Abstract
Recent research has identified neither consistent gendered stereotypes about legislators among constituents, nor a link between legislators’ gender and their electoral success, nor consistent gender coalitions in legislative voting and co-sponsorship. However, it would be premature to declare that gender is irrelevant to legislative politics. This article considers state legislators’ choice to discuss or omit family in their social media profiles and interprets the choice as a gendered presentation of self. The social media platform Twitter is notable for its status as an especially public stage and for the narrow choice that limited profile space imposes. Different strains in gender theory lead to competing predictions: that women will avoid references to family to avoid making gender salient, that women will highlight references to family to avoid backlash for non-normativity, or that differing gender expectations in different political parties lead to different gender effects. Apart from gender, family references may be more or less available to legislators as a function of their own family status and life stage. Analysis of Twitter profiles of state legislators in 10 US states (N = 911) reveals that party and gender effects interact; controlling for family and life stage circumstances, non-Republican women are moderately more likely to mention family than non-Republican men, but Republican women are much less likely to mention family than Republican men. Gender matters in the social media presentation of legislators’ selves, but in a manner strongly conditioned by party expectations.
Introduction
Is gender a force in American legislatures of the 21st century? The continuing numerical dominance of men in national and state legislatures is an undisputed fact. For every “success” story like the Vermont General Assembly (at 40.6% women in 2015, still a minority), there is a countervailing example like the Louisiana legislature (11.8% women in 2015). However, in the domains of elections, bill sponsorship, and roll call voting, gender distinctions are unclear at best. Although some evidence suggests that voter stereotypes regarding men and women legislators persist (Bauer, 2015; Bell & Kaufmann, 2015; Bligh, Schlehofer, Casad, & Gaffney, 2014; Dolan, 2014b; Fridkin & Kenney, 2009; Kittilson & Fridkin, 2008; Sanbonmatsu & Dolan, 2009; Stalsburg, 2010), it also appears that in comparison with dominant party biases, gender only weakly determines perceptions of candidates (Brooks, 2013; Dolan, 2014a; Hayes, 2011; Hayes & Lawless, 2013; Hayes, Lawless, & Baitinger, 2014; Sapiro, Walsh, Strach, & Hennings, 2011). Furthermore, voter stereotypes and electoral outcomes appear to be disconnected; despite the persistence of gender stereotypes about legislative candidates, women candidates manage to be elected to office as often as men candidates (Dolan, 2014b; Fox & Oxley, 2003; Fulton, 2012). Some research suggests that upon gaining office, men and women cluster by gender in patterns of bill co-sponsorship and voting (Bratton, 2005; Cook, 2011; Hogan, 2008; Swers, 2005). However, other research finds no significant difference between the bills supported by men and women when controlling for other factors such as party and ideology (Bratton & Rouse, 2011; Cook, 2015; Jenkins, 2012; Reingold, 2000). If there is an effect of gender on support for bills, it may be indirect, mediated by those factors (Clark & Caro, 2013).
Despite inconclusive evidence of divergent instrumental outcomes for men and women in American legislatures, it is far too soon to conclude that the legislature is not a gendered environment. After all, gender is not merely a dimension imposing discrimination or a moral compass directing instrumental action. Gender is also an ongoing act through which legislators continually establish, reinforce, and alter their public selves by successfully performing roles for audiences (Goffman, 1959; West & Zimmerman, 1987). When legislators enact a performance of self, they must select a role that is likely to lead to success, that fits on the stage provided, and that works within the limitations of a gender status to which they have already been ascribed. These roles are not simply limited to dichotomous aggregate master statuses (“men” and “women”) but take on multiple contingent forms (“mother,” “husband,” and “daughter”) that connote expectations dependent on the political context within which they are invoked (Burns, 2007; Rai, 2015).
This article considers the gendered presentation of self on the relatively new social media stage available to legislators. In particular, I ask whether state legislators use the text and image fields in Twitter profiles to present information about their family roles to the public. The use of family in presentation is not new or limited to social media; Sclafani (2015) recently identified family references as a rhetorical tool for presidential candidates to legitimize their political role aspirations. However, social media affords a public stage of a unique kind, one on which persistent identities can be created and broadly diffused to multiple audiences through performance with the discursive “props” of posts and profiles (Marwick & boyd, 2011; Papacharissi, 2012, 2014). The social media stage is always lit, even when the performer is asleep. Twitter is particularly interesting as a social media form because of its default-public status and because the limited space for Twitter profiles forces legislators to decide what information about themselves is essential.
