Experiments on Gender Differences in Behavioral Traits
Paper Session
Saturday, Jan. 6, 2024 10:15 AM - 12:15 PM (CST)
- Chair: Ragan Petrie, Texas A&M University
Signaling Confidence
Abstract
We study gender differences in confidence, and the impact of signaling confidence, on employment likelihood in an experimental hiring market. We document that moderate, as opposed to low, levels of confidence enhance the chances of being hired, while excessive confidence, characterized by high performance estimates and high certainty, diminishes employment prospects. Men display higher levels of confidence than women, and this gender gap widens in forward-looking scenarios, where performance estimates are provided ex ante rather than ex post. These findings serve as a cautionary reminder of the potential negative consequences of excessive confidence. Hence, encouraging both men and women to signal performance predictions that are aligned with reality instead of simply advising women to be more confident may well be a safer way to foster equity in labor market outcomes.Early Onset of Gendered Work Assignments
Abstract
Research has shown that women more than men perform “non-promotable tasks” (NPTs) in the workplace. This study investigates how early unequal task allocations emerge and the impact of unequal task assignments on career progression. We collect survey, experimental, and administrative data from cohorts of MBA students graduating from the Norwegian School of Economics before and after they enter the labor market and follow their career trajectories. Results show that as early as 1 year after entering the labor market, gender differences in task allocations emerge with women more than men reporting that they perform more non-promotable tasks than their peers. This is not driven by differences in preferences or by men and women initially wanting or expecting different things from work, or from selection into different industries, firms, or occupations. It is a result that is stronger in male-dominated sectors and which seems to be driven by gendered expectations of who will perform NPTs, consistent with prior findings (Babcock et al 2017).Lying Out of Obligation: Cheap Talk in an Ultimatum Game with Outside Obligations
Abstract
We introduce an “outside obligation” in an ultimatum game with a real effort task. Responders may have an obligation to pay a portion of their earnings to a third party, ranging from 0 to 50 percent. Responders may report their outside obligation, which is cheap talk, and propose a split. The Proposer receives the indicated obligation and the suggested split and offers a counter split. If the offer is accepted, the Responder completes a task to generate the surplus; if the Responder reject the counter split, both parties must complete the task to receive a small payment. The experimental procedure carefully controls for the gender of the Proposers and Responders. Overall, we find that Responders significantly misreport their outside obligation, over-reporting 80 percent of the time when the obligation is zero. Men inflate their outside obligation to the greatest extent more frequently and propose a higher split for themselves. Women also overstate their level of obligation but to a lesser extent. We find that Proposer’s offers are higher when they believe the Responder has a larger outside obligation, and consistent with environments that have no signal at all when they do not believe the signal received.He Said, She Said: Who Gets Believed When Spreading (Mis)Information
Abstract
We design an online experiment meant to closely mimic interactions on Twitter in order to gauge whether people’s trust or mistrust of information contained in a tweet depends on the (perceived) gender of its author. In particular, subjects view tweets appearing in a stylized “Twitter feed” in random order, each randomly assigned a unique avatar from a set containing images of men, women, and gender neutral images. Some tweets are objectively false and some are true, as rated by a third-party entity. On average, tweets posted by women are actually less likely to be flagged as misinformation as tweets authored by men. Women are also significantly better than men at identifying and flagging false tweets (and not flagging true ones). We further investigate whether the gender-congruency of the tweet’s topic impacts the likelihood of it being flagged as misinformation as opposed to being liked/retweeted, and by whom. Here, we find gender differences: men are more likely than women to be believed when posting on male-typed topics (military and finance), while women are more likely than men to be believed when tweeting about female-typed topics (education and healthcare). We also explore whether truthfulness or fakeness of the tweet matters in these gendered contexts.Discussant(s)
Olga Stoddard
,
Brigham Young University
Brianna Halladay
,
Trinity College
Siri Isaksson
,
NHH
Johanna Mollerstrom
,
George Mason University
Danila Serra
,
Texas A&M University
JEL Classifications
- C9 - Design of Experiments
- J1 - Demographic Economics