Combining Econometric Methods and Qualitative Inquiry to Uncover the Impact of a School Attendance Program in Brazil

Abstract Information: This presentation describes the impact evaluation process of Busca Ativa Escolar, Brazil's largest program aimed at combating school dropout rates, which is sponsored by Unicef. The program employs a methodology for monitoring school absences and mobilizes public social services to actively address the causes of each dropout case. The program currently reaches approximately 2,500 municipalities, covering half of all municipalities in Brazil. The impact evaluation utilized mixed methods, including an impact evaluation that utilized staggered differences-in-differences methodology, using a public microdata panel, in conjunction with a survey of 1,000 municipalities. Additionally, 130 qualitative interviews were conducted with implementers in 30 participating municipalities. The evaluation process not only demonstrated a positive impact, with 62,000 children no longer dropping out of school due to the program, but also provided detailed insight into the implementation mechanisms that contributed to the program's success.

Relevance Statement: Considering the context of the "The Power of Story" conference, we believe that our case study contributes to demonstrating how a set of individual stories, from very different contexts, can enrich the conclusions of an evaluation based on econometric methods. The evaluated program is distributed throughout all regions of Brazil, including the most remote ones, and thanks to the availability of public microdata on education (with a long historical series that covers all public schools in the country), it was possible to demonstrate, using robust econometric models, that the program contributes to reducing school dropout rates. However, like all statistical methods, it is not capable of revealing the reasons why. In light of this, listening to over 130 public agents (from education and public social services) responsible for the program was essential to give meaning to the quantitative results. With each story told, the evaluators were able to understand the program's operation in practice, revealing its strengths and weaknesses. At the end of the qualitative process, it was possible to return to the impact evaluation results and carry out a second stage of data exploration, in which we could replicate the statistical models in different contexts, seeking to reproduce the qualitative conclusions in quantitative terms. This process not only allowed the econometric evaluation to reveal more conclusions but also provided a better explanation for the reasons behind each quantitative result.