Applied to social media profiles of legislators, the literatures on status characteristics, workplace perception, and expectancy violation generate conflicting hypotheses regarding the choice to include or omit references to family. I test these hypotheses by gathering information on Twitter profiles created by legislators from 10 states. When controlling for legislators’ family status and life stage, gender differences in self-presentation over social media emerge that are contingent on political party membership.
Gender and Discussion of Family in the Legislature
While there is no prior research regarding the gendered social media profiles of legislators, significant bodies of research have been published regarding the gendered choices of men and women in the workplace in general and by legislators using more traditional media in particular. This existing literature does not point in a single direction, however, but rather draws from different theoretical bases that suggest hypotheses at odds with one another.
In one of these traditions, status characteristics theory identifies gender as a form of identity that can be more or less salient in a social setting. To the extent that gendered expectations are not visible in a setting such as a legislature, expectations of performance may not be organized according to gender. However, if and when the gender of individuals participating in that setting is made salient, then expectations associated with gender for those individuals are also made salient (Ridgeway, 2011). These expectations include the presumption of authority and dominance for men, which can be advantageous for men and disadvantageous for women in a competitive legislative environment (Ridgeway & Correll, 2004). To the extent that gender “ethic of care” stereotypes regarding nurturing motherhood and marriage are invoked for women legislators, they may further undermine an “ethic of justice” image of authoritative, committed competence in the workplace of the legislature (Deason, Greenleeb, & Langnerc, 2015). If gender stereotypes do not damage women legislators as candidates, the lack of harm may be due to women legislators’ active attempts to counter stereotypes by acting in a manner contrary to gender expectations (Hayes, 2011).
In a survey and field experiment on this point, Correll, Benard, and Paik (2007) find that gender and family status can interact to affect perceptions of men and women in the workplace differently. The indication of parenthood status by equally qualified pairs of men and women led to different evaluations by survey respondents and actual employers. Men applying for a job were judged to be more competent, more committed, and more likely to rise to leadership if they indicated they were parents as fathers. Women, on the other hand, were judged to be less competent, less committed, and less likely to rise to leadership when they indicated they were parents as mothers. This pattern has since been replicated in the employment context (Hoyt, 2012) and can be applied to understand the legislative setting, where men legislators feature images of their families in video advertising and websites more often than women legislators (Bystrom & Kaid, 2002; Bystrom, Banwart, Kaid, & Robertson, 2004). This decision has been tied to advantages for men and disadvantages for women in making gender identity salient (Bystrom, 2013).
If legislators are strategic in their presentation of self, then given the likelihood of different reactions, men and women should make different choices about referring to family. The tradition grounded in status characteristics theory leads to a first hypothesis:
Status characteristics hypothesis. Women legislators will refer to family less often in their profiles than men legislators.
However, another tradition in research on politics and gender leads to a contrary prediction. Kathleen Dolan (2014b) suggests a recursive cycle in which women legislators may choose to craft images of themselves that match the perceived gender stereotypes of the electorate in order to better appeal to the electorate, in the process reinforcing the perception that such gender stereotypes are dominant (p. 153). Work drawing from expectancy violation theory predicts that women legislators who fail to highlight their marriage or motherhood status (or who fail to be married or mothers) face punishment for non-normativity (Bell & Kaufmann, 2015; Deason et al., 2015). In a “backlash effect” similar to effects of status characteristics theory but with an effect in an opposite direction, the failure to properly enact gender norms of family focus may make them more salient (Okimoto & Brescoll, 2010; Rudman & Fairchild, 2004; Rudman, Moss-Racusin, Glick, & Phelan, 2012). To succeed as legislators, according to this argument, women must already violate gender norms by expressing assertiveness and dominance. To avoid negative sanctions, women must balance this gender norm violation with a countervailing embrace of some other gender norm. Highlighting family is one way to accomplish this balance (Heilman & Okimoto, 2007; Rudman et al., 2012). Research supporting this perspective includes experimental evidence in which references to motherhood ameliorate negative perceptions of women in leadership (Heilman & Okimoto, 2007), while references to non-marriage and non-motherhood lead to poorer evaluations of women candidates for legislative office (Bell & Kaufmann, 2015; Stalsburg, 2010). If we assume that women legislators are aware of and respond to such trends, this article’s second and competing hypothesis follows:
Expectancy violation hypothesis. Women legislators will refer to family more often in their profiles than men legislators.
