrease over time, reflecting the historical evolution of the party coalitions as social conservatives have expanded their influence within the Republican Party and women's groups have become more central to the Democratic coalition.
H2a: Over time, the influence of gender on abortion policy voting will decline and Republican women will be just as likely as Republican men to vote for pro-life initiatives.
H2b: Over time, the influence of gender on abortion policy voting will decline and Democratic men will be just as likely as Democratic women to oppose pro-life proposals.
Finally, we expect the influence of gender on abortion voting to vary across policy issue types. Despite vigorous contestation among activists, public opinion on abortion has remained relatively stable over time (Fiorina Reference Fiorina2017). However, there is important variation in opinion across different issues, as access to contraception has long enjoyed broad public support (Vansickle-Ward and Wallsten Reference VanSickle-Ward and Wallsten2019). Party issue ownership theory suggests that parties and their interest group allies will focus public debate on issues that favor their party and will frame issues in ways that highlight party strengths (Petrocik Reference Petrocik1996; Sellers Reference Sellers2010; Sides Reference Sides2006). As a result, Democrats’ and pro-choice groups’ emphasis on women's reproductive rights should resonate most with voters in the realm of contraception. Based on this broad public support, in combination with gender considerations that make them more sensitive to arguments about women's rights, Republican women should be more likely than Republican men to experience conflict between their gendered life experiences and the Republican Party's stance against broadening access to and funding for contraception. As a result, Republican women will be more inclined than Republican men to take pro-choice votes on contraception issues such as bills cutting family planning programs or defunding Planned Parenthood.
Similarly, issues that focus public attention on the vulnerability of the fetus favor Republicans and pro-life groups’ emphasis on the unborn child. After failing to overturn Roe through constitutional amendment, pro-life advocates have increasingly found success by focusing on incremental proposals to restrict abortion that are framed as protecting unborn children and pregnant women, such as the partial-birth abortion ban and parental notification laws (Ainsworth and Hall Reference Ainsworth and Hall2011; Rose Reference Rose2007). The eight-year congressional effort (1995–2003) to ban “partial-birth” abortions gained particular traction with the public. By adopting the “partial-birth” terminology to describe a late-term abortion procedure, Republicans successfully framed the method as akin to infanticide (Sellers Reference Sellers2010; Swers Reference Swers2013). When the bill became law in 2003, Gallup reported that 68% of the public felt that the “abortion procedure known as ‘late-term’ abortion or ‘partial birth’ abortion . . . should be illegal.” This number rose to 72% when the issue was upheld by the U.S. Supreme Court in 2007 (Saad Reference Saad2007). Given this overwhelming popular support, Democrats felt strong pressure to fall in line with public opinion. The dual influences of gender and party that make Democratic women committed advocates of reproductive rights may make them more resistant than Democratic men to this public pressure (Dodson Reference Dodson2006; Swers Reference Swers2002, Reference Swers2013). In sum, the type of abortion issue being debated and the efforts of partisans and their allies to focus public attention on women's rights or on the life of the unborn may further highlight gender considerations for some members, which will mediate the influence of partisanship and independently impact legislators’ voting decisions.
H3: The influence of gender on members’ voting behavior will vary by abortion policy issue. Public opinion and the parties’ efforts to shape that opinion will create more cross-pressures for certain party-gender groups.
H3a: Public perception and Democratic promotion of contraception and family planning as essential to women's autonomy will increase the significance of gender as a determinant of voting, such that Republican women will be more cross-pressured in their efforts to reconcile their gendered life experiences and partisan considerations and more likely than Republican men to oppose pro-life policies related to family planning and contraception.
H3b: Republican framing of late-term abortion bans as equivalent to infanticide and public discomfort with the procedure will create more cross-pressures for Democratic men. Because of the reinforcing alignment of gender and partisanship, Democratic women will be the most strongly pro-choice and more resistant to shifts in public opinion in favor of late-term abortion bans. In contrast, Democratic men will be more responsive to shifts in public opinion making these legislators more likely to support late-term abortion bans.
DATA AND METHODOLOGY
To evaluate the evolving influence of partisanship and gender on the voting behavior of members of Congress on abortion-related legislation, we analyze representatives’ support for pro-life positions on votes that were scored by the National Right to Life Committee (NRLC) from the 103rd (1993–94) to the 115th (2017–18) Congresses.Footnote 2 Founded in 1968, the NRLC is the oldest and largest pro-life organization and maintains the longest-running publicly available set of vote scores on issues related to abortion among all pro-life and pro-choice organizations.Footnote 3 In fact, the group has lobbied for restrictions on abortion for longer than some of the key pro-choice organizations have engaged in political advocacy. For example, the first presidential candidate that Planned Parenthood endorsed was John Kerry in 2004 (Richards Reference Richards and Peterson2018).Footnote 4 By using NRLC data, we are able to examine changes in voting behavior between the 1990s, during a period of party realignment on the issue, through numerous controversial votes taken in recent Congresses.
