Comparing methods for measuring park access and equity using US census microdata in metropolitan Miami

Knowledge from prior park equity studies has been limited by differences in the spatial analytic techniques applied to measure park accessibility, a lack of systematic methods comparisons, and overreliance on aggregated spatial data. Our study addresses these issues using individual-level, restricted-access US Census American Community Survey microdata ( n = 339,000) from 2015 to 2019 to compare (a) alternative techniques for measuring access to park space and (b) how those measurement techniques affect inferences regarding patterns of inequity. It applies spatial analytic techniques to generate four park area per capita measures—one gold standard measure at the census block level and three widely-used measures at the census tract level in metropolitan Miami, USA—and compares the four methods using multiple analyses. The four methods generate somewhat dissimilar park accessibility values for individual residents. Bivariate analyses indicate that racially/ethnically minoritized or lower socioeconomic status people have less park accessibility than people from more advantaged groups, no matter the measure used. Multivariable models for each park accessibility measure yield largely dissimilar results regarding associations with social indicators, with consistent results for only two social indicators: Hispanic/Latino and older (>64 years) people have less park area per capita relative to non-Hispanic White and middle-aged (18–64 years) people, respectively. Findings suggest that advanced, fine-scale spatial analytic techniques should be applied to accurately characterize park accessibility for academic research and decision-making about parks. Findings also highlight the need to increase park space access for marginalized groups in metropolitan Miami. • Compares park access measurement methods and implications for assessing equity. • Uses fine-scale, individual-level US Census microdata to improve on prior research. • Four methods produce weakly correlated and somewhat dissimilar park access values. • Four methods support mostly dissimilar equity inferences from multivariable models. • More accurate, finer-scale measurement methods are needed for research and practice.