Identifying socioeconomic distinctions in the built environment: a case study using dwelling and occupant characteristics recorded in Ghana’s Demographic and Health Survey of 2014

Keywords: Demographic and Health Survey, population geography, population synthesis, socioeconomic status
Abstract Type: Paper Abstract

Abstract

Material differences in resource use, mobility and population density within socio-ecological systems often play out along the lines of socioeconomic class. Understanding geographic variation in socioeconomic class can provide insights into the distributional consequences of planning and policy decisions in low and middle-income countries (LMICs). However, describing and locating class distinctions remains a challenge in such settings where traditional metrics like household income are not directly enumerated for the whole population. In this paper, we develop an ontology of socioeconomic distinction from geolocated data drawn from Ghana’s 2014 Demographic and Health Survey and map this breakdown across the country. We first quantify relationships among occupant and dwelling characteristics within a mid-sized city, including employment, asset ownership and building materials, looking for latent ‘class’ subgroups. We scale these variables relative to DHS wealth quintiles and run a clustering algorithm to segment ‘lower’ and ‘upper’ portions of the population. We then fit a model with geospatial covariates to predict where these occupant-dwelling typologies are likely to occur and build a synthetic population attributed by these categories for the country. We validate our modelled distribution of both building and demographic characteristics against administrative-level census totals. We conclude with ideas for future work including leveraging occupant-dwelling relationships to more precisely place synthetic socioeconomic subpopulations at the scale of neighborhoods and buildings.