This study addresses an important methodological issue in the study of the relationship between neighborhood characteristics and obesity. Given the complexity of the built environment, it is difficult to account for the many, and often interacting, variables in the environment that might influence obesity. This study attempts to disentangle the myriad of environmental variables that might contribute to obesity in adolescents by comparing three unique approaches to analyzing the data. The authors examined the relationships between adolescent weight status and 22 neighborhood characteristics (including parks/recreation spaces, as well as, for example, perceived safety and access to food sources) in 2682 diverse adolescents in the Minneapolis/St. Paul, MN area. 53% were girls, the average age was 14.5 years, and all were part of a population-based study on eating and physical activity. Height and weight measures were assessed during one academic year and GIS data were examined to characterize neighborhood environments. The statistical techniques used in the comparison included: 1) regression analysis on individual neighborhood characteristics; 2) exploratory factor analysis to determine whether the individual neighborhood characteristics could be combined into a smaller set of factors associated with adolescent weight; and (3) latent class analysis to identify clusters of neighborhood characteristics with the potential to influence obesity and test their associations with adolescent weight status.
Comparison of the statistical techniques demonstrated that: Although regression analysis was able to identify specific variables related to obesity status, it could not facilitate interpretation of a total environment effect or the interaction of multiple variables in the environment. Factor analysis was able to assist in the assessment of the relative importance of environmental variables that were related to each other. Spatial latent class analysis was able to demonstrate the importance of considering socioeconomic status combined with neighborhood characteristics in predicting obesity. Overall, the three analyses concluded that convenient access to unhealthy foods and decreased access to parks/recreation areas predicted higher rates of obesity.
Research Partner