The work of Evans, Cordova, and Sipole (2014) connects tangentially to this empirical prediction with reference to social media, finding that women legislators are more likely to make “personal” posts on Twitter than men legislators. This pattern of behavior is consistent with responses to the pressures described by Dolan and colleagues. In contrast, however, Banwart and Winfrey (2013) find that men and women in legislative politics publicly discuss their personal lives to a roughly equal extent. Personal posts in both pieces of research include, but unfortunately are not limited to, familial references, leaving the question of familial reference in social media unsettled.
Complicating the straightforward predictions made above, differing gender norms held by constituents of different parties may make those constituents liable to respond differently to legislators’ mentions of family. Gender expectations are contingent on social position, not monolithic and unchanging. Conservative Republicans may reward women politicians who make reference to family to the extent that conservatives more strongly expect women to define themselves according to family (Bell & Kaufmann, 2015; Deason et al., 2015; Dolan, 2014b; McCabe, 2005; Sanbonmatsu & Dolan, 2009). Liberal Democratic women, by contrast, may feel less pressure to refer to family to the extent that their supporters are less concerned with adherence to traditional gender norms about women’s involvement with family. This prediction of an interaction effect is grounded in the expectancy violation literature:
Party interaction with expectancy violation hypothesis. Republican women will be more likely to refer to family in their profiles than non-Republican women.
On the other hand, if references to family make gender more salient and thereby raise attention to broader gender norms of masculine dominance and feminine subordination, pressures should be felt most strongly by Republican women legislators whose constituencies more strongly embrace those broad gender norms (Hoyt, 2012; Stalsburg, 2010). Liberal Democratic women, by contrast, may be less vulnerable to punishment to the extent that broad norms of masculine dominance and feminine subordination are less salient among liberal constituencies. According to status characteristics theory, Republican men in a legislature have the most to gain from a strategy of making family more salient in communication, since their more conservative base more strongly embraces the gender norms that connect masculinity to dominance. Non-Republican men have comparatively less to gain from making family salient according to this line of thought if their more liberal base less strongly embraces gender norms of male dominance. These predictions are expressed as interaction effects for the status characteristics hypothesis:
Party interaction with status characteristics hypothesis. Republican women will be less likely to refer to family in their profiles than non-Republican women. Republican men will be more likely to refer to family in their profiles than non-Republican men.
Of course, references to family in a legislator’s social media profile may be due to factors other than gender. Legislators who choose to feature family in their social media profiles might simply have a more active family life than other legislators. Legislators who are married or legislators with children could reasonably be expected to mention family more often, especially legislators with younger children for whom child-rearing activities are more salient and time-consuming (Dukhovnov & Zagheni, 2015). Family portrayals might also be a marker of partisan identity, not gender identity. To the extent that the Republican party has worked for a generation to position itself as a defender of “family” values, legislators in such that may find it advantageous to highlight family relations regardless of their gender identity or family status (Elder & Greene, 2012).
Finally, Twitter literacy and activity matter in a very practical sense. Legislators may be more likely to refer to family in their social media profile simply because they have devoted more attention or resources to their account. Some legislators may spend a great amount of time using their account, posting thousands of Tweets; alternatively, they may have dedicated staff and volunteers to manage their social media presence. These legislators are more likely to have spent the time to complete and tailor their profile and include strategic references. Other legislators may know relatively little about Twitter as a social media platform and may have given less consideration to the content of their profiles. Some legislators may have never tweeted or have wholly neglected to add a text profile at all; these individuals are certainly less likely to have mentioned family in them. To account for these possibilities, level of Twitter activity and presence of profile information must be taken into account during analysis.
