We found a match
Your institution may have access to this item. Find your institution then sign in to continue.
- Title
Uncovering a positive and negative spatial autocorrelation mixture pattern: a spatial analysis of breast cancer incidences in Broward County, Florida, 2000–2010.
- Authors
Hu, Lan; Chun, Yongwan; Griffith, Daniel A.
- Abstract
Spatial cancer data analyses frequently utilize regression techniques to investigate associations between cancer incidences and potential covariates. Model specification, a process of formulating an appropriate model, is a well-recognized task in the literature. It involves a distributional assumption for a dependent variable, a proper set of predictor variables (i.e., covariates), and a functional form of the model, among other things. For example, one of the assumptions of a conventional statistical model is independence of model residuals, an assumption that can be easily violated when spatial autocorrelation is present in observations. A failure to account for spatial structure can result in unreliable estimation results. Furthermore, the difficulty of describing georeferenced data may increase with the presence of a positive and negative spatial autocorrelation mixture, because most current model specifications cannot successfully explain a mixture of spatial processes with a single spatial autocorrelation parameter. Particularly, properly accounting for a spatial autocorrelation mixture is challenging. This paper empirically investigates and uncovers a possible spatial autocorrelation mixture pattern in breast cancer incidences in Broward County, Florida, during 2000–2010, employing different model specifications. The analysis results show that Moran eigenvector spatial filtering provides a flexible method to examine such a mixture.
- Subjects
BROWARD County (Fla.); GEOGRAPHIC spatial analysis; BREAST cancer; AUTOCORRELATION (Statistics); SPATIAL filters; MIXTURES; STATISTICAL models
- Publication
Journal of Geographical Systems, 2020, Vol 22, Issue 3, p291
- ISSN
1435-5930
- Publication type
Article
- DOI
10.1007/s10109-020-00323-5