We include all votes identified by the NRLC that directly impact abortion policy. The votes cover a wide range of issues including limitations on federal funding for abortion, domestic Title X and international family planning programs, stem cell research, and bans on late-term abortions. We exclude votes on issues unrelated to abortion such as campaign finance bills and votes that cannot be directly connected to a member's abortion position. For example, we include votes on two amendments during the 104th Congress on a defense appropriations bill concerning abortions at overseas military bases, but we exclude a vote on the bill's conference report because it is not clear whether this vote is related to members’ abortion position or their stance on other defense issues.Footnote 5 Our analyses include 138 votes across a 25-year period.
Table 1 provides a descriptive overview of the distribution of men and women serving in Congress by party and session. The number of scored abortion votes ranges from 2 during the 111th Congress (2009–10), when Democrats controlled the House and Barack Obama was president, to 25 during the 104th Congress (1995–96), the first Congress after the Republican Revolution when Republicans won back the House majority after 40 years in the minority. Congress was most active on abortion during the Bill Clinton presidency (103rd–106th Congresses) as the Democratic president and Republicans clashed over Republican efforts to utilize appropriations bills to restrict federal funding for abortion and family planning programs. This period also saw the beginning of the campaign to ban “partial-birth” abortion and efforts to adopt a parental notification bill (Dodson Reference Dodson2006; Swers Reference Swers2002, Reference Swers2013). Many of these controversies continued during the George W. Bush presidency (107th–110th Congresses), and new issues emerged, including debates related to cloning and stem cell research as well as the Unborn Victims of Violence Act, which established additional federal punishments for harming the fetus when a pregnant woman is killed or injured in a federal crime. The final years of the Bush presidency also saw the first efforts to defund Planned Parenthood with an amendment sponsored by then-representative Mike Pence (R-IN).
Descriptive statistics
Although the number of abortion votes declined during Obama's presidency and into the first two years of the Donald Trump administration, new issues have come to the fore, including proposals to prohibit federal funds from being used for abortion in the Affordable Care Act health insurance exchanges and to ban taxpayer funding for abortion. The effort to defund Planned Parenthood also intensified during the Obama years. Multiple votes on this issue and proposals to adopt a new late-term abortion ban after 20 weeks led to the largest number of abortion policy votes in the Obama presidency during the 114th Congress (2014–15). It was during this Congress that the dispute over defunding Planned Parenthood led Speaker John Boehner to resign in order to avoid a government shutdown (Green Reference Green2019; Swers Reference Swers, Shames and Och2018). Throughout this 25-year period, women were underrepresented in Congress. Although the number of Democratic women grew from 13.5% to 32% of the Democratic conference, Republican women never reached more than 10% of their caucus. Table 2 provides an overview of the types of abortion policy issues and when they were considered.
Issue categories
To capture as much variation in members’ voting behavior as possible, we use a panel data set organized by member-vote, which allows for the inclusion of every vote cast on abortion, including votes taken by members who left or entered Congress in the middle of a session.Footnote 6 Our data include nearly 60,000 vote observations. Figure 1 provides an initial picture of the trends in abortion voting by party and gender. As expected, Republican men appear to be the most committed pro-life voters, casting almost 90% of their votes in support of the NRLC in the early Congresses and closer to 100% of their votes in support of the NRLC in later Congresses (111th–115th). Meanwhile, Democratic women are the most reliably pro-choice legislators, supporting the NRLC in barely 5% of their votes between the 103rd and 106th Congresses and universally opposing the NRLC in all votes since the Tea Party wave that ushered in the 112th Congress (2011–12).
Percent support for NRLC position by Congress.
Republican women and Democratic men are less consistent in their abortion voting behavior. Between the 103rd and 108th Congresses, Republican women cast between 30% and 50% of their votes against the NRLC's position. They continued to cast almost 25% of their votes against the NRLC position through the final Congress of the George W. Bush administration (110th Congress). However, by the Obama presidency (111th–114th Congresses), Republican women appear to have fallen in line with Republican men. Similarly, Democratic men defected from their party, supporting the NRLC position on 25% to 30% of their votes between the 103rd and 108th Congresses, but they voted as strongly pro-choice as Democratic women in the three most recent Congresses (113th–115th). Figure 1 supports our contention that Republican women and Democratic men are more cross-pressured in their voting compared with Republican men and Democratic women. In line with our hypotheses, though, Figure 1 also shows that the influence of gender on abortion votes has attenuated over time.
