Research
Publications
Gender Differences in Willingness to Guess (2014) in Management Science
We present the results of an experiment that explores whether women are less willing than men to guess on multiple-choice tests. Our test consists of practice questions from SAT II subject tests; we vary whether a penalty is imposed for a wrong answer and the salience of the evaluative nature of the task. We find that when no penalty is assessed for a wrong answer, all test-takers answer every question. But, when there is a penalty for wrong answers, women answer significantly fewer questions than men. We see no differences in knowledge of the material or confidence in the test-takers, and differences in risk preferences explain less than half of the observed gap. Making the evaluative aspect of the test more salient does not impact the gender gap. We show that, conditional on their knowledge of the material, test-takers who skip questions do significantly worse on our test.
Evidence on Self-Stereotyping and the Contribution of Ideas (2014) in The Quarterly Journal of Economics
We use a lab experiment to explore the factors that predict an individual's decision to contribute her idea to a group. We find that contribution decisions depend upon the interaction of gender and the gender stereotype associated with the decision-making domain: conditional on measured ability, individuals are less willing to contribute ideas in areas that are stereotypically outside of their gender's domain. Importantly, these decisions are largely driven by self-assessments, rather than fear of discrimination. Individuals are less confident in gender incongruent areas and are thus less willing to contribute their ideas. Because even very knowledgeable group members under-contribute in gender incongruent categories, group performance suffers and, ex post, groups have difficulty recognizing who their most talented members are. Our results show that even in an environment where other group members show no bias, women in male-typed areas and men in female-typed areas may be less influential. An intervention that provides feedback about a woman's (man's) strength in a male-typed (female-typed) area does not significantly increase the probability that she contributes her ideas to the group.
Link to Online Appendix, Note on Subsection III.E
Laboratory Evidence on the Effects of Sponsorship on the Competitive Preferences of Men and Women (2016) with Nancy R. Baldiga in Management Science
Sponsorship programs have been proposed as one way to promote female advancement in competitive career fields. A sponsor is someone who advocates for a protégé, and in doing so, takes a stake in her success. We use a laboratory experiment to explore two channels through which sponsorship has been posited to increase advancement in a competitive workplace. In our setting, being sponsored provides a vote of confidence and/or creates a link between the protégé’s and sponsor’s payoffs. We find that both features of sponsorship significantly increase willingness to compete among men on average, while neither of these channels significantly increases willingness to compete among women on average. As a result, sponsorship does not close the gender gap in competitiveness or earnings. We discuss how these insights from the laboratory could help to inform the design of sponsorship programs in the field.
Stereotypes (2016) with Pedro Bordalo, Nicola Gennaioli, and Andrei Shleifer in The Quarterly Journal of Economics
We present a model of stereotypes based on Kahneman and Tversky’s representativeness heuristic. A decision maker assesses a target group by overweighting its representative types, which we formally define to be the types that occur more frequently in that group than in a baseline reference group. Stereotypes formed in this way contain a “kernel of truth”: they are rooted in true differences between groups. They are also context dependent: beliefs about a group depend on the characteristics of the reference group. Because stereotypes emphasize differences, they cause belief distortions, particularly when groups are similar. In line with our predictions, beliefs in the lab about abstract groups and beliefs in the field about political groups are context dependent and distorted in the direction of representative types.
The Size of the LGBT Population and the Magnitude of Anti-Gay Sentiment are Substantially Underestimated (2017) with Lucas C. Coffman and Keith M. Marzilli Ericson in Management Science
We demonstrate widely-used measures of anti-gay sentiment and the size of the LGBT population are misestimated, likely substantially. In a series of online experiments using a large and diverse but non-representative sample, we compare estimates from the standard methodology of asking sensitive questions to measures from a “veiled” methodology that precludes inference about an individual but provides population estimates. The veiled method increased self-reports of anti-gay sentiment, particularly in the workplace: respondents were 67 percent more likely to disapprove of an openly gay manager when asked with a veil, and 71 percent more likely to say it should be legal to discriminate in hiring on the basis of sexual orientation. The veiled methodology also produces larger estimates of the fraction of the population that identifies as LGBT or has had a sexual experience with a member of the same sex. Self-reports of non-heterosexual identity rose by 65 percent, and same-sex sexual experiences by 59 percent. We conduct a “placebo test” and show that for non-sensitive placebo items, the veiled methodology produces effects that are small in magnitude and not significantly different from zero in seven out of eight items. Taken together the results suggest anti-gay discrimination might be a more significant issue than formerly considered, as the non-heterosexual population and anti-gay workplace-related sentiment are both larger than previously measured.
Beliefs about Gender (2019) with Pedro Bordalo, Nicola Gennaioli, and Andrei Shleifer in American Economic Review
We conduct laboratory experiments that explore how gender stereotypes shape beliefs about ability of oneself and others in different categories of knowledge. The data reveal two patterns. First, men's and women's beliefs about both oneself and others exceed observed ability on average, particularly in difficult tasks. Second, overestimation of ability by both men and women varies across categories. To understand these patterns, we develop a model that separates gender stereotypes from misestimation of ability related to the difficulty of the task. We find that stereotypes contribute to gender gaps in self-confidence, assessments of others, and behavior in a cooperative game.
