Working Papers: Economics
All SROs are not created equal: Heterogeneity in school policing (with Katherine Kwok) - Job Market Paper
The expansion of police presence in U.S. schools over the past two decades raises questions about the trade-off between student safety and exposure to the criminal justice system. In this paper, we study how school context moderates the effect that School Resource Officers (SROs) have on in-school offenses and school discipline. Using a difference-in-differences design that leverages a federal SRO hiring grant, we estimate bounds on the variance of the treatment effect distribution to detect treatment effect heterogeneity in SRO hiring and discover meaningful differences in how hiring SROs impacts the number of students suspended or referred to law enforcement. To explain the detected treatment effect heterogeneity, we consider observable school characteristics such as student demographics and school staff and unobservable differences in how schools implement their SRO program, measured by training text models to extract themes from grant narratives that describe the problems facing the school. We find that schools whose SRO programs focus on building community have decreased suspensions and referrals or arrests than schools that focus on the SRO’s fulfillment of traditional law enforcement duties. Our results highlight how a school’s context shapes the impact that SROs have in schools.
How many children are left behind? A meta-analysis of treatment effect heterogeneity in developing-country education programs (with Jason Kerwin, Juan Muñoz, Jeffrey Smith, Rebecca Thornton)
[Draft coming soon]
Extensive research on the effects of education programs in developing countries has left a key question unanswered: how many students actually benefit from these programs? We answer this question by re-analyzing the microdata from the universe of published education RCTs from developing countries. Despite the current enthusiasm for replication, we were able to obtain data for just 45% of the studies in our sampling frame. Our analytic sample includes more than half a million observations covering 123 different interventions run in 24 countries. Using this data, we construct non-parametric lower bounds on the across-student variance of treatment effects for each intervention. Our meta-analytic estimate of the standard deviation of treatment effects is 0.12 SDs of control-group test scores, which is nearly twice as large as the average treatment effect for the studies in our sample. The across-student variation in the effects of the same intervention is more than half as large as the variation in impacts across studies. Moreover, the standard deviation of impacts varies widely, from nearly zero for 20% of programs to almost half an SD of test scores for the top handful of interventions. The variance of treatment effects is strongly correlated with the average effect. However, we can explain almost none of the variance in treatment effects with the observed covariates from the studies, even if we use machine learning methods to estimate and partial out CATEs. Our results suggest that education interventions in developing countries leave over a quarter of children behind.
Working Papers: Other disciplines
All in the Family: Mothers of Children with Disabilities and Retirement (with Molly Costanzo, Lisa Klein Vogel, Victoria Knocke)
Revise & Resubmit at Children and Youth Services Review, 2025
Mothers of children with disabilities face unique tradeoffs as they approach retirement. These mothers often struggle to amass retirement savings due to their caregiving responsibilities and increased expenses but require greater economic resources in retirement as they continue to provide for their children’s financial and caregiving needs during retirement. We use a mixed-methods approach, leveraging data from the National Longitudinal Survey of Youth 1979 and data collected from interviews with mothers to examine differences in retirement outcomes for mothers of children with disabilities compared with other mothers and to understand how these mothers think about retirement options. Interviews and quantitative estimates identify caregiving responsibilities as a key consideration in mothers’ decisions about when, where, and how much to work. Qualitative data reveal that even among our relatively advantaged sample, caregiving, career trajectories, and expected financial needs of children affect retirement plans. Our findings suggest social workers can play an important role in supporting these families by connecting them to available benefits and resources. Findings also emphasize the importance of additional research on this topic with more diverse families.
Policy Reports
Wisconsin ELEVATE program evaluation
ELEVATE (Empowering Lives through Education, Vocational Assessment, Training, and Employment) was a child support-led program administered across five counties in Wisconsin that provided non-custodial parents who were behind on their child support obligations with case management, parenting services, and employment services. There was an impact evaluation of ELEVATE using a quasi-experimental design, and additional analysis using survey measures.
“Survey-Based Outcomes from the ELEVATE Evaluation,” with Molly Costanzo. University of Wisconsin-Madison Institute for Research on Poverty, CSRA 24-26-T10, 2025.
“Wisconsin’s ELEVATE Program: Final Evaluation Report,” with Hilary Shager, Molly Costanzo, Lisa Klein Vogel, Alexis Dennis, Yonah Drazen, Samina Hossain, Liesl Hostetter. University of Wisconsin-Madison Institute for Research on Poverty, 2025.
“Experiences, Characteristics, And Service Needs of Noncustodial Parents With Challenges Meeting Child Support Obligations: Evidence From ELEVATE Parents,” with Molly Costanzo, Yonah Drazen, Liesel Hostetter, Hilary Shager, and Lisa Klein Vogel. University of Wisconsin-Madison Institute for Research on Poverty, CSRA 22-24-T14, 2024.
State opioid policy data visualization
“OPTIC R Repository and Tutorial for Creating Policy Wheel Data Visualizations,” with Joshua Eagan, Beth Ann Griffin, Max Griswold, Pedro Nascimento de Lima, Seema Choksy Pessar, Rosalie Liccardo Pacula, Bradley D. Stein. RAND Corporation, 2025.
National Guard Youth ChalleNGe Program
“Developing Outcome Measures for the National Guard Youth ChalleNGe Program,” with Jennie W. Wenger, Stephani L. Wrabel, Thomas E. Trail, Louay Constant, Wing Yi Chan, Kathryn A. Edwards, Joy S. Moini. RAND Corporation, 2022.