However, it is possible that there are underlying factors beyond partisanship and gender that drive representatives’ voting behavior. Perhaps Republican women appear to vote more pro-choice in early Congresses because they are more likely than Republican men to be elected from more liberal districts. Similarly, it is possible that pro-life Democratic men in earlier Congresses were more likely to be elected from more conservative districts or that, in comparison with the women, more of the Democratic men were Catholics. To determine whether these gender-party patterns persist, we account for other factors that are known to influence congressional voting on abortion.
In addition to party and gender, the independent variables in our analyses incorporate individual member and district-level characteristics that are known to influence voting behavior. Since ideology is consistently one of the most important predictors of voting, we utilize first-dimension NOMINATE scores calculated by Lewis et al. (Reference Lewis, Poole, Rosenthal, Boche, Rudkin and Sonnet2019). The scores capture ideological liberalism and conservatism and range from –1, indicating most liberal, to +1, indicating most conservative. We also incorporate second-dimension NOMINATE scores, which capture within-party differences concerning cultural issues such as civil rights and abortion (Bailey Reference Bailey2013; Poole and Rosenthal Reference Poole and Rosenthal2007).
In addition to gender, we account for the race and ethnicity of members because African American and Latinx members anchor the liberal end of the Democratic Party, although African American and Latinx voters have historically leaned more conservative on abortion (Cook, Jelen, and Wilcox Reference Cook, Jelen and Wilcox1992). We also include measures of seniority (terms served), the proportion of the votes members received in their last election, and their status as a majority or minority party leader as indicators of experience and electoral security.Footnote 7 Since votes on abortion are often described as conscience votes (Oldmixon Reference Oldmixon2002, Reference Oldmixon2005; Tatalovich and Schier Reference Tatalovich and Schier1993), we account for whether a member is an evangelical Christian, Mormon, or Catholic. These measures come from Nicole Asmussen Mathew (Reference Mathew2018) and are based on the home church membership of representatives. Mathew has also created measures of the religious affiliation of members’ constituents by tracking church membership data at the district level, which we include to control for the evangelical, Catholic, and Mormon populations within a district.
Other district variables measure the Democratic presidential vote to account for the partisan and ideological leanings of the district. College-educated voters with higher incomes living in more urban areas are known to be more socially liberal and supportive of abortion rights, while older voters are generally more conservative (Abramowitz Reference Abramowitz2010; Cook, Jelen, and Wilcox Reference Cook, Jelen and Wilcox1992).Footnote 8 Additionally, research demonstrates that women legislators are historically more likely to be elected from more educated, racially diverse, and higher-income districts (Palmer and Simon Reference Palmer and Simon2008). Therefore, we include measures of the district's racial and ethnic composition, proportion of college-educated voters, median age, median household income, unemployment rate, and percentage of urban residents to ensure that differences attributed to gender do not actually stem from the fact that women represent different kinds of districts.
To analyze the relationship between members’ gender-partisan identities and the likelihood of casting a vote in support of the NRLC's position on abortion-related legislation, we estimate linear probability models using multivariate ordinary least squares regression on a panel data set structured by member-vote. The dependent variable in our analyses is a binary variable capturing whether a member voted in support of the NRLC's position on a given vote between the 103rd and 115th Congresses. To test our second hypothesis, that gender differences among partisans should attenuate over time, we incorporate a time trend variable to mark a given vote's distance from the 103rd Congress. By interacting this time trend with each gender-party group indicator, we evaluate whether and when the differences among Democrats and Republicans based on gender declined. We use linear probability models because logit or probit models can yield inconsistent coefficients when used with fixed effects on panel data (Greene Reference Greene2004). They also tend to produce similar estimates to logit or probit models estimated with maximum likelihood estimation with the benefit of being more easily interpretable (Angrist and Pischke Reference Angrist and Pischke2009).Footnote 9 We also cluster all standard errors by member to account for possible correlations in the error term between the votes of individual members.
