Party reputations are key to understanding the public’s voting decisions and their more durable party identities in the United States (Campbell et al. Reference Campbell, Converse, Miller and Stokes1960; Ondercin Reference Ondercin2017; Klar Reference Klar2014; Klar and Krupnikov Reference Klar and Krupnikov2016). Voters’ views of what policies, groups, and values each party represents usually move slowly over time (if at all) because substantial elite-level changes are rare, and what changes do occur can take decades to develop fully (Cohen et al. Reference Cohen, Karol, Noel and Zaller2009; Petrocik Reference Petrocik1981). As a consequence, partisan-social sorting in the public and broader partisan realignments tend to be slow-moving and incremental (Mason Reference Mason2018; Valentino and Sears Reference Valentino and Sears2005).
But mass partisanship does change individually and collectively, and elite party shifts can motivate partisans to change parties in notable numbers. In their largest form, “issue evolutions” (Carmines and Stimson Reference Carmines1990; Adams Reference Adams1997) begin when political events (elite-led or otherwise) dramatically increase the salience and importance of an issue, and the parties respond with clearly diverging positions on the issue. Attentive citizens receive those signals and update their views of the parties and the issue. Partisans whose views misalign with their own party can either update their views or realign their votes, and some voters may even begin to adjust their long-standing partisan identities (Carsey and Layman Reference Carsey and Layman2006). These changes may focus on a singular issue, as in canonical cases of issue evolution like race, or they may occur across many issues at once in a broad process of “conflict extension” (Layman and Carsey Reference Layman and Carsey2002a).
Building on these paradigms of dynamic partisanship, we focus on more rapid changes in party voting and identification that may occur when those conditions correspond with strategic issue ownership by parties. Voters’ views of party issue ownership change in response to salient events, policy performance, party messaging, and elite political behaviors (Petrocik Reference Petrocik1996; Dittmar Reference Dittmar2015; Egan Reference Egan2013). When parties engage in strategic issue-based action to capture voters and satisfy party constituencies, divergence between party elites clarifies public views of party positions, but it can also change views of party issue ownership. When those two reputational changes coincide, we expect to see accelerated shifts in voting and party attachments related to that issue.
Put a different way, issue ownership research (Egan Reference Egan2013) addresses party positions, performance, and emphasis, but it focuses less on implications for partisanship itself. At the same time, issue evolution and conflict extension literatures do not situate issue ownership in their theories of partisan change. Clearly, the two areas are related but they have not been brought fully into conversation. We bring all three perspectives together to study the vital case of #MeToo partisanship and to understand how a new issue became politically salient and polarized along party lines. Our results provide important insights supporting the three strands of research alone and in combination. We inform those theoretical perspectives without attempting to replace them.
We test our expectations for change in partisan views, votes, and identities in a new domain: sexual misconduct and #MeToo.Footnote 1 First, using survey data from the same question asked in 1991, 2016, and 2018, we show that sexual misconduct only recently became an issue that one party owns. Sexual misconduct by candidates and officials is far from new (e.g., Clarence Thomas, Bill Clinton, or Herman Cain), but the public only recently shifted from party balance towards disproportionate Democratic ownership of the issue (Hansen and Dolan Reference Hansen and Dolan2020; Klar and McCoy Reference Klar and McCoy2021a). Consistent with our expectations, observational patterns and results from a survey experiment show that elite party responses to the Kavanaugh confirmation shifted public views towards Democratic ownership and away from Republican ownership on handling sexual misconduct well. Voters respond to elite actions by adjusting their views of party issue performance (Egan Reference Egan2013).
According to issue evolution and conflict extension theories, shifting party reputations on sexual misconduct should produce a ripple effect causing increasingly polarized views among ordinary partisans in each coalition. Using dozens of survey questions about sexual harassment from 1986 to 2018, we show that while the parties have asymmetrically reacted to sexual misconduct at multiple points over the last forty years, misconduct-related attitudes among party members did not polarize until 2016. The key difference, we argue, is the emergence and persistence of differences in issue prioritization.
Misaligned views could cause shifts in voting and even party identity among partisans who consider the other party’s position as preferable and care strongly about the issue. Looking at vote choice in the 2018 election, we show that partisans who saw their own party performing worse on the issue reported much higher levels of vote defection. Finally, using data from a survey conducted after the Kavanaugh hearings, we find substantial changes in otherwise durable partisan identification among people whose views of #MeToo were incongruent with their party’s position.
Both parties have perpetrators of sexual misconduct and the problem is not unique to the #MeToo era. Yet issue ownership, asymmetric party positions, and increasing mass clarity on those party differences result in sexual misconduct playing a rapid polarizing role in partisan behavior and attachments. Building on work that evaluates the role of social identities in partisanship (Mason Reference Mason2018) and shifts in gendered attachments to the parties (Ondercin Reference Ondercin2017), we demonstrate the centrality of elite positioning, issue ownership, and issue salience to shifts in the mass public’s partisan preferences.
Partisanship and Issue Ownership
Partisanship is usually the most powerful orienting force in mass politics, guiding vote choice, policy preferences, political perceptions, and political action (Bartels Reference Bartels2002; Jacobson Reference Jacobson2013). Party identification largely arises from parent socialization—when parents have the same party attachments, their children tend to grow into adulthood with the same—and that identification endures across the lifespan for most people (Jennings, Stoker, and Bowers Reference Jennings, Stoker and Bowers2009). Policy views related to prominent social groups are more stable than most (e.g., Converse Reference Converse and Apter1964), but party leaders often guide party followers with cues even on these (Barber and Pope Reference Barber and Pope2019; Karpowitz, Monson, and Preece Reference Karpowitz, Monson and Preece2017; Layman and Carsey Reference Layman2002b). These views of norms can shift in cyclical patterns across time (Finnemore and Sikkink Reference Finnemore and Sikkink1998). Broader social identities and contexts matter too: the “funnel of causality” model of partisan vote choice places broad social divisions and group loyalties prior to party attachments, issues, and candidate evaluations (Campbell et al. Reference Campbell, Converse, Miller and Stokes1960). Beyond early-life partisan socialization, those group orientations pave a latent path for potential partisan change later in life.
