Published and Working Papers
“What We Teach About Race and Gender: Representation in Images and Text of Children’s Books”
(with A. Eble, E. Harrison, H.B. Runesha, T. Szasz) (July 2022, NBER Working Paper 29123, Revision invited at Quarterly Journal of Economics)
Books shape how children learn about society and norms, in part through representation of different characters. We introduce new artificial intelligence methods for systematically converting images into data and apply them, along with text analysis methods, to measure the representation of race, gender, and age in award-winning children’s books from the past century. We find that more characters with darker skin color appear over time, but the most influential books persistently depict a greater proportion of light-skinned characters than other books, even after conditioning on race; we also find that children are depicted with lighter skin than adults. Relative to their growing share of the U.S. population, Black and Latinx people are underrepresented in these same books, while White males are overrepresented. Over time, females are increasingly present but appear less often in text than in images, suggesting greater symbolic inclusion in pictures than substantive inclusion in stories. We then report empirical evidence for predictions about the supply of and demand for representation that would generate these patterns. On the demand side, we show that people consume books that center their own identities. On the supply side, we document higher prices for books that center non-dominant social identities and fewer copies of these books in libraries that serve predominantly White communities. Lastly, we show that the types of children’s books purchased in a neighborhood are related to local political beliefs. [Slides] [Research Brief] [AI Research Brief (SSN)] [Video] [Interactive Figures] [Google Case Study] [Press: School Library Journal, Wall Street Journal, Code Together, Inequalitalks, FutureEd, The 74, ParentData, New York Times] [Recognition: Named one of the ten most significant education studies of 2021 by George Lucas Foundation’s Edutopia, Received the 2021 Early Career Product Award from the Education Policy Collaborative, Received the first place poster prize at APPAM 2021, Received the 2021 Google Customer Award for Education and the 2021 Google Customer Award for Diversity, Equity & Inclusion]
“Portrayals of Race and Gender: Sentiment in 100 Years of Children’s Literature” (with C. Christ, A. Das, A. Raj) (June 2022, Proceedings of the ACM SIGCAS/SIGCHI Conference on Computing and Sustainable Societies (ACM COMPASS ’22))
The way that people of different identities are portrayed in children’s books can send subconscious messages about how positively or negatively children should think about people with those identities. These messages can then shape the next generation’s perceptions and attitudes about people, which can have important implications for belief formation and resource allocation. In this paper, we make two contributions: (1) we examine the depiction of race and gender in award-winning children’s books from the last century, and (2) we examine how consumption of these books relates to local beliefs. First, we analyze the sentiment associated with the famous individuals mentioned in these books. While the sentiment surrounding women is positive overall, on average, we see that Black women are more often portrayed with negative sentiment in Mainstream books, while White women are more often portrayed with positive sentiment. Because children’s books in the US depict more White women overall, this disguises the more negative intersectional portrayals of Black women. Books that center underrepresented identities are more likely to portray all characters with more positive sentiment. A century ago, women were much less positively spoken about than men, but the average sentiment of females and males has converged over time. The difference in sentiment connected with Black people and White people has also decreased over time, but there still remains a substantial gap. Second, we then analyze the relationship between book purchases and local beliefs to understand the potential messages being transmitted to children in different parts of society. We see that more purchases of books with positive sentiment towards Black characters are associated with a larger proportion of individuals who believe that White people in the U.S. have certain advantages because of the color of their skin and who are angry that racism exists. Understanding the messages that may be implicitly — or explicitly — sent to children through highly influential books can lend insight into the factors that may shape children’s beliefs and attitudes.
“From Retributive to Restorative: An Alternative Approach to Justice“ (with B. Feigenberg and F. Momeni) (May 2022, Working Paper)
Schools have traditionally employed punitive methods of discipline when trying to shape children’s behavior, but more recently, districts have been experimenting with restorative approaches to engaging with students. We use variation in the timing of the introduction of restorative practices across Chicago Public Schools to evaluate the impact of restorative approaches on student outcomes. We identify significant decreases in out-of-school suspensions and increases in perceived school climate in response to policy adoption. We also find evidence of a decrease in arrests that is consistent with spillovers in behavior outside of school. [Video] [Slides]
“Measuring Representation of Race, Gender, and Age in Children’s Books: Face Detection and Feature Classification in Illustrated Images” (with T. Szasz, E. Harrison, P-J. Liu, P-C. Lin, H.B. Runesha) (January 2022, Proceedings of the IEEE/CVF Winter Conference on Applications of Computer Vision (WACV))
Images in children’s books convey messages about society and the roles that people play in it. Understanding these messages requires systematic measurement of who is represented. Computer vision face detection tools can provide such measurements; however, state-of-the-art face detection models were trained with photographs, and 80\% of images in children’s books are illustrated; thus existing methods both misclassify and miss classifying many faces. In this paper, we introduce a new approach to analyze images using AI tools, resulting in data that can assess representation of race, gender, and age in both illustrations and photographs in children’s books. We make four primary contributions to the fields of deep learning and social sciences: (1) We curate an original face detection data set (IllusFace 1.0) by manually labeling 5,403 illustrated faces with bounding boxes. (2) We train two AutoML-based face detection models for illustrations: (i) using IllusFace 1.0 (FDAI); (ii) using iCartoon, a publicly available data set (FDAI_iC), each optimized for illustrated images, detecting 2.5 times more faces in our testing data than the established face detector using Google Vision (FDGV). (3) We curate a data set of the race, gender, and age of 980 faces manually labeled by three different raters (CBFeatures 1.0). (4) We train an AutoML feature classification model (FCA) using CBFeatures 1.0. We compare FCA with the performance of another AutoML model that we trained on UTKFace, a public data set (FCA_UTK) and of an established model using FairFace (FCF). Finally, we examine distributions of character identities over the last century across the models. We find that FCA is 34% more accurate than FCF in its race predictions. These contributions provide tools to educators, caregivers, and curriculum developers to assess the representation contained in children’s content. [Poster] [Video]