We include fixed effects by Congress, state, and policy issue type.Footnote 10 The inclusion of Congress fixed effects serves as a control variable for time that allows us to account for broader trends in the political context of each Congress such as the partisan composition of the chamber, ideological polarization, and the public saliency of abortion at the national level. State fixed effects control for state-level, time-invariant characteristics that could affect the types of legislators who are elected and those legislators’ abortion positions, such as state-wide cultural differences and policy norms. Finally, policy issue type fixed effects hold constant characteristics common across similar kinds of votes in terms of content. We identify 12 policy issue categories, as shown in Table 2. We created these policy categories by conducting a review of the vote descriptions provided by the NRLC and of media coverage surrounding congressional action on these abortion controversies. We also draw on academic work that includes case studies of particular abortion-related controversies and analyses of congressional action related to abortion (e.g., Ainsworth and Hall Reference Ainsworth and Hall2011; Dodson Reference Dodson2006; Oldmixon Reference Oldmixon2002, Reference Oldmixon2005; Swers Reference Swers2002, Reference Swers2013, Reference Swers, Shames and Och2018). Table 2 includes a description of the issues comprising each category and the Congresses that saw votes on these issues.
We test our third set of hypotheses, that specific issue areas are more likely to create cross-pressures for specific party-gender groups, by conducting analyses on the subsets of votes related to late-term and other abortion bans (the categories “partial-birth” abortion bans, 20-week abortion bans, and other abortion bans in Table 2) and the subsets of votes concerning international family planning and contraception (the categories foreign policy family planning and contraception in Table 2). These categories are most representative of policy issues that are framed as being related to protecting the unborn and protecting women's autonomy, respectively. These categories of votes also appear across the data set, allowing us to test whether men and women partisans’ abortion-related decisions are impacted by the type of issue considered.
RESULTS AND ANALYSES
We analyze the influence of gender and partisanship on abortion policy voting in Table 3 and Table 4 using six models. Model 1 estimates the bivariate relationship between gender-party group and abortion-related voting behavior. Model 2 includes member-specific characteristics, and Model 3 adds district-level factors. Models 4, 5, and 6 add Congress, state, and vote type fixed effects.
Likelihood of supporting NRLC position (Republican man omitted)
Likelihood of supporting NRLC position (Democratic woman omitted)
The models in Table 3 evaluate the likelihood that a member will vote in support of the NRLC's pro-life position on a given vote, using Republican men as the omitted category to easily highlight the differences between Republican men and women. In line with H1a, Table 3 estimates that Republican women are significantly less likely to support the NRLC position compared with Republican men. After adding a robust set of member-specific and district-level characteristics, Congress fixed effects, state fixed effects, and issue type fixed effects in Model 6, we estimate that the votes of Republican women are approximately 9 percentage points less likely to align with the NRLC stance than those of Republican men over our period of analysis. More ideologically conservative members are more likely to support the NRLC position on legislation, as are Catholics and legislators with more seniority. Members from districts with older constituents and those with higher levels of unemployment are also more likely to cast pro-life votes.
Table 4 repeats the same analyses but utilizes Democratic women as the omitted category to clearly estimate the underlying differences between Democratic men and women. In contrast with Republicans, after controlling for district characteristics, there are no significant differences in the likelihood of Democratic men and women voting against the NRLC position. This suggests that apparent differences between the voting records of Democratic men and women shown in Figure 1 may be driven by differences in the types of districts that elect Democratic women.
Supporting this finding, Pearson and McGhee (Reference Pearson and McGhee2013) find that in comparison with the districts that elect Democratic men, the districts that elect Democratic women demonstrate higher levels of support for the Democratic presidential candidate. Similarly, Palmer and Simon (Reference Palmer and Simon2008) note that historically, the districts that elect women are more educated, higher income, and more racially and ethnically diverse. In contemporary politics, such districts increasingly elect Democrats (Abramowitz Reference Abramowitz2010). Looking at the Democrats serving during our 25-year period spanning the 103rd–115th Congresses, districts represented by Democratic women on average had 7 percentage points higher levels of Hispanic constituents (24% versus 17%), were 11 percentage points more urban (93% versus 82%), had 6 percentage points higher support for the Democratic presidential candidate (68% versus 62%), and had 4 percentage points more college-educated constituents (30% versus 26%) than the districts represented by Democratic men.