Party reputations related to prominent social groups are especially important to partisanship as leaders cue relevant identities and attitudes, given the scarcity of more ideologically principled motivations in the public (Campbell et al. Reference Campbell, Converse, Miller and Stokes1960; Kinder and Kalmoe Reference Kinder and Kalmoe2017; Klar Reference Klar2013; Ondercin Reference Ondercin2017). These social groups and other interests strongly influence party platforms (Freeman Reference Freeman2002; Kittilson Reference Kittilson2011), but groups may also decouple from a party that ignores their support, or groups may even realign when the opposing party moves to represent their interests better (Bishop Reference Bishop2009; Mason Reference Mason2018; Ondercin Reference Ondercin2017).
Partisan predominance does not make party loyalties impervious to change. As political problems rise and fall in salience, parties choose alternative paths in addressing those issues (Petrocik Reference Petrocik1996). These shifts in elite preferences and clarity on issues in each party’s coalition can change vote choice and even party identity. We know sociopolitical contexts influence party reputations and public preferences. For example, the gender gap in voting is shaped by the gendered context, party behavior, and representation within the parties (Ondercin Reference Ondercin2017; Cassese and Barnes Reference Cassese and Barnes2018; Ondercin and Lizotte Reference Ondercin and Lizotte2020) and gendered events during and within campaigns (Ondercin and Bernstein Reference Ondercin and Bernstein2007; Sapiro and Johnston Conover Reference Sapiro and Conover1997; Cassese and Holman Reference Cassese and Holman2019). Sometimes these shifts are temporary and last for a single election, while other shifts occur slowly over decades (Ondercin Reference Ondercin2017). But sometimes issue-related change among party leaders reshapes mass partisanship in a relatively short period among many who care strongly about that issue.
Issue evolution involves four steps: 1) changes in elite positions on the issue, 2) clarity to the general public as to the differences between the parties on the issue, 3) shifts in affect towards each party along those lines, and 4) a mass realignment in response. In the most prominent example, dramatic changes in national party reputations on racial issues realigned white Southerners from Democratic dominance to large Republican majorities (Achen and Bartels Reference Achen and Bartels2016; Piston Reference Piston2010; Kuziemko and Washington Reference Kuziemko2018; Valentino and Sears Reference Valentino and Sears2005; Carmines and Stimson Reference Carmines and Stimson1986, Reference Carmines1990). Individuals may also move against their party habits based on group-related predispositions when those issues are salient (e.g., Piston Reference Piston2010; Kalmoe and Piston Reference Kalmoe and Piston2013; Tesler Reference Tesler2016; Cassese and Barnes Reference Cassese and Barnes2018; Cassese and Holman Reference Cassese and Holman2019).
Layman and Carsey’s (Reference Layman and Carsey2002a, Carsey and Layman Reference Carsey and Layman2006) theory of “conflict extension” involves similar steps but elite and mass partisan polarization spreads across many issues simultaneously rather than one focal issue rising up to replace another. Conflict extension highlights the prevalence of opinion leadership—in which many partisans learn of and adopt their party’s position on the new issue—as an alternative to changing party identification among those who prioritize an issue and who disagree with their party about it. #MeToo partisan changes probably involve both types.
Bringing In Issue Ownership
Party elites can find strategic advantage in “borrowing” particular issues during an election and responding diametrically to them (Petrocik Reference Petrocik1996; Pope and Woon Reference Pope and Woon2009). In the original formulation of issue ownership, party reputations are “regularly tested and reinforced” by both context and performance (Petrocik Reference Petrocik1996, 828). For example, voters saw the Republican Party of the early 2000s as more capable in foreign affairs. After September 11, 2001, members of the party received a boost from the party’s ownership (Merolla and Zechmeister Reference Merolla and Zechmeister2009) and voters evaluated the party’s performance by how they handled national defense issues (Pope and Woon Reference Pope and Woon2009). Parties and campaigns also strategically leverage issue ownership reputations in an attempt to “wedge” groups of voters away from their opponents (Hillygus and Shields Reference Hillygus and Shields2008). More generally, changes in party issue ownership can shift votes in elections, but the issue must be salient to voters for it to influence their behavior (Bélanger and Meguid Reference Bélanger and Meguid2008), and one party must be seen as distinctively better on the issue over the other (Guinaudeau and Persico Reference Guinaudeau and Persico2014).
In his central work, Petrocik (Reference Petrocik1996) suggests issue ownership is generally more fleeting than issue evolution as parties “borrow” (and return) issues from election to election, and it will not lead to a realignment of members from one party to another if party action only cements preferences that already exist within the parties (thereby leaving affect towards the parties unchanged). However, we posit that the issue ownership process can contribute to an accelerated sequence of partisan change.
Building out the roots of party issue ownership, Egan (Reference Egan2013) identifies three potential influences on those judgments. On controversial issues, like abortion, he finds that people tend to perceive party ownership based on a match with their policy view. On less controversial issues, people are more likely to regard the party that publicly emphasizes and prioritizes the issue most as the one that owns the issue. Actual performance in government on the issue seems to have little to do with ownership, contrary to conventional characterizations of the concept.
Our Integration
We combine these three party-issue perspectives for tests in the critical case of #MeToo and partisanship. The opportunity to take ownership of a newly salient issue may motivate elite position taking and action that supplies more clarity to the public, and those clarified public perceptions of reputations provide momentum motivating the parties to commit more to that issue ownership in a reciprocal relationship. Issue ownership related to groups may have the most potential in this regard.
Issue ownership intersects with issue evolution and conflict extension as one part of a party’s broader policy reputation. The party’s reputation can work to sort and lead voters when that reputation is clear to voters. When an issue becomes salient through proactive decisions by a party or in reaction to an event, this amplifies the divergent positions of the parties. It also provides clarity for partisans on what their preferences should be on the issue (Layman and Carsey Reference Layman and Carsey2002a; Carsey and Layman Reference Carsey and Layman2006).