“Sanitation and Education” (April 2017, American Economic Journal: Applied Economics)
I explore whether the absence of school-sanitation infrastructure impedes educational attainment, particularly among pubescent-age girls, using a national Indian school latrine-construction initiative and administrative school-level data. School-latrine construction substantially increases enrollment of pubescent-age girls, though predominately when providing sex-specific latrines. Privacy and safety appear to matter sufficiently for pubescent-age girls that only sex-specific latrines reduce gender disparities. Any latrine substantially benefits younger girls and boys, who may be particularly vulnerable to sickness from uncontained waste. Academic test scores did not increase following latrine construction, however. Estimated increases in enrollment are similar across the substantial variation in Indian district characteristics. [Slides]
“Educational Investment Responses to Economic Opportunity: Evidence from Indian Road Construction”
(with S. Asher and P. Novosad) (January 2020, American Economic Journal: Applied Economics)
The rural poor in developing countries, once economically isolated, are increasingly being connected to outside markets. Whether these new connections crowd out or encourage educational investment is a central question. We examine the impacts on educational choices of 115,000 new roads built under India’s flagship road construction program. We find that children stay in school longer and perform better on standardized exams. Treatment heterogeneity supports the predictions of a standard human capital investment model: enrollment increases are largest where nearby labor markets offer the highest returns to education and lowest where they imply high opportunity costs of schooling. [Slides]
“Spillover Impacts on Education from Employment Guarantees” (January 2022, Education Finance and Policy)
Programs that guarantee some basic level of low-skill employment are a popular anti-poverty strategy in developing countries, with India’s employment-guarantee program (MGNREGA) annually employing adults in 23% of Indian households. An important concern is these employment programs may discourage children’s education and, thus, more-sustained long-run income growth. Using large-scale administrative data and household survey data, I estimate precise spillover impacts on education that reject substantive declines in children’s education from the rollout of MGNREGA. These negative spillovers are inexpensive to counteract, and small compared to immediate effects of MGNREGA on rural employment and poverty alleviation. [Takeaway (one-pager)]
“Religion and Sanitation Practices”
(with M. Alsan, K. Babiarz, J. Goldhaber-Fiebert, and L. Prince) (May 2021, World Bank Economic Review)
In India, infant mortality among Hindus is higher than among Muslims, and religious differences in sanitation practices have been cited as a contributing factor. To explore whether religion itself is associated with differences in sanitation practices, this study compares sanitation practices of Hindus and Muslims living in the same locations using three nationally representative data sets from India. Across all three data sets, the unconditional religion-specific gap in latrine ownership and latrine use declines by approximately two-thirds when conditioning on location characteristics or including location fixed effects. Further, the estimates do not show evidence of religion-specific differences in other sanitation practices, such as handwashing or observed fecal material near homes. Household sanitation practices vary substantially across areas of India, but religion itself has less direct influence when considering differences between Hindus and Muslims within the same location.
“Economic and Social Development along the Urban-Rural Continuum: New Opportunities to Inform Policy”
(with A. Cattaneo, D. Brown, L. Christiaensen, D.K. Evans, A. Haakenstad, T. McMenomy, M. Partridge, S. Vaz, D. Weiss) (September 2022, World Development)
The economic and social development of nations relies on their population having physical access to services and employment opportunities. For the vast majority of the 3.4 billion people living in rural locations, this largely depends on their access to urban centers of different sizes. Similarly, urban centers depend on their rural hinterlands. Building on the literature on functional areas/territories and the rural-urban continuum as well as insights from central place theory, this review paper advances the notion of catchment areas differentiated along an urban-to-rural continuum to capture these urban-rural interconnections. It further shows how a new, publicly available data set operationalizing this concept can shed new light on policy making across a series of development fields, including institutions and governance, urbanization and food systems, welfare and poverty, and access to health and education services. Together the insights support a more geographically nuanced perspective on development.
Other Writings: Policy Briefs, Op-Eds, Blog Posts
Road connectivity and rural economic development: Evidence from India’s rural roads programme (with S. Asher and Paul Novosad), VoxDev, July 2022.
India’s National Education Policy: A need to look beyond the classroom to improve results, VoxDev, November 2019.
How sanitation facilities in schools can improve educational outcomes, Ideas for India, August 2018.
Promoting Education through School Sanitation, World Bank Development Impact, March 2014.
“The Impact of the Financial Crisis on Tertiary Education World Wide” (with B. Long) World Bank, September 2009.
“The Broadmoor Project New Orleans Community Engagement Initiative: Progress Report” (with D. Ahlers, M. Blakley, L. Cole, M. El Dahshan, A. Hodari, H. Ko, J. Maeso, A. Noble, D. Radcliffe, M. Richards, C. Valentine, A. Van, D. Walsh, A. Watson, A. Woods, C. Wood, J. Wright, K. Yang). Harvard Kennedy School – Broadmoor Initiative, March 2007.