Given the continuing polarization of the parties on abortion, we also hypothesize that the influence of gender on members’ votes should diminish over time as members increasingly vote with their party. To test our second set of hypotheses, we interact a time trend variable with each gender-party group. Figure 2 reports the resulting estimates of the marginal effect of time on the voting behavior of Republican women and Democratic men, respectively.Footnote 11 Figure 2 suggests that the likelihood of Republican women taking a pro-life position on NRLC-scored votes has increased over time. While Republican women were more likely to defect from the NRLC's position in the 103rd–110th Congresses (1993–2008), by the 111th (2009–10) and 112th (2011–12) Congresses, Republican women are just as reliable pro-life voters as Republican men. By contrast, we find that Democratic men's voting behavior quickly converges with Democratic women's with the confidence intervals of our estimates crossing zero as early as the 105th Congress (1997–98), indicating that Democratic men are as likely as Democratic women to oppose the NRLC's position over our period of analysis.
Marginal effect of time
Finally, we predict that the importance of gender considerations will vary with the type of abortion issue addressed by a given vote, as members’ willingness to stand with the party on particular abortion issues in the face of public backlash or criticism from colleagues will be impacted by their gender. We anticipate that the public perception of contraception as fundamental to women's autonomy will elevate the importance of gender as a determinant of Republican women's votes. To test this hypothesis, we evaluate the subsets of votes related to contraception and domestic and international family planning. These votes include proposals to limit funding for Title X family planning, forbid funding for research of abortion inducing drugs, block Planned Parenthood from receiving Title X funds, and prohibit funding for international groups that perform abortions.
Table 5 estimates the relationship between gender-party group and the likelihood of a vote being aligned with the NRLC position among only those votes related to contraception and family planning. We find evidence to support our hypothesis that gender considerations play a role in Republican women's voting behavior on contraception and other family planning policies. Holding constant individual-, district-, state-, and Congress-level characteristics, we estimate the votes of Republican women to be about 11 percentage points less likely to align with the NRLC position on contraception-related legislation compared with the votes of Republican men. The influence of moving from a Republican woman to a Republican man is comparable to the impact of religion, as Catholic members are about 10 percentage points more likely to adopt the pro-life position on contraceptive issues that is favored by the NRLC.Footnote 12
Likelihood of supporting NRLC position, contraception and international family planning votes (Republican man omitted)
Yet we also find that gender becomes a less important predictor of Republican women's votes on these issues over time. Figure 3 suggests that Republican women were more likely to vote distinctively on family planning issues as late as the 110th Congress (2007–08).Footnote 13 However, by the Obama presidency, Republican women's support for the NRLC's pro-life position was indistinguishable from that of Republican men.
Marginal effect of time for Republican women, contraception votes
The evolution of Republican women's voting behavior on contraception issues likely reflects changes in the electoral coalitions of the parties that, in turn, altered the type of Republican woman elected to Congress and changes in public opinion on these issues that reflect partisan efforts to shape public perceptions in their favor. According to VanSickle-Ward and Wallsten (Reference VanSickle-Ward and Wallsten2019), polling through the early 2000s demonstrated that the public perceived contraception as access to health care rather than through the lens of morality politics. General Social Survey results in the 1970s and 1980s indicated that more than 90% of Democrats, Republicans, and independents believed that women should have access to information about birth control. This broad public support and the strong connection to women's health might have spurred some Republican women to oppose the NRLC, an important party interest group, on votes related to family planning. Thus, in the 104th Congress, 53% (9) of Republican women compared with 20% (44) of Republican men opposed eliminating Title X family planning funding and redirecting it to other health programs for the poor.Footnote 14 As late as 2008 (110th Congress), when Mike Pence (R-IN) offered the first of multiple amendments to deny Title X family planning funding to Planned Parenthood, 28.6% (6) of Republican women opposed it compared with only 9.4% (17) of Republican men.Footnote 15
During the Obama administration, the landscape changed. The influence of women's groups grew. EMILY's List and Planned Parenthood played major roles in supporting Obama and congressional Democrats, and they were a driving force behind the “War on Women” message that helped reelect Obama in 2012 and continues to be a part of Democratic messaging. As a result, taking a pro-choice stance has become an even more important prerequisite for Democratic candidates.