The traditional theory of issue ownership is about long-term reputations, but voters’ views of those reputations may shift more rapidly due to salient events (e.g., #MeToo) that open opportunities for strategic shifts by party leaders. In the case of #MeToo, Democratic moves to own the issue of sexual misconduct may have been aided by the party’s reputation for being better on “women’s issues” in general (Ondercin Reference Ondercin2017; Winter Reference Winter2010), facilitating a quicker reputational shift on sexual misconduct specifically. We do not claim the issue’s rise was entirely exogeneous from politics, but it was relatively more exogeneous than most political issues given that #MeToo originally gained salience in entertainment, not politics. Democratic elites in the #MeToo era then moved to establish a clear reputation after those events, and we show that the public did not perceive that reputation clearly until Democratic leaders acted. Indeed, the Democratic Party’s defense of Bill Clinton’s sexual misconduct in the 1990s and 2000s and nomination of John Edwards as a vice-presidential candidate show the relatively recent shift in how elites in the party viewed issues of sexual harassment. Our ideas about issue ownership have portability to a wide-range of situations where parties react to events that are out of their control—such as economic downturns or international events.
We focus on the consequences of those reputational changes in the public, integrating issue ownership into models of issue-based partisan change. In this environment, partisans who prioritize the issue, whose party does not own the issue, and who disagree with their party’s position will shift their affect towards the parties and rapidly realign to the other party—made more rapid by the ownership mismatch. Figure 1 illustrates our expectations for this process and provides the structure for our empirical tests.
Issue ownership, issue evolution, and rapid realignment
We test our expectations in the domain of sexual misconduct. Beyond our efforts at theoretical integration, we consider our novel investigation into several facets of #MeToo partisanship to be an important contribution all its own.
Several key events during the 2016 presidential election campaign and subsequent years pushed the parties to contest issue ownership over sexual misconduct: the candidacies of Hillary Clinton and Donald Trump, the sudden prominence of the #MeToo movement, several high-profile political resignations associated with sexual misconduct, and the nomination and confirmation of Brett Kavanaugh to the U.S. Supreme Court, despite sexual assault allegations against him. Testing the process of rapid realignment, as set forth in figure 1, we present a wide range of public-opinion evidence on sexual misconduct to show how elite-level party statements and actions clarified party positions, with consequences for views of the parties, voting, and party membership.
Elite Positions. We start by describing how political elites engaged with the issue around the 2016 election, #MeToo, and the Kavanaugh hearings. Recall that we expect issue ownership will push party elites to differentiate themselves (line A in figure 1) and will provide clarity to voters about the differences between the parties in performance on the policy (line B in figure 1).
Clarity. We expect voters will recognize the shifts in party images, including issue ownership and positions. To test increased clarity, we examine perceptions of party ownership over time. We expect to find that the share of Americans who will favor the Democratic Party’s performance in handling sexual misconduct will rise in absolute terms starting in 2016 (Hypothesis 1a) and that Democrats will gain on that dimension relative to the Republican Party (Hypothesis 1b). We explicitly test sequence C in figure 1 using experimental data to evaluate if signals from elites produce clarity in issue ownership among voters, seeing the Democratic Party as comparatively better at handling sexual misconduct policy (Hypothesis 1c).
Issue evolution also requires shifts in how much the public cares about the issue (Carmines and Stimson Reference Carmines1990), patterns that can shift asymmetrically across groups (Ondercin and Lizotte Reference Ondercin and Lizotte2020). By contrast, conflict extension simply requires clear public signals on the issue from party leaders (Layman and Carsey Reference Layman and Carsey2002a). We expect no partisan differences in viewing sexual misconduct as a problem prior to 2016 due to a lack of party clarity (Hypothesis 2a). Starting in 2016 a partisan gap in views of sexual misconduct will emerge, as Democratic partisanship becomes associated with more concern about sexual misconduct than will Republican partisanship (Hypothesis 2b).
Affect. Next in the sequence, preferences around the issue should involve “a heavy dose of affection and disaffection for the parties” (Carmines and Stimson Reference Carmines1990, 161). In adjacent work, Ondercin and Lizotte (Reference Ondercin and Lizotte2020) find that women report more partisan affective polarization than men, partly due to the strength of their abortion-related attitudes. We expect that Americans who view sexual misconduct with more (less) favorability will have warmer (colder) feelings toward the Democratic Party, with the opposite pattern for Republicans (Hypothesis 3).
Vote and partisan change. Beyond party reputations, the combination of shifts in ownership, asymmetric elite reactions, and clarity of views among partisans should shape vote choice and party identity. We anticipate that Democrats will be more likely than Republicans to report that a candidate’s stance on sexual misconduct is important in their vote choice (Hypothesis 4a). Individuals who see the opposing party as better able to handle issues of sexual misconduct will switch their vote to that party (Hypothesis 4b). Finally, people whose views on sexual misconduct misalign with their party will defect to the more amenable party; we expect that those who see #MeToo favorably (unfavorably) are more likely to become Democrats (Republicans) (Hypothesis 4c).
Data and Methods
We use the increased salience of sexual misconduct starting in 2016 as a case study to test our expectations. In doing so, we present evidence from a lab experiment, data from twenty-nine surveys from 1986 to 2018, and more detailed and nuanced analysis from three nationally representative surveys (2018 Kaiser Family Foundation Health Tracking Poll, 2018 Data for Progress survey, and the 2018 Cooperative Congressional Elections Study). The online appendix contains details of all the data.
Survey Experiment. To test the effects of party responses to sexual misconduct in a highly prominent real-world case, we fielded a survey experiment about the Brett Kavanaugh Supreme Court nomination with 1,724 participants provided by Lucid on Sept. 19, 2018. This was two days after the Senate scheduled a new hearing to investigate misconduct allegations with testimony from him and his accuser, Dr. Ford. The experiment had three conditions: a control without a news article, a basic news story, and the same story with details mentioning each party’s response to the allegations.
Cooperative Congressional Elections Survey (CCES). We use a team module from the 2018 CCES to evaluate party reputations on the issue of sexual misconduct and whether party reputations about sexual misconduct relate to party vote defection. The CCES was conducted online by YouGov, with common content questions asked of 60,000 individuals both pre- and post-election. The questions from our team module with 1,000 respondents include views of which party is better able to handle the issue of sexual harassment, which party is better able to punish their members for sexual misconduct, and vote choice, along with other demographic and political controls.
Kaiser Family Foundation Health Tracking Poll (KFF). We examine partisanship and candidate preferences within the frame of sexual misconduct using data from a KFF telephone poll, fielded June 11–20, 2018 answered by a nationally representative sample of 1,492 respondents. Social Science Research Solutions conducted the survey, which includes a question on the importance of a candidate’s support for the #MeToo movement for the respondent’s vote choice, along with political and demographic measures.
