Geospatial analysis of tobacco outlet density in New Jersey: A weighted clustering approach

A preponderance of literature suggests that higher proportion of retail tobacco distributors often found in densely populated areas, increases access to these products, thereby augmenting tobacco-related consumption. Even when controlled for in geospatial weighted regression, still the locations targeted for placement of retail tobacco stores appear to be in census tracts with higher concentrations of minorities particularly Latino and Black origin, coupled with lower income. Although previous studies have demonstrated a relationship between higher percentage of ethnic variation, lower educational attainment and lesser wages in neighborhoods where more stores sell tobacco, in conjunction with higher smoking rates, the higher concentration may merely be an artifact of higher population density in a region. Therefore, in order to test whether larger numbers of NJ tobacco retail stores (n = 11,848) are found in ethnically diverse areas as a function of increased population, Voronoi diagrams as a geospatial analysis technique are applied. Using Geographic Information Systems (ArcGIS) software and Matlab to generate the clusters, Voronoi polygons with a weighted clustering design allow for comparison of groups based on geospatial location relative to a center point. Relationship of tobacco retails stores and percent minority, high school education or less and income between Voronoi-designated regions now distributed relative to equal population is assessed using weighted regression analysis methods. Results show that tobacco outlets are no more prevalent in highly populated areas with higher percent of minorities. Therefore, limiting the number of outlets on a health planning level may not curtail smoking or reduce tobacco use.