Link to Experimental Materials
The Impact of Penalties for Wrong Answers on the Gender Gap in Test Scores (2020) with David Klinowski in Proceedings of the National Academy of Sciences
Multiple-choice exams play a critical role in university admissions across the world. A key question is whether imposing penalties for wrong answers on these exams deters guessing from women more than men, disadvantaging female test-takers. We consider data from a large-scale, high-stakes policy change that removed penalties for wrong answers on the national college entry exam in Chile. The policy change reduced a large gender gap in questions skipped. It also narrowed gender gaps in performance, primarily among high-performing test-takers, and in the domains of math, social science, and chemistry.
The Role of Beliefs in Driving Gender Discrimination (2020) with Christine Exley and Muriel Niederle in Management Science
(formerly circulated as "When Gender Discrimination is Not About Gender")
While there is ample evidence of discrimination against women in the workplace, it can be difficult to understand what factors contribute to discriminatory behavior. We use an experiment to both document discrimination and unpack its sources. First, we show that, on average, employers prefer to hire male over female workers for a male-typed task even when the two workers have identical resumes. Second, and most critically, we use a control condition to identify that this discrimination is not specific to gender. Employers are simply less willing to hire a worker from a group that performs worse on average, even when this group is instead defined by a non-stereotypical characteristic. In this way, beliefs about average group differences are the key driver of discrimination against women in our setting. We also document some evidence for in-group preferences that contribute to the gender discrimination observed. Finally, our design allows us to understand and quantify the extent to which image concerns mitigate discriminatory behavior.
Link to Instructions
Memory and Representativeness (2020) with Pedro Bordalo, Nicola Gennaioli, Frederik Schwerter, and Andrei Shleifer in Psychological Review
We explore the idea that judgment by representativeness reflects the workings of episodic memory. In our model, subjects assess the probability distribution of types in a group in terms of the ease of retrieval of types when cued with the group. Retrieval is driven by a measure of similarity which exhibits interference across groups, and is thus context-dependent. In a new laboratory experiment on cued recall, participants are shown two groups of images with different distributions of colors. In line with the model's predictions, we find that i) decreasing the frequency of a given color in one group significantly increases the recalled frequency of that color in the other group, ii) for a fixed set of images, different cues for the same objective distribution entail different interference patterns and different probabilistic assessments. A calibration of the model yields a good quantitative fit with the data. Selective retrieval and interference may offer a foundation for the representativeness heuristic, but more generally for understanding the formation of probability judgments from experienced statistical associations.
Link to Appendix
Gender Stereotypes in Deliberation and Team Decisions (2021) with Clio Bryant Flikkema and Olga Shurchkov in Games and Economic Behavior
We explore how groups deliberate and decide on ideas in an experiment with communication. We find that gender biases play a significant role in which group members are chosen to answer on behalf of the group. Conditional on the quality of their ideas, individuals are less likely to be selected in gender incongruent domains (i.e. male-typed domains for women; female-typed domains for men). Individuals are also less likely to promote themselves when they are in the gender minority within their group. These patterns are not well-explained by objective or subjective differences in conversational behavior, nor by differences in beliefs about idea quality. Our results seem most consistent with a preference for promoting and rewarding group members in a way that conforms with gender norms.
Link to Appendix and Instructions
Whether to Apply? (2023) with Manuela Collis and Leena Kulkarni, accepted at Management Science
Labor market outcomes depend, in part, upon an individual’s willingness to put herself forward for different opportunities. We use a series of experiments to explore gender differences in willingness to apply for higher return, more challenging work. We find that, in male-typed domains, qualified women are significantly less likely to apply than similarly well-qualified men. We provide evidence both in a controlled setting and in the field that reducing ambiguity surrounding required qualifications increases the rate at which qualified women apply. The effects are more mixed for men. Our results suggest a path for increasing the pool of qualified female applicants.
Stereotypes and Belief Updating (2023) with Manuela Collis and Leena Kulkarni, accepted at the Journal of the European Economic Association
We explore how self-assessments respond to feedback about own ability across a range of tasks, with a particular focus on how gender stereotypes impact belief updating. Participants in our experiments take tests of their ability across different domains. Absent feedback, beliefs of own ability are strongly influenced by gender stereotypes: holding own ability fixed, individuals are more confident in gender congruent domains (i.e., male-typed domains for men, female-typed domains for women). We then provide noisy feedback about own absolute performance to participants and elicit posterior beliefs. Gender stereotypes have significant predictive power for posterior beliefs, both through their influence on prior beliefs, as predicted by the Bayesian model, but also through their influence on updating, a non-Bayesian channel. Both men and women’s beliefs are more responsive to information in gender congruent domains than gender incongruent domains. This is primarily driven by differential reactions to exogenously-received good news about own ability: both men and women react more to good news when it arrives in a gender congruent domain than when it arrives in a gender incongruent domain. Our results have important implications for understanding how feedback shapes, and perpetuates, gender gaps in self-assessments.
