Assessing national vs subnational population heterogeneities in a global context

Quantitative demographic research is often conducted at the national level, largely due to data limitations and a lack of hypotheses about within-country variation. Yet national-level data can obscure important internal diversity, leading to potentially misleading conclusions about demographic patterns. Using the CORESIDENCE Database, we analyse subnational variation in household arrangements across 142 countries divided into 1788 subnational units. We study four key household variables: household size, number of children, number of spouses or partners and number of other members. Results reveal substantial internal heterogeneity, particularly in Africa and Asia, where subnational variation often rivals or exceeds national-level variation. In contrast, European countries show greater homogeneity, making national datamore representative. These findingsemphasise the importance of integrating subnational data into demographic research. Relying solely on nationallevel indicators risks masking meaningful demographic patterns and misrepresenting population dynamics. As traditional census data become increasingly scarce, fine-grained data are crucial for capturing the complexity of demographic variation within countries.