A Focus on Twitter and State Legislatures
Why focus on Twitter? In the two decades during which websites have been used in political campaigns, they have been repeatedly scrutinized for differences by gender (Banwart & Winfrey, 2013; Bystrom, 2013; Bystrom et al., 2004; Dolan, 2005). On the other hand, Twitter has only existed for a decade and has only recently been adopted by a majority of national legislators in the United States (Evans et al., 2014; Golbeck, Grimes, & Rogers, 2010), providing little opportunity for the consideration of Twitter as a gendered platform in politics. Regardless, the relevance of Twitter to legislative politics has grown quickly as Twitter adoption has diffused down to the state level. As of 2014, a majority of legislators have adopted Twitter in 7 of the 10 states described in Table 1, a super-majority in 5 of the 10 states, and a near-majority of legislators in the remaining 3 states.
Characteristics of 10 State Legislatures.
Source: Ziegler (2014) and National Conference of State Legislatures (NCSL, 2009, 2013, 2014).
Quintile ranking among 50 states in parentheses (1 = top, 5 = bottom).
One feature that makes Twitter a particularly interesting social media platform for study in the legislative context is that it comes closest to re-creating a “public sphere” in which all members may fashion a civil society based on horizontal discourse (Habermas, 1989; Papacharissi, 2014: 145). While Standage (2014) hails social media as a return to the idealized public sphere of old after an era of mass-media dominance, Facebook, Instagram, Pinterest, and blogs remain closer to a mass-media form with the account-holder dominating the start of conversation and maintaining moderation power over visitor comments. Twitter, by contrast, comes closest to the Agora or classical coffee house in which all visitors (or at least, all those with the coin or status required to visit) are able to partake in conversation on an equal footing. Posts, mentions, replies, and retweets on Twitter are all subject to the same size and format restrictions. Any account can speak to any other, and although Twitter users may mute or block other users from their personal timelines on a one-to-one basis, no Twitter account has the ability to wholly delete the posts of other users or even to delete replies they have received. As in the Agora and the coffee house, some voices on Twitter gain markedly greater visibility over time through more and more influential followers (Dormagen, 2014), but that possibility remains open to all. No one is forced to listen, but anyone can take to the stage.
Compared to other social media platforms, then, the Twitter environment most closely approaches the dramaturgical stage envisioned by Erving Goffman (1959) on which actors strive to maintain a legitimate performance through the playing of parts and the sharing of cues in interaction (Papacharissi, 2014). This performance begins as each player adopts a persona, the modern equivalent of the classic Greco-Roman dramatic mask indicating some central, telling essence of character (Gibson, 1969). Before ascending to the stage, Twitter asks each account-holder to put on a face in the form of a profile picture and a brief snippet of profile text. It is this face that actors seek to fulfill, or at least to save, in their social media interaction. As Marwick and boyd (2011) point out, the more interactive a social media platform is, and the less exclusive control any one participant holds over others, the greater the attention participants must pay to role performance through identity management. Public scrutiny further prods Twitter users’ attention to the perceived expectations of others in the environment. The importance of appropriate presentation of self over Twitter is highlighted by the Sun Foundation’s project to document missteps in politicians’ Twitter performance by saving deleted Tweets on its Politwoops website (http://politwoops.sunlightfoundation.com/) and by the development of a second career for Pennsylvania Representative Mike Schlossberg (2014) as a social media consultant who is paid to help other politicians avoid “social media fails” on Twitter.
The Twitter profile is also useful as an object of study because it forces users to make difficult choices. Of course, profile graphics and profile texts are not all there is to Twitter; the content of Twitter posts (“Tweets”) is another area in which a user may make communications about her or his family. It is quite possible that a user might discuss family in the text of a Twitter post but never in a profile. But this possibility is telling. Tweets, unlike profile expressions, are ephemeral, quickly moving down the timeline and out of followers’ attention. Tweets are also, while limited to 140 characters per post, unlimited in number. As many politicians have found to their delight, it is possible to post as many Tweets as a person wishes. The relative strength of a choice made about the text of a Tweet is, therefore, diluted compared to the strength of the message conveyed in a profile. While the volume of self-identification on a candidate’s website is limited only by imagination and bandwidth, Twitter limits profile descriptions to 160 characters and profile pictures to 128 square pixels. With only so much space for an enduring expression of self, a legislator who uses Twitter must decide what elements of self are especially important to present in a profile (Papacharissi, 2012). It is reasonable to conclude that Twitter profiles in text and words offer an especially strong glimpse of a legislator’s most central, essential presentation of self.