Data for Progress Survey (DFP). We test the longer-term consequences of #MeToo attitudes on affect towards the parties and party switching with a nationally representative midterm election survey of 2,954 individuals collected by DFP in October 2018, a survey conducted online through YouGov Blue. The survey included questions evaluating the #MeToo movement, sexism, party identification, party favorability ratings, and change in partisan support, along with political and demographic measures.
Surveys conducted over time. We analyze forty-four questions about sexual harassment from surveys conducted from 1986 to 2018; data was obtained from the Roper Center at Cornell University. We also look at changes in responses to two sets of questions that were asked in a similar manner over time to examine responses to the question “Which party does a better job of handling the issue of sexual harassment?” A full list of all surveys and questions is available in the online appendix.
Issue Ownership of Sexual Misconduct
Our tests of change in the issue ownership component of sexual misconduct proceed in three parts: 1) a description of diverging partisan elite behavior and rhetoric around the issue, 2) survey evidence of recent mass shifts in issue ownership perception over decades, and 3) a combination of natural and survey experiments providing causal evidence of elite behavior and rhetoric as sources of shifts in public perceptions of party reputations.
The 2016 Presidential Election
The 2016 presidential candidates and their campaign choices raised the salience of sexual misconduct and pushed the parties to contest issue ownership over specific domains within the issue. On the Democratic Party’s side, Hillary Clinton’s historic presidential campaign as the first woman atop a major political party’s ticket in U.S. history included an explicit focus on her “first woman” status and on feminine issues (Wilz Reference Wilz2016; Conroy Reference Conroy2018; Heldman, Conroy, and Ackerman Reference Heldman, Conroy and Ackerman2018).Footnote 2 At the same time, Donald Trump’s nomination to the Republican Party ticket, particularly his displays of “hyper” masculinity in the primaries (Conroy Reference Conroy2018) and his gendered attacks on women in journalism and on Clinton (Cassese and Holman Reference Cassese and Holman2019; Heldman, Conroy, and Ackerman Reference Heldman, Conroy and Ackerman2018), provided a clear foil to Clinton’s woman-focused campaign.
Like many political leaders before him, Trump’s campaign was haunted by the specter of allegations of sexual misconduct, prompted by accusations from several women including a May 2016 article from the New York Times that detailed allegations of “unwelcome romantic advances, unending commentary on the female form, a shrewd reliance on ambitious women, and unsettling workplace conduct” (Barbaro and Twohey Reference Barbaro and Twohey2016). The October 2016 release of a 2005 Access Hollywood video where Trump bragged about sexual assaulting behavior and used derogatory language towards women (Farenthold Reference Farenthold2018) raised the salience of sexual misconduct even more. Clinton’s campaign capitalized on the allegations, attacking him in debates and crafting ads in a move to own the issue in the 2016 election. Trump responded by bringing women who accused Clinton’s husband of sexual misconduct to the debate. Research on voting behavior in 2016 suggests that the Access Hollywood tape galvanized some opposition to Trump, but may have also solidified support for him among some Republicans (Cassese Reference Cassese2019; Klar and McCoy Reference Klar2021b).
#MeToo, Kavanaugh, and the Parties
Events following the 2016 election continued to raise the profile of sexual misconduct. In October 2017, a viral hashtag (#MeToo) emerged following abuse allegations against filmmaker Harvey Weinstein (Farrow Reference Farrow2017; Lee and Murdie Reference Lee and Murdie2020). The hashtag built off the MeToo movement that Tarana Burke founded in 2006 to provide survivors of sexual violence—particularly Black girls and women—with opportunities to access resources and create solutions to sexual violence (Burke Reference Burke2019). The #MeToo hashtag gained national and international attention in 2017, largely without support from political elites. Promoted by several celebrities, many of whom suffered sexual abuse themselves, the movement succeeded in bringing attention to sexual misconduct by other entertainment moguls, sports figures, and members of college faculty. The #MeToo movement eventually spread to politics, revealing sexual harassment and assault by leaders in Congress and state legislatures (Gessen Reference Gessen2017), with several states passing legislation to strengthen workplace protections (Rampell Reference Rampell2018). Several high-profile political leaders eventually resigned because of sexual abuse allegations, including Democratic Congressman John Conyers and Senator Al Franken. In Franken’s case, pressure from public outrage and Democratic women in the Senate, who unanimously called for his resignation, presents an interesting contrast with Republican responses to #MeToo events. This action showed Democratic prioritization beyond position-taking, sending a strong public signal of commitment on the issue, with major implications for ownership views (Egan Reference Egan2013). In many ways, the rise of #MeToo follows patterns of other international norms diffusions (Htun and Weldon Reference Htun and Weldon2012), where political elites take up the issue after it is sufficiently “ripe” from actions by non-political actors (Lee and Murdie Reference Lee and Murdie2020; Finnemore and Sikkink Reference Finnemore and Sikkink1998).
Brett Kavanaugh’s nomination pushed the parties to take distinct positions on sexual misconduct yet again. President Trump nominated Kavanaugh to fill an empty seat on the U.S. Supreme Court in early July 2018, with hearings scheduled for September. In mid-September, Dr. Christine Blasey Ford came forward with sexual misconduct allegations against Kavanaugh; specifically, she alleged that Kavanaugh attempted to sexually assault her in high school by locking her in a room and pinning her on a bed despite her verbal and physical resistance (Britzski Reference Britzski2018). Another accuser emerged on September 23, 2018. On September 27, 2018, both Kavanaugh and Ford testified before the Judiciary Committee in the Republican-controlled Senate. After a delay for a brief investigation, the Senate voted for Kavanaugh’s confirmation on October 6, 2018, and he assumed a lifetime appointment on the U.S. Supreme Court shortly thereafter.
