Greenspace is increasingly examined as a low-cost way to increase standardized test scores in public schools. However, the evidence for this intervention is mixed. One potential explanation is the variety of ways that greenspace is measured using remotely sensed data. For instance, aggregate measures can be captured from tree, grass, and shrub cover classifications in high-resolution (1 m2) land cover datasets or they can be measured with normalized difference vegetative index (NDVI) values from sensors at different resolutions (e.g., 30 m2 or 250 m2). In the current cross-sectional observational study, we tested the relationship between five greenspace measures and third-grade math and reading standardized tests scores in Maryland public schools (n = 668) around schools and in children's neighborhoods. Low- and high-resolution greenspace measures were highly correlated with each other, but moderate-resolution measures were not. Multivariate regression models revealed positive associations between academic performance and low-resolution NDVI measures around schools and in neighborhoods as well as between performance and tree cover in neighborhoods. These effects were attenuated when an understudied confounder in this body of literature was included: population density as a measure of urbanicity. Grass cover showed associations with performance in models adjusted for urbanicity, but the direction of these associations was negative. These findings suggest that the possible association between greenspace and academic performance is complex and tenuous when examined with observational, cross-sectional study designs in limited geographic regions.
Read the article in Landscape and Urban Planning.