Why focus on state legislatures? To the extent that Twitter is studied as an avenue of self-presentation in legislatures, state legislatures may be the most appropriate bodies to observe. In the wake of Rep. Anthony Weiner’s poorly managed acts of gender self-expression, the management of social media self-presentation has been outsourced at the national level to congressional staff who carefully manage social media accounts (Williams & Gulati, 2015). State legislators, who are less supported or entirely unsupported by staff and whose social media profiles are therefore likely to be self-produced, offer a less-polished glimpse at social media presentation of self. In this environment, legislators tend to make choices themselves about how to construct their profiles, making any observed patterns less liable to be an artifact of organizational policy and more a result of social forces. State legislatures are also typical points of entry into representative politics in the United States, representing just one step beyond initial recruitment, where men and women from the population are most strongly sorted by gender (Lawless & Fox, 2010). It is important to know whether gendered differences persist or are magnified at this early stage of political careers.
Data and Methods
In order to assess the relationship between gender and presentation of family in social media, I observed the Twitter profiles of state legislators from 10 of the 50 United States in June 2014. I chose the states listed in Table 1 to reflect a diversity of size (from 63 members to 253 members), a diversity of region (two Southern, three Mountain, one Central, one Pacific, one Midwestern, one Mid-Atlantic, one New England), a diversity of professionalization (two citizen legislatures, two professionalized legislatures and six hybrids—National Conference of State Legislatures [NCSL], 2013), a diversity in the extent of Twitter adoption (from 48.1% to 88.7% of legislators), a diversity in partisanship (five legislatures with a minority of Republicans and five with a Republican majority—NCSL, 2014), and a diversity in the representation of women (five below the median of 24.5% women and five above that median—Ziegler, 2014). The effort to simultaneously select for diversity along each of these dimensions from only 50 states means that perfect representativeness is not possible, but quintile rankings in Table 1 indicate a fairly diverse set along these dimensions when compared to the range of all 50 states. For each of the 10 selected states, I curated a list of all Twitter accounts discoverable through Twitter search, Google search, and snowball sampling of known accounts’ lists of followers.
The result of this process was a set of 911 legislators with Twitter accounts, 64.7% of the 1,409 state legislators serving in the 10 states as of June 2014. As Table 2 documents, Twitter adopters are different from one another from state to state, although some commonalities emerge. Majorities of Twitter adopters in every state have completed a text profile, majorities have married, and majorities have children. In each state, the average Twitter user has made many hundreds of Twitter posts (“Tweets”) over the lifetime of her or his account, and in each state the large standard deviation indicates that there is a wide range in volume of Twitter activity. In all, 51 Twitter users have not made a single post; in contrast, Rep. Garry Smith of South Carolina had made 27,020 Twitter posts as of June 2014 (as of April 19, 2016, his grand total of posted Tweets had reached 38,065).
Characteristics of Twitter Adopters in 10 State Legislatures.
Standard deviations in parentheses.
The most noticeable differences between states emerge in partisan and gender composition, but these largely reflect differences in the partisan and gender composition of the legislatures overall. In t-tests for each state, no statistically significant differences emerged between the percent of Twitter adopters who are women and the percent of all legislators who are women. Statistically significant differences did emerge between percent Republican for Twitter adopters and percent Republican for all legislators, but only in three states. The three statistically significant differences involved two states (Michigan and Nevada) in which Republicans were underrepresented among Twitter adopters and one state (South Carolina) in which Republicans were overrepresented among Twitter adopters.
A Twitter profile image and a text profile (if any existed) were collected for each legislator account in June 2014. Family references are counted if an account profile refers to some combination of the legislator’s children, parents, or spouses; in profile images, the identity of individuals other than legislators was checked against captioned images in Facebook posts, legislators’ websites, and newspaper articles to verify the nature of their relationship to legislators. To control for the extent to which a legislator is an incomplete or thorough user of her or his Twitter account, I obtained a count of the number of Twitter posts made by each legislator account as of June 2014 and also noted instances in which a text profile was left blank.
