Data Note: Challenges when combining housing data from multiple sources to identify overcrowded households

• Published in: International journal of population data science Vol. 8; no. 5
• Main Authors: Scott, Laura, Weigang, Yan, Ucci, Marcella, Sheringham, Jessica
• Format: Journal Article
• Language: English
• Published: Swansea University 20.05.2025
• Subjects: administrative data; data linkage; housing; overcrowding; public health
• ISSN: 2399-4908; 2399-4908
• DOI: 10.23889/ijpds.v8i5.2927

BackgroundThis project in one urban local authority in London (England) sought to assess the feasibility of generating locally-derived indices of overcrowding using data available to local councils on the population and their homes. We merged data at household level using the Unique Property Reference Number from publicly available Energy Performance Certificates and commercial property platforms, with data available to councils on the population and their housing characteristics, drawn from multiple sources including council tax bands and council housing databases. Multiple imputation was used to address missing data. Using the dataset, it was possible to generate two indices of overcrowding for households with dependent children, based on the bedroom standard and the space standard, which could be compared with nationally derived estimates. Data challengesWe encountered three challenges with data. 1. Individuals in the population were excluded through linkage with household-level data. 2. Definitions of overcrowding are ambiguous and variably applied. 3. Many local areas face high proportions of missing household data, particularly numbers of bedrooms. We discuss how we addressed such problems and illustrate with a local example how they could affect estimates of overcrowding prevalence. Lessons LearnedFurther clarity is needed in how bedrooms are defined to compare overcrowding prevalence generated locally and nationally. Access to national records on bedroom numbers would facilitate local areas to identify overcrowding in their own populations. Despite these challenges, we demonstrate it is feasible to generate overcrowding indices that can be useful for researchers and local policy makers seeking to develop or evaluate strategies to address household overcrowding.