Policies and interventions to remove gender‐related barriers to girls' school participation and learning in low‐ and middle‐income countries: A systematic review of the evidence

Background Gender disparities in education continue to undermine girls' opportunities, despite enormous strides in recent years to improve primary enrolment and attainment for girls in low‐ and middle‐income countries (LMICs). At the regional, country and subnational levels gender gaps remain, with girls in many settings less likely to complete primary school, less likely to complete secondary, and often less likely to be literate than boys. The academic and policy literatures on the topic of gender‐related barriers to girls' education are both extensive. However, there remain gaps in knowledge regarding which interventions are most likely to work in contexts with different combinations of barriers. Objectives This systematic review identified and assessed the strength of the evidence of interventions and exposures addressing gender‐related barriers to schooling for girls in LMICs. Search Methods The AEA RCT Registry, Africa Bibliography, African Education Research Database, African Journals Online, DEC USAID, Dissertation Abstracts, EconLit, ELDIS, Evidence Hub, Global Index Medicus, IDEAS‐Repec, Intl Clinical Trials Registry, NBER, OpenGrey, Open Knowledge Repository, POPLINE, PsychINFO, PubMed, Research for Development Outputs, ScienceDirect, Sociological Abstracts, Web of Science, as well as relevant organization websites were searched electronically in March and April of 2019. Further searches were conducted through review of bibliographies as well as through inquiries to authors of included studies, relevant researchers and relevant organizations, and completed in March 2020. Selection Criteria We included randomized controlled trials as well as quasi‐experimental studies that used quantitative models that attempted to control for endogeneity. Manuscripts could be either published, peer‐reviewed articles or grey literature such as working papers, reports and dissertations. Studies must have been published on or after 2000, employed an intervention or exposure that attempted to address a gender‐related barrier to schooling, analyzed the effects of the intervention/exposure on at least one of our primary outcomes of interest, and utilized data from LMICs to be included. Data Collection and Analysis A team of reviewers was grouped into pairs to independently screen articles for relevance, extract data and assess risk of bias for each included study. A third reviewer assisted in resolving any disputes. Risk of bias was assessed either through the RoB 2 tool for experimental studies or the ROBINS‐I tool for quasi‐experimental studies. Due to the heterogeneity of study characteristics and reported outcome measures between studies, we applied the GRADE (Grading of Recommendation, Assessment, Development and Evaluation) approach adapted for situations where a meta‐analysis is not possible to synthesize the research. Results Interventions rated as effective exist for three gender‐related barriers: inability to afford tuition and fees, lack of adequate food, and insufficient academic support. Promising interventions exist for three gender‐related barriers: inadequate school access, inability to afford school materials, and lack of water and sanitation. More research is needed for the remaining 12 gender‐related barriers: lack of support for girls' education, child marriage and adolescent pregnancy, lack of information on returns to education/alternative roles for women, school‐related gender‐based violence (SRGBV), lack of safe spaces and social connections, inadequate sports programs for girls, inadequate health and childcare services, inadequate life skills, inadequate menstrual hygiene management (MHM), poor policy/legal environment, lack of teaching materials and supplies, and gender‐insensitive school environment. We find substantial gaps in the evidence. Several gender‐related barriers to girls' schooling are under‐examined. For nine of these barriers we found fewer than 10 relevant evaluations, and for five of the barriers—child marriage and adolescent pregnancy, SRGBV, inadequate sports programs for girls, inadequate health and childcare services, and inadequate MHM—we found fewer than five relevant evaluations; thus, more research is needed to understand the most effective interventions to address many of those barriers. Also, nearly half of programs evaluated in the included studies were multi‐component, and most evaluations were not designed to tease out the effects of individual components. As a result, even when interventions were effective overall, it is often difficult to identify how much, if any, of the impact is attributable to a given program component. The combination of components varies between studies, with few comparable interventions, further limiting our ability to identify packages of interventions that work well. Finally, the context‐specific nature of these barriers—whether a barrier exists in a setting and how it manifests and operates—means that a program that is effective in one setting may not be effective in another. Authors' Conclusions While some effective and promising approaches exist to address gender‐related barriers to education for girls, evidence gaps exist on more than half of our hypothesized gender‐related barriers to education, including lack of support for girls' education, SRGBV, lack of safe spaces and social connections, inadequate life skills, and inadequate MHM, among others. In some cases, despite numerous studies examining interventions addressing a specific barrier, studies either did not disaggregate results by sex, or they were not designed to isolate the effects of each intervention component. Differences in context and in implementation, such as the number of program components, curricula content, and duration of interventions, also make it difficult to compare interventions to one another. Finally, few studies looked at pathways between interventions and education outcomes, so the reasons for differences in outcomes largely remain unclear.