On the Republican side, groups such as the NRLC and the Susan B. Anthony List have long opposed federal family planning programs. The money that Planned Parenthood invested in ads portraying Republicans as engaged in a war on women's rights, its president, Cecile Richards, acting as a prominent surrogate for Obama and Democratic congressional candidates, and the money the group donated to Democratic candidates increased the group's association with the Democratic Party and fueled more widespread Republican anger at the group (Swers Reference Swers, Shames and Och2018; Vansickle-Ward and Wallsten Reference VanSickle-Ward and Wallsten2019). Gallup polling indicates that in 1989 and 1993, about 80% of the public had a favorable view of Planned Parenthood, including 69% of Republicans and 83% of independents. While Democrats support for Planned Parenthood remains high, in 2015 and 2018, only 35% of Republicans and 56% of independents viewed Planned Parenthood favorably (Norman Reference Norman2018). Indeed, a Republican strategist who advises moderate candidates lamented that Planned Parenthood's problem “is they have a political arm and they attack Republicans. And they run their 527 as Planned Parenthood so everyone knows who is attacking them. Why do you want to support anything that an organization who wants to defeat you wants?”Footnote 16
Just as women's groups increased their influence within the Democratic Party in the Obama years, social conservatives expanded their power in the Republican coalition. Grassroots anger over Obamacare fueled the rise of the Tea Party. Beyond economics, social issues from abortion to immigration drove conservative anger propelling more strongly committed pro-life members from the South and West into Congress (Deckman Reference Deckman2016; Skocpol and Williamson Reference Skocpol and Williamson2012). The long fight over how broadly to interpret Obamacare's contraception mandate that employer's insurance plans, including religious nonprofits, provide free contraception to their employees went to the U.S. Supreme Court and upended the long-standing bipartisan public consensus on access to contraception. VanSickle-Ward and Wallsten (Reference VanSickle-Ward and Wallsten2019) report that between 2011 and 2014, public polling showed a widening partisan gap in support for the contraception mandate as Democrats responded to party messaging about women's health and Republicans were influenced by party messaging that emphasized religious freedom. Reflecting these changing coalitions, across three votes in the 111th and 112th Congresses (2009–12), only two Republican women opposed amendments to restrict or eliminate funding to Planned Parenthood. By the 114th and 115th Congresses (2015–18), no Republican women and only three Republican men opposed legislation to defund Planned Parenthood. Diane Black (R-TN) led the charge as the sponsor of these proposals. The transformation of contraception into a partisan battleground appears complete.
Taking a closer look at the votes of individual Republican women, the change appears largely driven by replacement rather than conversion. Most of the Republican women serving during the Obama and Trump presidencies were elected since the 2010 Tea Party wave. As Republican women who took pro-choice stances on family planning issues such as Deborah Pryce (OH), Connie Morella (MD), Nancy Johnson (CT), Mary Bono (CA), and Judy Biggert (IL) left Congress, newly elected women from party strongholds in the South and Midwest were strongly pro-life including Diane Black (TN), Vicki Hartzler (MO), Martha Roby (AL), and Renee Ellmers (NC).
Among the few Republican women with long congressional careers, there were individual instances of conversion. For example, Kay Granger (TX) opposed the NRLC on several family planning votes prior to the Obama presidency and the Tea Party wave, including votes related to the development of RU486, the United Nations Population Fund, and defunding Planned Parenthood. She later reversed her position on Planned Parenthood and supported defunding in the four votes taken on the issue during the Obama presidency. Despite her pro-life record, Granger's 2020 primary challenger sought to sow doubt with conservative voters by running an ad highlighting a 2007 interview in which Granger described herself as “a pro-choice Republican.” To counter this attack Granger highlighted her endorsements from the NRLC and the Susan B. Anthony List, but Texas Right to Life still endorsed her opponent (Bowman Reference Bowman2020). The importance of pro-life activists in today's Republican Party demonstrates that the electoral consequence of any defection on the issue can be severe, making it less likely that gender considerations surrounding women's autonomy will drive members’ abortion policy votes. Indeed, Republican women who take moderate stances on other women's issues such as the four Republican women who voted for the Democrats’ 2019 reauthorization of the Violence Against Women Act all maintain a consistently pro-life voting record.Footnote 17 In sum, the changing nature of the party coalitions and increasing power of social conservatives in the Republican Party and pro-choice groups in the Democratic Party has likely reduced the importance of gender as an influence on Republican women's votes on all types of abortion issues.
We also analyze the behavior of Democratic men on various late-term abortion bans. Public opinion surveys demonstrate that the public is ambivalent about abortion and receptive to restrictions (Ainsworth and Hall Reference Ainsworth and Hall2011; Saad 2007). Republicans and their pro-life interest group allies have successfully portrayed late-term abortions as “partial-birth” abortions that were unnecessary and akin to infanticide (Dodson Reference Dodson2006; Swers Reference Swers2013). The conflict over “partial-birth” abortion spanned the 104th–108th Congresses (1995–2004). Pro-life groups have since used similar framing to ban abortion after 20 weeks, to ban sex-selection abortions, and to criminalize providers for harming fetuses born alive. Given the gruesome imagery associated with debates over late-term abortion and their relative popularity with the public, as evidenced by the almost 70% support among the public for a “partial-birth” abortion ban when the legislation was voted into law (Saad 2007), Democratic men in particular might be swayed by the messaging efforts of Republicans and pro-life groups and more likely to support these bans than Democratic women, whose partisan leanings are reinforced by a commitment to reproductive rights based on their social identity as women.