The Kavanaugh hearing thus became another flash point for public discussions of #MeToo issues in politics, with the Republican Party reluctant to address issues associated with sexual assault, given the importance of the seat on the Supreme Court, and the Democratic Party portraying the Republican Party’s acts as evidence of a general acceptance of sexual misconduct by its members. In mid-October 2018, 62% of those surveyed by Gallup said that Kavanaugh’s confirmation would be important to their vote in the 2018 midterm Congressional elections.Footnote 3
Clarifying Party Ownership on Sexual Misconduct
In our account, issue ownership motivates diverging elite party positions and subsequent clarity for the public about party differences, even as ownership is also a product of those shifts. To evaluate clarity, we start by examining which party Americans view as more capable of handling issues associated with sexual misconduct, using all available data on the question. Table 1 presents the share of respondents in surveys from 1991, 2017, and 2018 that view the Republican or Democratic Parties (or both or neither) as better able to handle the issue of sexual harassment.
Party reputations on sexual harassment
Note: Share of respondents who identify the Democratic Party (row 1), Republican Party (row 2), or both or neither (row 3) as better at handling the issue of sexual harassment. Percentages in table. For simplicity’s sake, we have condensed the neither, both, no difference, not sure responses into a single category. Refer to online appendix 1A for question wording.
Consistent with Hypothesis 1a, survey respondents see the Democratic Party in 2018 as more capable of handling sexual harassment policy (˜50%) as compared to 1991 (25%) or even 2017 (34%); we also find that a larger share of respondents see the Democratic Party as owning the issue, as compared to the Republican Party, consistent with Hypothesis 1b. The relative Democratic advantage is twelve points in 1991, twenty-four points in 2017, thirty points in early 2018, and fourteen points in late 2018. The latter decline in Democratic advantage reflects a 2018 growth in respondents favoring the Republican Party’s handling of the issue and not a decline in Democratic ownership. This suggests that views of party ownership of sexual harassment policy may also be subject to defensive partisan responses and negative partisanship among Republicans as the issue gained salience (Bankert Reference Bankert2020a). In sum, we find longstanding Democratic ownership of the sexual misconduct issue, but we also find substantially stronger party ownership in recent years.
Clarity from Elite Messaging
We next test whether party messaging clarifies party reputations on the issue in the public’s view. To assess, we use a randomized survey experiment around the Kavanaugh hearings. The experiment directly tests the effects of elite misconduct cues on related party reputations in the public.
Our design includes a true control condition (no information), a basic information treatment (which describes the allegations but provided no party-specific information), and a party treatment (which provided information about each party’s reactions to Kavanaugh). The party treatments headline read “Republicans and Democrats scramble to respond to sexual assault allegations against Supreme Court nominee.” The text noted that Republican leaders intended to “move forward with voting on Kavanaugh” and that Democrats, including “Senator Dianne Feinstein, the top Democrat on the Judiciary Committee” saw the accusations as serious and wanted to delay the hearings (refer to online appendix A2 for the full treatments). Treatments were based on newspaper coverage.
We asked respondents to identify which party was better at handling sexual harassment policy and at punishing its own offending members. We use two measures to try to understand a broader evaluation of how elite behavior might shape partisanship: one path is via policy (as we try to capture in our first question) but another might be how the party acts towards bad behavior within its own ranks. As an example, in a survey of news coverage of how state parties reacted to sexual misconduct claims by their members, we found that the parties recommended punishment or resignation in less than half of the cases. Variation between the parties over punishing their members primarily emerged in the post-2016 era. We drew on these (and other, more high profile) examples and asked respondents about their views of the parties’ ability to punish their own members for bad behavior.
For clarity, we estimate a multinomial logit model for our tri-part measure (which party is better: Democrats, Republicans, or both) and present the post-hoc predicted probabilities from those models, predicting movement towards Democrats or Republicans (see figure 2 and online appendix 2A). Because party identification is unbalanced by chance in our randomized treatments, we control for party in our models. In the online appendix, we investigate the effects of the treatments across partisan strength, showing that all partisan views are shifted by the party cues treatment in the same direction—reducing the view that Republicans are better and increasing the view that Democrats are better. These effects occur in a normal curve shape, so that strong Democrats and strong Republicans (whose views were already polarized on the issue) are least moved by the treatment and weak partisans are the most moved (refer to online appendix 2B).
Kavanaugh hearings and issue ownership (Survey Experiment)
Note: Posthoc predicted probabilities. Party identification is included as a control. See online appendix 2A for full model. Error bars indicate 95% confidence intervals.
Participants in the control condition generally favored the Democratic Party on handling policies and punishing leaders for sexual misconduct, confirming the survey. The party treatment (detailing how party leaders responded to the Kavanaugh accusations) leads respondents to rate the Democratic Party as significantly better at handling the issue of sexual harassment, as compared to the control condition (p < 0.05, one-tailed test), but the treatment did not significantly improve Democratic evaluations on punishing leaders (p < 0.62). Evaluations of the Republican Party decline among those in the party treatment compared to the control group for handling policy (p < 0.05, one-tailed test) and punishing leaders (p < 0.05, one-tailed test). In terms of relative issue ownership, the Democratic policy advantage grows from eleven points in the control to twenty-three points in the party treatment, and it grows more modestly for in-party policing from nine points to fifteen points, respectively. The differing effects of the party information treatment versus the basic treatment show the importance of not just issue salience, but also messaging from party elites. The basic treatment produces small, insignificant shifts among respondents towards viewing the Democratic Party as owning the issue and away from seeing the Republican Party owning sexual misconduct.
We have presented two types of evidence that indicate shifting views of party ownership over sexual misconduct (surveys and a survey experiment) to demonstrate “clarified mass cognitive images of the parties” (Carmines and Stimson Reference Carmines1990, 161). We also have data from an “unexpected event during survey design” evaluation that shows similar patterns, with results available in online appendix 2D. We believe that these results suggest even more robust party effects, given the widespread attention to the Kavanaugh hearings; in essence, our control group was already treated with information about how the parties acted on sexual misconduct. For example, 60% of our respondents had heard about the Kavanaugh hearings prior to our experimental treatment. As we show in table 2, these effects are more pronounced if we restrict the sample just to those who had not heard of the Kavanaugh hearings. For example, looking at the policy question, the Democratic party goes from a 5% advantage in the control group to a 19% advantage in the party treatment group.
Party ownership evaluations among low-information respondents (Survey Experiment)
Note: Post-hoc predicted probabilities from multinomial logit models, restricting the sample to those who had not heard of the Kavanaugh hearings.
The issue of sexual misconduct thus meets the criteria for issue ownership, distinct reactions from elites in each party, and a corresponding shift in views of the parties on the issue. Importantly for our theoretical framework, those differences manifested abruptly in just the past few years.