The more extensive a legislator’s family, the more likely it is that a legislator may refer to family, and so information on marital and parental status of the legislator was collected from social media accounts, official legislative biographies, third-party legislative databases, newsbanks, and broad web searches. These sources were also used to gather information on the age of legislators. For most legislators, the exact year of birth could be gathered, but for a significant minority no specific year of birth could be found. However, in all of these cases, a combination of indirect information (years of graduation, employment, marriage, parenting, and service) could be gathered to determine whether or not a legislator was over the age of 50 years, the age after which hours spent in family care markedly diminish (Dukhovnov & Zagheni, 2015).
Data from all states were aggregated to estimate a single set of logistic regression models predicting the odds that a legislator’s Twitter profile refers to a family member, with the dependent variable equal to 1 if the profile image, the profile text, or both refer to family, and equal to 0 otherwise. Because state legislatures are social units in which reference to family by one legislator may be correlated with reference to family by a fellow legislator due to network effects, organizational effects, or common culture, robust standard errors to account for possible clustering by state are calculated. It is unfortunately not possible to incorporate state-specific variables in multivariate analysis due to very low numbers of Republican women using Twitter in some states. Of the 10 states studied here, 6 counted fewer than 10 Republican women with Twitter accounts; only 3 Republican women held Twitter accounts in the Nevada legislature of 2014.
Results
Table 3 presents the results of logistic regression models predicting reference to family in a legislator’s Twitter profile. Five logistic regression models are run, starting with a simple bivariate model in which legislator gender is the sole independent variable, then introducing party as a second independent variable, then introducing an interaction term between gender and party, then adding control variables to control for family and age status, and finally adding control variables to account for the presence or absence of a text profile and the extent of Twitter activity by the holder of an account.
Logistic Regression Models for Reference to Family in a Legislator’s Twitter Profile.
LR: likelihood ratio.
Robust standard errors in parentheses.
p < .10; * p < .05.
When no other factors are controlled for (Model 1), and when only controlling for party (Model 2 and Model 3), no statistically significant association between gender and reference to family is evident, either in main or interaction terms. However, moderate unobserved heterogeneity leads to biased estimates in these initial models, masking gender effects. Republican legislators are more likely to be married with children than non-Republican legislators (correlation +.25), while women legislators are less likely to be married with children than men legislators (correlation −.15). When legislators’ family and age status are controlled for (Model 4), statistically significant effects emerge for gender as a main effect and in its interaction with political party, and statistically significant effects of legislator marriage and age are also evident. These effects are mainly as predicted; legislators who are married are more likely to refer to family in their social media profiles, while legislators who are of an age associated with less family involvement are less likely to refer to family. In something of a surprise, legislators who have children are not significantly more likely to refer to family in their social media profiles; however, the effect of children in Model 4 and Model 5 nearly attains a significance level of p < .05, and the strong correlation between marriage and children among legislators (+.53) indicates the standard error of the effect of children is likely to be inflated, leading to an underestimation of the significance of children for family references.
These effects persist with the addition of controls for level of Twitter engagement (Model 5). Not surprisingly, accounts with a text profile are more likely to mention family than accounts without such a profile; this simply represents an increase in available space for family to be mentioned. Twitter activity (measured as thousands of Tweets posted) is not significantly associated with reference to family, suggesting that the decision to craft a profile comes prior to the posting of a large number of Tweets and that further inclusion of family information does not come as legislators continue to post more Tweets.
Turning to the four hypotheses suggested by different theoretical strains, the combined effect of legislator gender, legislator party, and their interaction is made clearer when the estimated odds ratios of Model 5 are converted to the predicted probability of a legislator making references to family under various conditions. These probabilities are shown in Figure 1, which reveals that the combined effects of family status and age are considerable. However, within the set of those legislators aged 50 years or less, regardless of family status there is a consistent and substantively important ordering in the probability of referring to family that is reducible to neither gender nor party alone. Republicans are not always more or less likely to refer to family than non-Republicans despite their gender status, and in a blow to the simpler versions of the status characteristics and expectancy violation hypotheses, Women are not always more likely to refer to family than Men despite their party status. Contradicting the predictions of the party interaction with expectancy violation hypothesis, Republican women are not more likely to refer to family than non-Republican women. Rather, non-Republican women are the most likely to refer to family, followed by Republican men. Non-Republican men are still less likely to refer to family, and the least likely group to refer to family are Republican women. The pattern of empirical results is most compatible with the party interaction with status characteristics hypothesis. In this group of states, Republican legislators are drawn most strongly apart in their presentation of self over social media, as conservative gender ideology makes it seem advantageous for Republican men to highlight their family via Twitter but disadvantageous for Republican women to do so.