Table 6 reports regression results for the subset of abortion votes related to bans. We do find evidence that Democratic men are cross-pressured by these kinds of bills, and they are less likely than Democratic women to vote against the NRLC position. While Republican men and women are about 30 percentage points more likely than Democratic women to vote in line with the NRLC, Democratic men are about 6 percentage points more likely to do so, even after incorporating a robust set of controls and using Congress and state fixed effects. The impact of having a Democratic man for a representative is slightly smaller than the influence of having a Catholic representative, as Catholic legislators were 8% more likely to vote with the NRLC on late-term abortion bans.Footnote 18
Likelihood of supporting NRLC position, abortion ban votes (Democratic woman omitted)
“Partial-birth” abortion was a particularly successful policy campaign for Republicans. The term “partial birth” is not a medical term and in fact was coined by the NRLC (Rovner Reference Rovner2006). The imagery of infanticide favored the Republican Party's focus on the vulnerability of the unborn, while Democrats struggled to refocus attention on women's rights and issues related to the health of the mother. Sellers (Reference Sellers2010) notes that Republicans were much more likely than Democrats to promote the party message on the issue through press conferences and floor speeches, and that Republicans dominated press coverage of the issue. In contrast, Democrats shied away from engaging with the issue fearing that the position of pro-choice groups would alienate their broader electorate. Furthermore, Catholic Democrats were particularly worried about the views of the church (Sellers Reference Sellers2010; Swers Reference Swers2013).
Public opinion was very responsive to the gruesome imagery of partially delivered fetuses and the rhetoric of infanticide. According to Gallup, public support for a partial-birth abortion ban increased from 57% in 1996 to close to 68% in 2003, and American National Election Studies data indicated that public support for the ban rose from 56% in 1997 to 70% in 2000. The increase in support was largely driven by people who identified as pro-choice (Freedman Reference Freedman1997, Reference Freedman2003). Ultimately, several high-profile Democratic men voted for the ban including Minority Leader Richard Gephardt (D-MO) and Minority Whip David Bonior (D-MI). Examining the first votes on the issue in the 104th Congress demonstrates the discomfort of Democrats, particularly of Democratic men. Twenty-four Democratic men and two Democratic women voted with their party on the rule for debate that would not allow amendments, a party loyalty vote, while also voting with the NRLC position in favor of the ban.Footnote 19 In sum, Republicans owned the partial-birth abortion issue. The nature of the debate favored the Republican focus on the life of the unborn. Republican party messaging moved public opinion strongly toward their position. Democrats, particularly Democratic men, felt pressured to align with public opinion rather than with the position of the women's rights groups that are a key component of the party coalition.
Looking beyond the two larger categories of contraception and family planning issues and late-term and other abortion bans that were the subject of our regression analyses, we can see similar qualitative patterns among our other policy area categories, some of which contain too few votes or votes not sufficiently spread across our period of study to allow for quantitative analysis. Looking descriptively, policy categories of issues that highlighted women's autonomy drew defections from Republican women, and policy categories of issues focusing on the vulnerability of children yielded more defections from Democratic men.
For example, compared with votes related to abortion bans, votes concerning abortions in federal facilities, particularly military bases, more readily drew defections by Republican women. On the 15 votes on amendments to Defense Appropriations and Authorization bills taken during the 104th–109th (1995–2006) Congresses, a subset of Republican women opposed the NRLC position and supported proposals that would allow women in the military to obtain abortions at overseas military bases utilizing their own private funds. Republican women also opposed complete bans on access to abortion at military bases and bans with only limited exceptions for rape, incest, and the life of the mother.Footnote 20 Women in the military are a highly sympathetic population, and these votes were seen as potentially jeopardizing the ability of service members to access medical care.
While largely loyal to the party on votes related to contraception, Democratic men defected more frequently on a set of proposals that would make it a federal crime to transport a minor across state lines to obtain an abortion in order to circumvent a state's parental consent law. Members took 11 votes on versions of this legislation and related amendments in the 105th, 106th, 107th, and 109th Congresses. Across these votes, nearly one in three Democratic men supported the NRLC's position compared with one in seven who defected on contraception votes, likely responding to Republican arguments concerning parental rights and the protection of young girls from exploitation by sexual partners or other adults (Saul Reference Saul1998).