Partisan Clarity on Recognizing the Problem
To spur issue-based mass partisan change, “the issue must matter” (Carmines and Stimson Reference Carmines1990, 166). Here, we test for shifts in sexual misconduct problem importance in the public by partisanship. We expect that perceptions of sexual misconduct as a problem were largely partisan neutral prior to 2016. Starting in 2016, however, we expect to see significant differences between Democrats and Republicans as they receive diverging cues from party elites about the problem.
Ideally, we would test these expectations on a time-series of questions about sexual misconduct. Unfortunately, no such data exist as surveys have not routinely asked questions about the issue across time. As an alternative test, we turn to an analysis of aggregate survey questions about sexual harassment. We cast a broad net to find all survey questions with available data on attitudes towards sexual harassment since the 1980s from the Roper survey catalog, supplemented with data from Pew and ICPSR. We identified forty-four survey questions on sixteen surveys fielded sporadically from 1986 to 2018 that asked generic questions about sexual harassment like “How much of a problem is sexual harassment?” (refer to online appendix A4 for all questions). For each question, we coded the responses so that more positive responses were friendlier towards victims of sexual harassment with each response rescaled between 0 and 1. We also code partisanship (0 = Republicans, 1 = Democrats, leaners coded with partisans and Independents excluded) and gender (0 = male, 1 = female) in identical ways across the surveys.
We estimated a multi-level model with party, gender, and time, as well as party interacted with time with clustered errors at the survey level (Barabas et al. Reference Barabas, Jerit, Pollock and Rainey2014). We present the post-hoc predicted difference of the interaction between party and time in figure 3, which provides the difference between Democrats and Republicans in all the surveys for each year for which we have survey data: 1986, 1991, 1997, 1998, 2013, 2015, 2016, 2017, and 2018.
Polling data over time
Note: The figure presents the post-hoc predicted difference between Democrats and Republicans from a model predicting views of sexual harassment with controls for party, gender, year, and party interacted with year, with clustered errors at the survey level. See online appendix A3 for full list of surveys and questions, full results of the model that produced results for the graph, and gender effects.
We find considerable evidence supporting our expectation: prior to 2016, Democrats and Republicans did not differ in their views of the importance, existence, or frequency of sexual misconduct. Even in periods where the issue was made salient and partisan by a well-known offender in a specific party (i.e., Clarence Thomas in 1991, Bill Clinton in 1997–1998, Herman Cain in 2011), we do not see mass partisan differences in views. Yet starting in 2016 and continuing through 2018, Democrats expressed significantly more sympathetic views of sexual harassment victims, occurrence, and whether sexual misconduct is a problem. If we look at the patterns for Democrats and Republicans separately (refer to the online appendix), we find that the 2016-onward divergence is the result of both Democrats expressing more positive views and Republicans expressing more negative views.
As a comparison, we also examine gender differences, both overall and within the parties (Cossette and Craig Reference Cossette and Craig2019; refer to online appendix 3A). While women began to see sexual misconduct as a larger problem than did men starting in the 1990s, the MeToo era did not increase this gap; if anything, the gap shrinks in size. In comparison, looking at partisan differences within the genders (so, comparing Republican men to Democratic men and the same for women), we find consistent patterns of growing party differences starting in 2016 (refer to online appendix 3A). We also see similar patterns of partisan differences in views of #MeToo specifically (refer to online appendix 3A).
#MeToo and Affect toward the Parties
Do shifting views of issue ownership and issue importance then translate into shifting affect towards the parties? We do not have feeling thermometers to construct the standard measure of affective polarization in these data, but we do have party favorability ratings, which function similarly. In fact, the wording of feeling thermometer questions explicitly equates warmth with favorability, and so favorability serves as a reasonable substitute. #MeToo favorability is the key independent variable, and we control for partisanship, demographics, racial resentment, and hostile sexism to rule out alternative explanations for differences in partisan affective polarization.
Figure 4 shows that #MeToo views are strongly associated with affective partisan polarization. Positive views toward the movement are associated with significantly warmer views toward Democrats, by a large margin (a Democratic advantage of 0.6), while negative views are associated with more positive views of Republicans (a Republican advantage of 0.45), even when controlling for party identity and other correlates. The relationship with #MeToo favorability is substantially greater than a parallel test with labor union favorability, another issue owned by one party (refer to online appendix A4). These correlational results do not enable tests of directionality or causality, but they do show that affect towards the parties is connected to views of #MeToo, even when controlling for demographics and partisanship.
Party affect by #MeToo views
Note: Post-hoc predicted probabilities from ordinary least squares regression models predicting affect towards Democratic Party and Republican Party on a 1-4 scale with controls for party ID, gender, race, education, age, whether someone has children and is married, their employment status, income, sexism, racial resentment, and views of the country headed in the right or wrong direction. Full results are available in online appendix A4. Error bars are 95% confidence intervals.
Political Consequences of Sexual Misconduct Views
Voting is arguably the most fundamental mass political act in democracies, and partisan identification is the anchor that usually undergirds longer-term behavioral patterns—except when undermined by other group attitudes and identities. Here, we test the impact of sexual misconduct views on partisanship in three ways: expressed importance of candidate positions on sexual misconduct for vote choice, partisan voting loyalty in the 2018 U.S. House elections, and longer-term changes in party support.
Prioritizing Candidate Sexual Misconduct Positions in Vote Choice
We first look at partisan differences in prioritizing a candidate’s views on sexual misconduct in vote choice, using data from a June 2018 survey from the Kaiser Family Foundation. Respondents were asked: “As you may or may not have heard, the MeToo Movement is an international movement that is working to raise awareness about sexual harassment and assault. Are you more likely to vote for a candidate who is an outspoken supporter of the MeToo Movement, does not address issues raised by the MeToo Movement, or does a candidate’s position on this issue not make a difference in who you vote for?” We estimate support for the “outspoken supporter” response with demographic and partisan controls and present the post-hoc partisan differences in figure 5.
Vote preference by candidate behavior (KFF data)
Note: Data from Kaiser Family Foundation Health Survey, July 2015. N = 1000. Post-hoc predicted probabilities from logistic regression models predicting response that they would vote for the candidate who is an outspoken supporter of the MeToo Movement, with controls for gender, race, education, whether someone is married, their employment status, and income and survey weights. Full results are available in online appendix A5. Error bars are 95% confidence intervals.
