Expectations, Perceptions, and Social Roles: The Effects and Performance of Gender in Campaigns for the United States Congress
In my dissertation, I argue that analyzing the effects of gender in the context of other social identities can provide a better understanding of why women continue to struggle to get elected to Congress. Specifically, I show how the expectation that women will be parents negatively affects their ability to get elected to public office. Motherhood, and the expectation that women will be mothers, places a distinct bind on women as candidates, and this bind intersects with other known binds that women candidates face, such as the double bind. To conceptualize this third bind, I trace societal expectations of mothers using feminist theory, public opinion polls, and popular portrayals of women. Using original survey experiments, surveys, and content analysis of the communication of women candidates for Congress, I demonstrate how expectations of motherhood present added difficulties for women. Taken together, my research shows the need to add a third bind to our understanding of the difficulties women candidates face when they run for office.
Brehm, John and Meg Savel. (In Press)"What do survey measures of trust actually measure?" In Trust in Contemporary Society, edited by Masamichi Sasaki. Leiden; Boston: Brill.
“Generalized trust” is the idea that individuals have a predisposition to trust others, affected by particulars of the trustee and the purpose of trust, but consistent nonetheless. Opinion surveys are one of the most widely used instruments available to social scientists (and the media) to document generalized trust, and what appears to be a catastrophic decline in trust in public institutions, and even in trust in one another. Yet like all instruments by which one might empirically assess trust, survey measures suffer from specific weaknesses. While one of the most persuasive ways to assess generalized trust would be to measure the respondent’s reported trust in a wide range of actors, the act of repeatedly measuring multiple trustees with the same question introduces the problem of response set. Response set arises when respondents retreat to an anchoring point to answer questions instead of considering each question on its own. Further, any appraisal of risk or even of making simple assessments is prone to effects due to the mood of the respondent. We show in two separate analyses that these two factors do play a role in assessments of trust, but that trust itself still systematically covaries with more well–known substantive measures, if to a lesser degree.
"A new approach to modeling attitudes in two dimensions" with John Brehm
Public opinion scholars have long known that people may simultaneously hold more than one opinion about other people, matters of politics, and even their own choices. The conventional response would be to consider the multiple dimensions separately, or to reduce the dimensional space to a single construct. There is a tractable solution that measures the attitudes of multiple dimensions simultaneously. We use the bivariate beta density, a statistical distribution bounded over the unit square. We offer two demonstrations. One, conducted in summer 2016, measures the compassion and skill of women and men politicians, demonstrating greater uniformity of reactions to women than men, higher combined skill and compassion ratings for women over men, and a greater appraisal of the compassion of women over men. The second, conducted in 2007–8, measures positive and negative assessments of politicians, showing that ambivalent, unipolar, and nonpolar reactions are equally abundant.
"Gender Stereotypes and Family in 2016 Senate Races"
Women candidates integrate references to family with their discussion of policy to demonstrate that they fulfill traditional gender norms. Previous literature has found ambiguous re- sults when it comes to the effects of gender stereotypes on vote choice. However, women as candidates continue to be concerned with gender stereotypes and consciously consider them when creating their public image. Political candidates fear they will be penalized if they are viewed as norm violators by constituents, and one of the biggest norm violations a woman can make is not appearing traditionally feminine and maternal. This paper provides a theoretical argument for how gender and family factor into the images women candidates create, in addition to providing qualitative support from candidate controlled communication with constituents: debate performances, campaign webpages, and campaign advertisements. Women candidates emphasize gender norms and family even when they do not have "traditional families” themselves, and do so to the greatest degree when the majority of their audience is unlikely to have high levels of political knowledge or interest (e.g. in campaign advertisements).
"Policy without Partisanship: The Direct Appeals of First Ladies" with Shu Fu
Presidents make public appeals on behalf of their policy priorities, but they are not the only members of presidential administrations who address the public. First ladies are highly visible presidential surrogates. We argue that first ladies make direct appeals to advance presidents' policy initiatives, but do so without being overly partisan. To support these claims we present evidence from the public remarks of the last three first ladies whose husbands have completed their terms (Obama, Bush, and Clinton). We use structural topic modeling to show that the remarks of first ladies are primarily concerned with policy, rather than ceremonial topics. We measure the partisanship of public remarks using a dictionary-based approach with Bayesian shrinkage and regularization to illustrate how the remarks of first ladies are not overly partisan. We demonstrate why the remarks of first ladies should be characterized as direct appeals, even though they are not partisan.