Imagining the Future: Memory, Simulation, and Beliefs with Pedro Bordalo, Giovanni Burro, Nicola Gennaioli, and Andrei Shleifer, accepted at the Review of Economic Studies
How do people form beliefs about novel risks, with which they have little or no experience? Motivated by survey data we collected in 2020, which showed that beliefs about Covid’s lethality depended on a range of personal experiences in unrelated domains, we build a model based on the psychology of selective memory. When a person thinks about an event, different experiences compete for retrieval, and retrieved experiences are used to simulate the event based on how similar they are to it. The model yields predictions on how experiences interfere with each other in recall and how non domain-specific experiences bias beliefs based on their similarity to the assessed event. We test these predictions using data from our Covid survey and from a primed-recall experiment about cyberattack risk. Experiences and their measured similarity to the cued event successfully help explain beliefs, with patterns consistent with our theory. Our approach offers a new, structured way to study and jointly account for systematic biases and substantial belief heterogeneity.
Replaces previous working paper Older People are Less Pessimistic about the Health Risks of Covid-19
A Dynamic Investigation of Stereotypes, Belief Updating, and Behavior with Maria Paola Ugalde Araya and Basit Zafar, accepted at Economic Inquiry
Many decisions – such as what educational or career path to pursue – are dynamic in nature, with individuals receiving feedback at one point in time and making decisions later. Using a controlled experiment, with two sessions one week apart, we analyze the dynamic effects of feedback on beliefs about own performance and decision-making across two different domains (verbal skills and math). We find significant gender gaps in beliefs and choices before feedback: men are more optimistic about their performance and more willing to compete than women in both domains, but the gaps are significantly larger in math. Feedback significantly shifts individuals' beliefs and choices. Despite this, we see substantial persistence of gender gaps over time. This is particularly true among the set of individuals who receive negative feedback. We find that, holding fixed performance and decisions before feedback, women update their beliefs and choices more negatively than men do after bad news. Our results are not well-explained by motivated reasoning; participants accurately recall their feedback at high rates, and negative feedback is more likely to be recalled than positive feedback. Overall, our results highlight the challenges involved in overcoming gender gaps in dynamic settings.
Gender and Preferences for Performance Feedback with David Klinowski, accepted at Management Science
Across multiple studies, we investigate whether there are gender differences in preferences for receiving performance feedback. We vary many features of the feedback context: whether the performance task is a cognitive test or a mock interview, whether the feedback is objective or subjective, and whether it is possible for the provider of the feedback to discriminate on the basis of gender. Consistent with past work, we find that women are less optimistic about their performance than men, and that, on average, more optimistic individuals have greater demand for feedback.. Results like these have been hypothesized in the literature to imply that women will shy away from performance feedback more so than men. And, when we survey participants from a similar population, they also anticipate that women will demand feedback at lower rates than men. Yet, across our two incentivized studies, we find that women are no less eager to receive performance feedback than men.
Assent-maximizing Social Choice (2013) with Jerry R. Green in Social Choice and Welfare
We take a decision theoretic approach to the classic social choice problem, using data on the frequency of choice problems to compute social choice functions. We define a family of social choice rules that depend on the population’s preferences and on the probability distribution over the sets of feasible alternatives that the society will face. Our methods generalize the well-known Kemeny Rule. In the Kemeny Rule, it is known a priori that the subset of feasible alternatives will be a pair. We define a distinct social choice function for each distribution over the feasible subsets. Our rules can be interpreted as distance minimization—selecting the order closest to the population’s preferences, using a metric on the orders that reflects the distribution over the possible feasible sets. The distance is the probability that two orders will disagree about the optimal choice from a randomly selected available set. We provide an algorithmic method to compute these metrics in the case where the probability of a given feasible set is a function only of its cardinality.
Representative Democracy and the Implementation of Majority-Preferred Alternatives (2016) in Social Choice and Welfare
In this paper, we contrast direct and representative democracy. In a direct democracy, individuals have the opportunity to vote over the alternatives in every choice problem the population faces. In a representative democracy, the population commits to a candidate ex ante who will then make choices on its behalf. While direct democracy is normatively appealing, representative democracy is the far more common institution because of its practical advantages. The key question, then, is whether representative democracy succeeds in implementing the choices that the group would make under direct democracy. We find that, in general, it does not. We analyze the theoretical setting in which the two methods are most likely to lead to the same choices, minimizing potential sources of distortion. We model a population as a distribution of voters with strict preferences over a finite set of alternatives and a candidate as an ordering of those alternatives that serves as a binding, contingent plan of action. We focus on the case where the direct democracy choices of the population are consistent with an ordering of the alternatives. We show that even in this case, where the normative recommendation of direct democracy is clear, representative democracy may not elect the candidate with this ordering.