Early View article: “A Multi-Scale Analysis of Single-Family Residential Water Use in the Phoenix Metropolitan Area,” by Yun Ouyang, Elizabeth A. Wentz, Benjamin L. Ruddell, and Sharon L. Harlan.
Single-family residential water use research is often limited to using aggregated water data due to, for example, confidentiality restrictions on household-scale data. Water use data that are accessible to researchers may be aggregated to areal units, such as census blocks, census block groups, census tracts, and cities. To match the water use data, similarly aggregated data for the factors considered to influence water use should also be used. This may lead to an ecological fallacy problem, which can occur when the statistical analysis and conclusions based on aggregated data are not applicable at the individual scale. There is little, if any, empirical analysis that assesses whether spatial scale may cause an ecological fallacy problem in residential water use research. The goal of this study is to address this issue by using the Phoenix area, Arizona as a case study.
Three panel datasets with different spatial scales are used in this study: household scale, census tract scale, and city/town scale. The authors found that the census tract scale data produce similar results compared to the household-scale data when they use the econometric models to study the relationship of single-family residential water use and its determinants in Phoenix, Arizona. No significant ecological fallacy problem was identified by this comparative statistical analysis that is based on the signs, magnitude, and confidence intervals of the parameter estimates.
[Please note: I have quoted and paraphrased freely from the article, but the interpretation is my own.]