An area that saw a significant number of defections from both Republican women and Democratic men was the debate over the Unborn Victims of Violence Act, which created separate punishments for harming a fetus during a federal crime against a pregnant person. However, Democrats and abortion rights groups argued that the legislation sought to give personhood rights to the fetus to undermine Roe v. Wade. The bill passed the House in both the 106th and 107th Congresses and became law in the 108th Congress. The gruesome murder of Laci Peterson by her husband gave additional momentum to the legislation and Republicans ultimately renamed the bill Laci and Conner's Law (Perine Reference Perine2003). Across the six votes taken on the bill and Democratic alternative proposals, about a quarter of Democratic men defected from their party and supported the NRLC position, likely persuaded by the need to protect unborn children who are injured or killed in a federal crime. By contrast, nearly 40% of Republican women defected from their party and opposed the NRLC position, likely responding to the argument that the bill would harm women's autonomy by threatening the right to an abortion established in Roe v. Wade.
Clearly, the nature of an issue and the ability of parties and interest groups to frame arguments in ways that galvanize public opinion have an impact on how members vote. A significant group of Democratic men responded to broadly popular arguments about punishing criminals who attacked pregnant women and their unborn children and the need for parental consent to protect young girls. Meanwhile, some Republican women were motivated to oppose the NRLC when they felt women's autonomy was threatened by prohibiting women from accessing abortions at overseas medical facilities and potentially striking at the right to an abortion by creating personhood rights for a fetus harmed in a federal crime.
CONCLUSION
Going forward, abortion politics will likely remain a staple issue on the congressional agenda. This project is the first to examine congressional voting behavior on this issue over an expansive, 25-year period. We demonstrate that gender has shaped the dynamics of voting on the issue but find a nuanced relationship between gender and abortion-related voting behavior over time. First, the importance of gender as a predictor of voting on policies related to abortion has diminished over time. From the early 1990s to the final years of the George W. Bush presidency, Republican women were significantly more likely than Republican men to oppose pro-life initiatives. The more recent convergence of Republican men and women's voting records on abortion is largely explained by replacement, as strongly pro-life women from the South and Midwest replaced their more ambivalent counterparts. There were also a few instances of conversion in which longer serving Republican women aligned their voting with the majority position of the party and the social conservatives who are a key element of the party base. In contrast, apparent gender differences among Democrats are largely explained by the fact that Democratic women are elected from more liberal districts.
Although public opinion varies by abortion policy issue, no existing studies distinguish between types of abortion policies when examining the voting behavior of members of Congress. Our research demonstrates the importance of issue types and demonstrates that not all abortion votes are the same. The policy content of particular abortion debates impacts the importance of gender on members’ votes. The perception of access to contraception as significant for women's autonomy may have influenced Republican women and contributed to their increased support for initiatives related to family planning through the mid-2000s. Moreover, the nature of an issue impacts public perceptions of a policy and influences party messaging strategies as partisans and their interest group allies try to move public opinion by encouraging the broader electorate to view an issue through a lens that is favorable to their party's issue strengths.
As pro-choice and pro-life forces have gained a more dominant position within the two parties’ electoral coalitions, it will be increasingly important to focus on how issues emerge on the agenda and how each party works to shape the definition of abortion-related policy. Women in each party are likely to play key roles in determining the policy parameters of the abortion policies that ultimately reach the floor. State-level research indicates that Republican-controlled states with more pro-life organizations, stronger evangelical movement influence in their state governments, and more Republican women in the party caucus pass more restrictive abortion policies (Bentele, Sager, and Aykanaian Reference Bentele, Sager and Aykanaian2018). By contrast, the presence of more Democratic women in the legislature reduces the probability of adoption of some types of restrictive legislation (Bentele, Sager, and Aykanaian Reference Bentele, Sager and Aykanaian2018; Kreitzer Reference Kreitzer2015; Tatalovich and Schier Reference Tatalovich and Schier1993). Moreover, Reingold et al. (Reference Reingold, Kreitzer, Osborn and Swers2020) find that conservative Republican women are more likely to sponsor bills that are framed as protecting women, such as the recent wave of “TRAP” (targeted restrictions on abortion providers) laws that require abortion clinics to meet standards including requiring clinic physicians to have hospital privileges. As more liberal Democratic women and conservative Republican women are elected to Congress, we can expect that their gendered life experiences will inform how they define these issues and that women in each party will play pivotal roles in party messaging efforts to sway the public to support their policy vision.
