Dataset for the subnational seats of G20 countries subnational units

Assembly size has been a central question in the study and quest for proper political representation, going back thousands of years, at least back to Ancient Greece with discussions pertaining to the number of citizens per representative. Since then, political systems have evolved and formed rules to elect or appoint representatives. More recently, at least since Penrose (1946) and Taagepera (1972), social sciences have tried to understand the link between population growth and assembly size. At the national/federal levels, this subject is relatively adequately studied with hundreds of papers. Surprisingly enough, almost no research has tried to describe this relationship at the subnational level. A few have studied the political effects of assembly size at the micro level (municipal and district). This is probably due to the difficulty of gathering thousands of data points for subnational territories. This data paper aims at filling this gap, thus allowing for further systematic study of the link between assembly size, population size, population density, and number of subnational units. To this end, data for all 19 sovereign countries of the G20 group were collected. This sample provides a compromise between data availability, political and economic diversity and world population coverage. Collection was done by cross-comparing national/state election reports with each subnational unit's website. At total of 805 subnational units were identified, with a total of 69,110 subnational seats gathered, as of December 2025, constituting the main innovation of the present paper. The dataset has two forms, either an individual country perspective with 19 files or a single general file, compiling all data, sorted from smallest to biggest subnational unit population, thus pooling all countries. Australia, France, U.S.A., P.R. China and the United Kingdom have at least two sheets per file to differentiate between mainland only perspective or with overseas and special territories. For the general file compiling all data, there are also two sheets, one with island and special territories and the other with mainland only. This constitutes a minor innovation in the literature since very small and overseas territories (with <500 000 people) are never included in existing studies, thus potentially overlooking a different dynamic for the scaling relationship between population size and legislature size. Researchers are encouraged to study the included country's subnational units, in part or as a whole, as well as adding their own observations, as the dataset forms a basis for future expansion. One should note that as election results, and population size change quickly, one should understand the present paper as a late-2025 snapshot and look for an updated version in the future.