Multidimensional measures of quality of life: A comparison of methods using U.S. county-level data
比较六种方法（包括空间与非空间主成分分析、贝叶斯因子分析）构建美国县级生活质量指数，发现方法选择影响城乡极端地区的排名，强调敏感性分析的重要性。
There are numerous measures of place-based quality of life, accompanied by ongoing debate over which factors should be included. Past studies have predominately relied on aspatial, subjectively weighted combinations of different place-based characteristics. Little attention has been given to the rigor and appropriateness of the methodologies used to develop these measures. In this study we conduct a sensitivity analysis of methodological choices in the construction of quality of life indices using U.S. county-level data. We evaluate six different methods for developing a composite index, comparing Geographically Weighted Principal Component Analysis (GWPCA), Spatial Principal Component Analysis (SPCA), and Spatial Bayesian Factor Analysis (SBFA) with their aspatial counterparts. Our evaluation is twofold: conceptually, by examining underlying assumptions of each method, and practically, by assessing the consistency of results. We compare U.S. county-level quality of life ranking across methods and levels of rurality. While quality of life measures are generally highly correlated across methods, we find that the choice of spatial method does affect the rankings of individual counties. Notably, the most urban and most rural counties exhibit the greatest sensitivity to methodological choice. These findings underscore the importance of sensitivity analysis, particularly in urban and rural contexts, to better understand and account for the complex and spatially varying factors shaping quality of life.