Predicted probability that a legislator refers to family under varying values of independent variables.
It is important to note that this pattern in the behavior of legislators does not require constituents to actually respond in the manner described by status characteristics theory. It may be that constituents’ gender stereotypes regarding candidates are on the wane (Brooks, 2013; Dolan, 2014a; Hayes, 2011). However, legislators need to only subscribe to the belief that their constituents will respond in such a manner and then act according to that belief when creating a persona on the social media stage. A political culture of beliefs about what gender norms are that do not match empirically observed trends regarding existing gender norms may nevertheless produce gendered behavior that is real in its consequences.
Conclusion
This article seeks to advance the discussion of gendered legislative behavior into the domain of social media, contrasting the predictions of status characteristics theory with those from expectancy violation theory regarding the presentation of family in space-restricted profiles on Twitter. Studying 911 legislators with Twitter accounts in 10 state legislatures, I find no simple main effect of gender upon the likelihood of mentioning family in a profile. Instead, an interaction between gender and political party leads to a circumstance in which male Republicans are the most likely to mention family in social media profiles, while female Republicans are the least likely to do so. These results are consistent with a version of status characteristics theory in which women subject to judgment in a conservative context appear to be at risk for negative judgment when gender is made salient, while men seem to benefit in the same context.
While the results of this research are suggestive, they are not conclusive for a number of reasons. First of all, this research only studies the legislatures of 10 states, not all 50. Future research should return to this topic with attention given to all 50 states. While a dataset of 911 legislators may seem to be sufficiently large, the marginal statistical significance associated with the interaction of party and gender is at least in part due to the relatively low number of legislators in the category seemingly most affected by gendered status expectations. While there are 387 Republican men, 301 non-Republican men, and 151 non-Republican women in these 10 state legislatures, there are only 72 Republican women—too few Republican women to conduct an analysis for each state in which party and gender interact. The statistical challenge in studying gender in a legislative environment stems at least in part from the gendered nature of the environment itself. A follow-up study that includes all 50 states will be both more generalizable and more amenable to detailed study and a more subtle specification of context, at least by region.
Second, this article suffers from the problem of remaining theoretical in its description of the perceptions and motivations of legislators. Data as gathered for this study, limited to categorical profile information, render that task infeasible here. An ideal future study should dive deep, with either detailed interviews or detailed qualitative observation of social media content in at least one legislature to discover, empirically speaking, what legislators’ perceptions of the gender stereotypical context actually are. The significant association between legislator’s marital status and family references, combined with the lack of consistent association between legislator’s parenthood and family references, provides another avenue for future qualitative inquiry.
Third, this article only gathers data regarding current legislators, and fails to obtain data regarding the campaign context for legislators. Are legislators perhaps more attentive to concerns regarding gender normativity when they have been in a close race for election or re-election? Are legislators preparing to retire from office immune to these concerns? Does majority or minority status by party influence the gender normative environment? Do non-incumbent candidates for legislative office in these states follow the same pattern in their reference to family, and if so, is the extent of the pattern stronger or weaker?
Finally, it seems warranted to consider expressions of gender in social media beyond family references in a Twitter profile. While Twitter profiles are a conveniently terse mode of communication, collection of the Tweets posted by legislators over some period would provide another mode of measurement, one more sensitive to the interesting dimension of time. Gender is about more than family. The feminine-normative subject of family and the countervailing masculine-normative subject of work may be reflected in the content of women’s and men’s posts. The graphic nature of profile images also allows for the observation of more subtle gendered expressions of assertiveness or deference in stance, in direction of gaze, and in facial expression.
As the emergence of social media has widened the possibilities for communication by legislators, so it has widened the possibilities for observation of those communications. Findings presented here suggest a number of directions for future research. As that research progresses, an opportunity presents itself to more fully bridge the gap between the actual state of gendered reactions to legislators, academic understanding regarding that state, and the understanding of the situation among legislators. This opportunity should not be squandered.
Footnotes
Declaration of Conflicting Interests
The author(s) declared no potential conflicts of interest with respect to the research, authorship, and/or publication of this article.
Funding
The author(s) received no financial support for the research, authorship, and/or publication of this article.