Why women underperform relative to men in high-stakes exams while excelling under lower stakes remains a puzzle. Previous research suggests differential responses to pressure as an explanation. We evaluate the impact of a unique intervention assigning STEM program applicants in Uruguay to in-exam stress management exercises involving positive stress interpretation and a brief meditation. Relative to control women, treated women omit fewer questions, resulting in a 0.13 SD boost in overall performance. Consequently, the admissions gender gap closes and 10% more women are admitted to the program as a result of the intervention. The main insight is that omitting fewer questions is key for reducing gender gaps in performance. We also highlight that stress management exercises can go a long way in increasing the share of women in STEM education with potential downstream effects on gender earnings and diversity within these fields. 

We study whether the known gender composition of the pool of candidates affects the performance of women in a high-stakes exam. We use data from a natural experiment that changed the gender composition of the candidates for a nation-wide admission test to a coding educational program. Our identification strategy exploits the fact that both men and women were accepted to the admission test in all years except for 2019 when only women were allowed to take it. We find that in absence of male competitors, women perform better in areas where men typically do better, such as math and logical reasoning, while no significant effects are observed in verbal tasks, where men typically do not do better. Our findings are consistent with the hypothesis of that stereotype threat is deactivated in absence of male competitors. We also find that in absence of men, women exert higher effort, particularly in male-favored domains. Our results highlight the importance of gender composition when evaluating women's performance in fields traditionally dominated by men.

Adequate nutrition is essential for children's physical and mental development. Existing evidence of food insecurity and child behavior problems is limited, especially in low and middle-income countries, due in part to a lack of data. I leverage new longitudinal data for a middle-income country in Latin America, a region with a high incidence of food insecurity. The sample follows 1,539 Uruguayan children aged 2, 4 and 7 years. I investigate household food insecurity trajectories between early childhood and preschool age and its predictive power on behavior problems during school age. I employ the cumulative value-added model to adjust for unobserved omitted variables related to the lagged outcome variable. The findings show that household food insecurity is associated with more child behavior problems, regardless of when the household was classified as food insecure. Results are larger for externalizing behavior, particularly for boys. When children are 7, the persistence of household food insecurity predicts an increase in child behavior problems ranging from 0.25 to 0.30 standard deviations for internalizing and externalizing behavior, respectively. Evidence suggests that maternal mental health and parental warmth may act as mediators. The results were robust to different specifications, and especially to omitted variable bias.

Recent trends indicate a growing level of protest worldwide, particularly in developing countries. However, there is variation in protest strategies, with some countries opting for non-violent means while others resort to violence. This paper aims to investigate the factors underlying this difference and whether violent and non-violent protests are interconnected within the same underlying process. To achieve this, we propose an econometric model that, by controlling for observable and (time varying) unobservable variables, estimates the dynamic relationship between non-violent and violent protests, extending the class of finite mixture models for longitudinal count data to the bivariate case. In our model, the number of non-violent and violent protests are jointly modelled and influenced by a latent process following a first-order Markov chain. We utilize data from the Social Conflict Armed Dataset (SCAD), focusing on African countries. Our findings reveal a high level of persistence in protests, with past protests significantly correlated with present protests. Specifically, an additional violent protest in the past leads to a 15%-20% increase in the number of present violent protests, while an additional non-violent protest results in a 7%-8% increase in the number of present non-violent protests. 

Multiple-choice exams are used worldwide to select candidates in educational and job programs. Empirical evidence has found a significant gender gap in performance in these exams, partly because women skip more questions than men. The aim of this study is to investigate the potential factors behind this behavior in a unique setting where information on how incorrect answers are scored remains undisclosed. I focus on examining differences in choice consistency and risk preference. I use revealed preference theory as a criterion for choice consistency: Choices from a finite collection of budget lines are consistent with maximizing a utility function if and only if they satisfy the Generalized Axiom of Revealed Preferences (GARP). I focus on choice consistency in the risk domain. The findings indicate that consistency is strongly correlated with the accuracy rate for both men and women, while the gender gap in omitted questions is largely attributable to differences in risk aversion, with women exhibiting higher levels of risk aversion compared to men.

This project has been approved by the Research Ethics Committee (CEI) of the Faculty of Economics and Administration, University of the Republic; and the Ethics Committee on Animal and Human Experimentation (CEEAH), Universitat Autónoma de Barcelona. The pre-analysis plan (PAP) was preregistered in the Open Science Framework (OFS). 

This paper explores the effect of an educational intervention using adaptive technology combined with personalized teaching on grit and growth mindset. To do so, we conduct a quasi-experimental elementary public school design in association with a local government. Using a sample of 100 schools, where half of the schools are treated, we estimate the causal effect of a High-Tech High-Touch intervention on grit and growth mindset for children of 10 years old. Since the intervention implies personalized teaching and the use of an adaptive math platform adequate to the student's academic level, we expect positive effects on grit and a growth mindset.

This project has been approved by the  Ethics Committee on Animal and Human Experimentation (CEEAH), Universitat Autónoma de Barcelona. The pre-analysis plan (PAP) was preregistered in the Open Science Framework (OFS).