Research

Job Market Paper

“Be Prepared” for School: Scouting and School Outcomes in England

Previously circulated as “The Impact of Scouting on School Absences and Crime Rate: Evidence from England”

Winner of the María Jesús San Segundo Award (AEDE), 2026

Abstract. Scouting and Guiding are among the largest and most enduring youth organisations in the world, yet their effects on children have rarely been studied with causal methods. Exploiting the staggered opening of Scout and Girlguiding groups within walking distance of English schools between 2006 and 2018, I estimate the effect of gaining local access to a group on school absence, behaviour and attainment using a doubly-robust difference-in-differences design. The opening of a group within 0.75 km of a school reduces the unauthorised absence rate by about 9 percent and the suspension rate by about 10 percent, with effects that build up over the years after the opening, while permanent exclusions do not change. Openings are not predicted by schools’ prior outcomes, authorised absence does not fall, and neither placebo openings in time nor openings of sports charities reproduce the effects, which also hold across distances, treatment definitions and estimators. The fall in unexplained absence and truancy is concentrated in schools where persistent absence and suspensions are common, and it is in the secondary schools of this kind that GCSE attainment shows suggestive gains. Individual-level evidence from Understanding Society points to the same channel: children in areas where a group opens are less likely to report truanting, watch less television and spend more time with friends outside the home. These effects are comparable in size to those of school-based attendance and behaviour interventions, but come from a volunteer-run organisation operating outside school at almost no public cost.

Publications

Do LVLMs See Stereotypes as We Do? A Social Science Framework for Evaluating LVLMs’ Recognition and Explanation of Stereotypes in Images

with Marco Madeddu, Chiara Ferrando, Horacio Jarquín Vásquez, Marina Della Giusta and Viviana Patti

Proceedings of CLiC-it 2026

Abstract The strong impact that stereotypes have in shaping society made them a debated research topic. In this contribution, we adapt a test originally from behavioral economy to investigate the abilities of LVLMs in recognizing and explaining stereotypes in images. In order to collect higher quality data, we rely on images drawn from a real-life context (textbooks) rather than synthetically generated content. To achieve this purpose, we adapt the Stereotype Identification Test (SIT), created for evaluating human ability to identify biases in images. The original dataset consists of 100 images from educational materials that 614 Italian teachers rated as stereotypical on a scale of 1 to 5. In addition to the ratings, the dataset includes over 12,900 free-text comments in which participants explain their judgments and describe the stereotypes depicted. In the experimental phase, we evaluate five visual LLMs under different configurations of the SIT to measure their abilities in recognizing stereotypes in images. Our results reveal substantial discrepancies between model and human behavior: models consistently assign lower stereotypes ratings and tend to mitigate the impact of stereotypes, suggesting a systematic underestimation of stereotypes relative to human perception.

Working Papers

Seeing Stereotypes

with Pietro Biroli, Marina Della Giusta and Florent Dubois

arXiv:2503.02146

Abstract Reliance on stereotypes is a persistent feature of human decision-making and has been extensively documented in educational settings, where it can shape students’ confidence, performance, and long-term human capital accumulation. While effective techniques exist to mitigate these negative effects, a crucial first step is to establish whether teachers can recognize stereotypes in their professional environment. We introduce the Stereotype Identification Test (SIT), a novel survey tool that asks teachers to evaluate and comment on the presence of stereotypes in images drawn from school textbooks. Their responses are systematically linked to established measures of implicit bias (Implicit Association Test, IAT) and explicit bias (survey scales on teaching stereotypes and social values). We additionally asked several large language models (LLMs) to complete our SIT test to capture the wisdom of the crowd about the currently perceived level of stereotype associated with each image. Our findings demonstrate that the SIT is a valid and reliable measure of stereotype recognition and teachers’ perception of stereotypes is similar to the one obtained with the wisdom of the crowd. Teachers’ ability to recognize stereotypes is linked to trainable traits such as implicit bias awareness and inclusive teaching practices. Moreover, providing personalized feedback on implicit bias improves SIT scores by 0.25 standard deviations, reinforcing the idea that stereotype recognition is malleable and can be enhanced through targeted interventions.

Work in Progress

Sex Education: Mandatory School Programmes and Adolescent Health in Europe

with Chiara Gastaldi and Andrea Rebucini

Abstract This paper examines the effects of mandatory sex education on adolescent reproductive and sexual health in Europe. We combine cross-country variation in the timing of policy adoption with measures of births, fertility, abortions, and sexually transmitted infections (STIs) among adolescents aged 15–19 from the Global Burden of Disease (GBD) study, produced by the Institute for Health Metrics and Evaluation (IHME). We estimate average treatment effects using a staggered difference-in-differences design that exploits cross-country variation in the timing of policy adoption. The results show gradual and persistent reductions in adolescent births and fertility, with effects increasing as exposure to the policy accumulates. Abortion rates also decline, particularly at longer post-treatment horizons, suggesting that mandatory sex education reduces adolescent pregnancies rather than merely shifting pregnancy outcomes from births to abortions. The policy is additionally associated with a reduction in the STI rate. This aggregate effect is smaller than the fertility-related effects and masks substantial heterogeneity across individual infections. Overall, the findings indicate that the implementation of mandatory sex education programmes in schools improves adolescent reproductive and sexual health, although its effects emerge gradually and vary across outcomes.

Words That Matter: Gender, Ethnicity, and Performance Assessment in the Workplace

with Marina Della Giusta, Sylvia Jaworska, Giovanni Razzu and Almudena Sevilla

Abstract Performance evaluations determine pay, retention pay and career progression, but the criteria against which employees are judged often embed a narrow, biased and historically male template of what good performance looks like. This can result in misallocation of talent within firms. We ask whether rewriting the language of evaluation criteria changes who is recognised as a top performer. We study a large international services firm in which three of six teams evaluated staff using revised, more inclusive guidance while the other three did not. Our unique dataset combines 1,054 written evaluations with direct observation and linguistic coding of the panel meetings in which final ratings were agreed. Under the revised criteria, the gap in the probability of being rated as outperforming between non-white and white employees is about 18 percentage points more favourable to non-white employees. This estimate is positive in every specification we run, although with only six teams it is not statistically significant. In treated panels, discussion relies less on peer comparisons and sales figures, gives more attention to a broader set of competencies, and is less negative in tone towards women. Redesigning the language of evaluation therefore offers a low-cost institutional lever that targets the instrument rather than the evaluator.