In June 2018, after the 2016 misconduct attention but well before any information about accusations against Kavanaugh emerged, Democrats were 57 percentage points more likely than Republicans to say they would support a candidate who is an outspoken supporter of the MeToo movement, even with demographic and political controls.
Partisan Voting Loyalty and #MeToo Party Leadership Views
Self-reported voting rationales well before an election might diverge from actual voting behavior near election time. To test whether party issue ownership views shape actual partisan voting loyalty, we return to our module questions in the post-election wave of the 2018 CCES’s two-wave election panel, where respondents indicated which party was better at addressing the issue of sexual misconduct (issue ownership) and holding their leaders accountable (punish leaders).
Not surprisingly, partisans generally reported their own party was better (approximately two-thirds), or at least equal to the other (approximately one-quarter), with only 4% choosing the opposing party on each item. Republicans were a little less favorable toward their own party than Democrats on policy and leader accountability. We combine these party assessments with self-reported party loyalty, disloyalty, or abstaining from voting in 2018. Table 3 shows levels of party voting loyalty for the U.S. House and party performance.
Partisan voting loyalty in House elections by #MeToo party attitudes (CCES data)
We find large loyalty differences between people who rate the other party as better and those who see their own party as equal or better. While the 80% of partisans who view their own party as better on policy vote loyally for their party (3% vote against, 17% no vote), that rate drops to just 44% among those who see the other party as better (19% vote against, 37% no vote). Loyal party voting drops even further among those who view the other party as better at punishing leaders, with respondents reporting voting more for the other party than their own party by ten points (37% versus 27%).
Next, we test the robustness of these patterns with multivariate tests that address potential confounds. We dichotomize the measure to indicate a party-loyal perception (1) or disloyal perception (0). Voting loyalty is coded in order of most to least, based on harm to the party: +1 is a vote for one’s own party, 0 is non-voting, and -1 is voting for the other major party; we choose to measure voting loyalty in this fashion based on scholarship that shows that individuals dissatisfied with their party may first stay home or abstain from voting in a particular race, rather than voting for the other party (for example, see Banda and Cassese Reference Banda and Cassese2021).Footnote 4 Table 4 presents the results of an ordered probit model.
#MeToo views and party loyalty in voting for Congress (CCES data)
Note: Data from Congressional Cooperative Elections Survey module. Ordered probit model (+1 loyal, 0 non-voting, −1 disloyal) with survey weights. Partisans only, including leaners. ^ p < .10, * p < .05, ** p < .01, *** p < .001.
We find some evidence that the relationship between #MeToo party views and party voting loyalty for the U.S. House stays robust even after controlling for partisan strength and the respondent’s match with their party on a range of policy views. We find a clear relationship between #MeToo evaluations for how the party handles its own accused leaders, but a weaker relationship between the evaluation of #MeToo policy evaluations. In sum, Americans told us that #MeToo views mattered for their vote choice, and less obtrusive tests supply support for the claim. When they viewed the other party as better on the issue, they were more likely to abstain or event defect, even after accounting for other factors affecting their party loyalty.
Party Switching
We have presented a wide set of evidence on changing views of sexual misconduct and its effects on partisan evaluations and voting. Whether it ultimately matters for partisan identification is the last piece in our sequential model of rapid realignment. This poses a significantly harder test for partisan change given the general stability and endurance of party identities and one where we are unsuited methodologically to make causal claims. We can, however, use observational data to identify trends in data. The 2018 DFP survey asked whether partisans had “always” been a supporter of their party or if they were previously a supporter of one party “but have switched to supporting” the other party; unfortunately, respondents did not provide any information about when they had previously been a supporter of another party beyond the distinction of not “always,” which would potentially provide more evidence implicating #MeToo or suggest alternatives. This is not the canonical party identity question but it does get at party support over the long-haul. This is a single-shot survey rather than panel data to observe party change, but it is self-reported party change that adds a temporal component. In the survey sample, 10% of Democrats and 16% of Republicans reported having supported the opposing party at some point in the past.
We compare the change in party support to views of the #MeToo movement; post-hoc predicted effects are presented in figure 6. There, we find substantial volatility in partisan support oriented around #MeToo views, even when controlling for demographic factors, political attitudes, and partisan strength. Views of the #MeToo movement are a strong predictor of change from Democratic to Republican Party support, somewhat more so than change in the other direction. Of particular interest are the opposing poles, whereby both Democrats and Republicans with very favorable attitudes towards #MeToo were roughly equally likely to report switching parties. However, among those with very unfavorable attitudes, partisan change into the Republican Party was four times more likely than a change into the Democratic Party by post-estimation differences in probabilities. In essence, the unfavorable side of the measure is doing the heavy lifting: hostility to #MeToo was a more potent motivation for a change in party ID than favorable views toward the movement. The asymmetry in partisan change is intriguing. We do not have a definitive explanation for the difference, but one plausible cause is that Republicans were nearly twice as likely to report being ex-Democrats (39%) as Democrats who were ex-Republicans (21%). In other words, the baseline rate for switching in each direction was much different, which would also affect the rates of switching conditional on #MeToo evaluations.
Party switching and MeToo favorability
Note: Post-estimation predicted probabilities; OLS models with controls for race, education, age, children in the home, married, employed, income, racial resentment, and views of direction of country with survey weights. Full models are in online appendix A6. Error bars are 95% confidence intervals.
For a final time, we see substantial evidence of a dynamic relationship between partisanship and views about sexual misconduct, this time for the most stable trait in mass political behavior, suggesting the disruptive power of misconduct views and group orientations more broadly, especially among those with hostile attitudes toward #MeToo. Consistent with our expectations, changes in durable party voting patterns and identifications were smaller than the more malleable issue-related views of party reputations and viewpoints among partisans. Nonetheless, these dynamics show the potential for rapid shifts in partisan attitudes and the potential for longer-term, larger-scale party change around the issues of #MeToo politics.
