Vol.:(0123456789) Natural Hazards https://doi.org/10.1007/s11069-021-04938-9 1 3 ORIGINAL PAPER The constituent components and local indicator variables of social vulnerability index Gainbi Park 1  · Zengwang Xu 1 Received: 24 March 2021 / Accepted: 10 July 2021 © The Author(s), under exclusive licence to Springer Nature B.V. 2021 Abstract Social vulnerability index (SoVI) has been widely used to measure the extent to which people or places are socially vulnerable. The SoVI is an aggregate composite index that linearly combines a few principal components resulted from the principal components analysis on a number of selected social vulnerability indicator variables, and it can quantify the relative level of overall social vulnerability but cannot inform the specifc local social indicators that contribute to the vulnerability in various degrees. The specifc social indica‑ tors that either attenuate or amplify local social vulnerability are of much need in policy making to reduce social vulnerability. This study explores the diferential contributions of the constituent components of SoVI and investigates how the local indicator variables have evolved over time and across the Greater Houston metropolitan area in the USA using the geographically weighted principal components analysis. It found that the overall social vul‑ nerability as measured by SoVI has exhibited persistent spatial patterns in the Greater Hou‑ ston area since 1970; however, the spatial patterns of the SoVI are not equally constituted by the components of the SoVI. In particular, the high social vulnerability of suburban areas is mainly the result of one principal component that highly correlates with the per‑ centage of mobile homes. It also found that the indicator variables of social vulnerability have exhibited great spatial heterogeneity and dependence at local scale, and they vary over time but persist on disadvantages in economic condition, mobility, and family structure. Keywords Social vulnerability · Principal components analysis · Geographically weighted principal components analysis · Spatial heterogeneity · Houston 1 Introduction Social vulnerability has been widely adopted in identifying the vulnerable people and places for disaster preparedness, response, recovery, and mitigation. However, social vul‑ nerability is not directly measurable, as it is latent in a broad range of social conditions that in various ways situate people or places in vulnerable conditions. Developing reliable and * Zengwang Xu xuz@uwm.edu 1 Department of Geography, University of Wisconsin Milwaukee, Bolton Hall Room 410, Milwaukee, WI 53201‑0413, USA