Discussion and Conclusions
How might an issue drive a rapid change in party reputations, with corresponding effects on party affect, voting loyalty, and party identification? In this article, we argue that issue preferences drive partisan change among people who prioritize the issue when the political parties diverge in their responses to newly salient issues, while others adopt their party’s newly salient position. We adapted the issue evolution model (Carmines and Stimson Reference Carmines1990) and the conflict extension model (Layman and Carsey Reference Layman and Carsey2002a, Reference Layman2002b) by adding a more explicit role for partisan issue ownership. We then presented a trove of data focused on the case of sexual misconduct, including dozens of surveys, a natural experiment, and a survey experiment. In doing so, we identified changes in party reputations, in the views of party followers, in partisan affect, and consequent effects on party voting loyalty and broader party-switching.
We first found an abrupt breakpoint in partisan views of sexual misconduct in survey data across several decades, with few clear party distinctions until the intersection of #MeToo politics with the 2016 presidential election and later events. Those circumstances dramatically shifted public perceptions of partisan issue ownership on the issue, and both our natural experiment and survey experiment show the added impact of exposure to elite party rhetoric surrounding the Kavanaugh Supreme Court confirmation hearing. #MeToo views then conditioned affect toward the parties, above and beyond other factors. Finally, our surveys identified the self-described importance of #MeToo views in vote choice, real shifts in party voting loyalty among those few who favored the opposing party’s stance, and substantial shifts in broad party support over time.
Notably, we found that political salience alone is not enough to shift partisan preferences. Indeed, our survey data showed that several high-profile instances of sexual misconduct by politicians failed to shift party reputations on the issue. Likewise, our experimental tests showed little movement in party reputations when hearing about misconduct alone. Instead, learning about (or being reminded of) how party elites responded differently to sexual misconduct is key to changing party reputations. In this way, our research on #MeToo provides further evidence of slow-moving patterns of partisan realignment (Carmines and Stimson Reference Carmines1990), driven both by the shifting patterns of the membership in each party and the actions of party members (Ondercin Reference Ondercin2017). Our research also confirms strengthening partisan identities (Mason Reference Mason2013, Reference Mason2018).
Our research also suggests that this is, in part, due to the conduct and messaging of elites within each party. What, then, could we expect if the Democratic Party confronts a Kavanaugh-like figure within their midst in the near future? One possibility is that the Democratic Party would eject such a member (or, in the case of a Supreme Court nominee, not nominate such a member), as they did after allegations of abuse by Senator Al Franken. Another possibility is that the party could leverage its solidified issue ownership reputation to escape critique, just as the Republican Party continued to own national security even after the security failures that led to 9/11. Future research on the durability of issue ownership, particularly as it accompanies a party-issue change, could investigate these possibilities.
The research presented here has a variety of implications for understanding social identities and movements, the role of gender in political attitudes, and the degree to which voters can hold leaders accountable in a two-party system. Researchers point to both the successes and failures of social movements that tap into social identities (Silber Mohamed Reference Heather2017; Banaszak and Ondercin Reference Banaszak and Ondercin2016; Kaufman Reference Kaufman2000); the full extent of how sexual misconduct might change partisan politics remains to be seen. The 2016, 2018, and 2020 elections were dominated by issues of identity, with group-based identities like gender, race, immigration status, and sexual orientation all key aspects of candidate self-presentation, campaign attacks, and voter decision-making. Of particular importance in 2016 and 2020 was the role of hostile sexism (Cassese and Barnes Reference Cassese and Barnes2018; Cassese and Holman Reference Cassese and Holman2019; Archer and Kam Reference Archer and Kam2021); future research might evaluate how these attitudes map on to both views of #MeToo and partisan identities, shaping candidate choices.
Whether these shifts in party reputations, vote choice, and identification continue to build remains to be seen, but partisan change on sexual misconduct has clearly already begun. Our results are consistent with other work that finds liberals and conservatives today holding different views of whether particular acts constitute sexual harassment, even when the harassment is blatant (Warren, Gothreau, and Schneider Reference Warren and Gothreau2020; see also Bankert Reference Bankert2020b for a discussion of discrimination). Future research might evaluate the degree to which sexual misconduct has motivated individuals to run for political office, both to promote policies to protect victims of sexual misconduct and to block such efforts. For example, a large group of women (largely Democrats) were spurred to run for office in 2018 in response to negative emotions in response to #MeToo and the 2016 presidential election (Dittmar Reference Dittmar2019).
As race-related issue evolution eventually increased the share of candidates of color in the Democratic Party, so might the sexual misconduct evolution reinforce the Democratic Party’s reputation as more favorable to gendered policies and women candidates. Conflict extension involving opinion leadership beyond party switching might encourage the same, though Democrats adopting their party’s #MeToo view would presumably provide less impetus on the issue than those who join the party for that reason. In that way, they may mirror recent liberalizing shifts in racial attitudes among white Democrats in response to the Black Lives Matter movement.
Either way, if present trends continue, one party may come to be seen as the party that defends perpetrators and their enablers while the other advocates for survivors of sexual harassment and violence. That will continue to realign partisans out of step with their party at the ballot box and in their party identifications. Against those trends, Democrats may struggle to uphold their consistent reputation when political expedience encourages some partisans to support leaders credibly accused of misconduct. To the extent that women are more sympathetic to #MeToo appeals than men (for example, women across the parties have more favorable views of #MeToo than do men in the 2018 DFP data), those differences may tend to reinforce and even expand the ten-point partisan gender gap in voting that routinely appears in recent presidential elections.
Supplementary Materials
Appendix A1. Party Ownership Question Wording (table 1)
Appendix A2. Survey Experiment (figure 2)
Appendix A3. Survey Data Compilation (figure 3)
Appendix A4. MeToo and Party Affect (figure 4)
Appendix A5. Candidate MeToo Stance and Vote Intention (figure 5)
To view supplementary material for this article, please visit http://doi.org/10.1017/S1537592721001912.
Acknowledgment
This paper was previously presented at the Louisiana State University’s Research Roundtable at the Manship School; the University of California, Riverside; the 2019 Midwest Political Science Association; and the 2020 Southern Political Science Association meetings. Research in this paper was funded by the Duran Fund and the Newcomb Institute at Tulane University and the Gittinger Professorship at Louisiana State University’s Manship School of Mass Communication. Thanks to Monica Schneider, Jennifer Merolla, Brian Schaffner, and Nichole Bauer for their comments on the paper and to Natalie Sharpe for her stellar research assistance.
Open access
