Why do structurally similar regions vote so differently?
DemoAtlas sets the voting behaviour and social structure of Germany's 400 districts side by side, and reveals a pattern that structure alone cannot explain: East and West German districts with very similar indicators (unemployment, income, education) often vote in opposite directions.
The atlas traces this East-West difference back to its roots: the Treuhand shock of the early 1990s (unemployment in the East rising from around 10 to over 15 % within two years), the persistent under-representation of East Germans in top positions, and a declining trust in the media. It makes no claim of causality. It assembles the data so that you can examine the connections for yourself.
Election data
- GERDA, German Regional Election Database
- Party-list (second) votes per district, federal elections 1990–2025 and state (Landtag) elections by federal state (1990–2026).
- Repräsentative Wahlstatistik der Bundeswahlleiterin
- Voting behaviour by age group (federal elections 2021 and 2025); officially counted ballots in sample precincts, not a survey.
- Die Bundeswahlleiterin
- Structural data on the 299 constituencies (including education) as the basis for the structural types.
- Wahlnacht-Erhebung (Wählerwanderung)
- Net vote transfer between parties 2021→2025 (federal level).
Structural & regional data (district level)
- INKAR (BBSR)
- GDP per capita, share of foreign nationals, population density, school-leavers (without a qualification / with the Abitur), average age, share aged 65+, and workers by economic sector, as boundary-adjusted time series.
- Statistik der Bundesagentur für Arbeit
- Unemployment rate: districts (from 2001), West/East and federal states (from 1991).
- Regionalstatistik / GENESIS (Destatis)
- Disposable income and further structural indicators.
- Bundesamt für Kartographie und Geodäsie (BKG)
- District boundaries for the maps, © GeoBasis-DE / BKG (via opendatasoft georef-germany-kreis), dl-de/by-2-0.
East-West time series (aggregated)
- VGR der Länder (Arbeitskreis VGRdL)
- Disposable income of private households per capita and population trends, East/West (as calculated in 2024).
- Bundesinstitut für Bevölkerungsforschung (BiB)
- Relocations across the East-West border (based on the Federal Statistical Office).
Societal context (national)
- Deutscher Elitenmonitor (Uni Leipzig / Jena / HS Zittau-Görlitz)
- Share of East Germans in top positions by sector.
- Mainzer Langzeitstudie Medienvertrauen · Reuters Digital News Report
- Trust in the media / news, Germany.
Spatial level. Maps and structural data are held at the 400 current districts (present-day boundaries). Time series are boundary-adjusted, recalculated onto consistent borders across district reforms (including Saxony in 2008) using INKAR, so that comparison across the years holds.
Structural types. The clusters on the structural-types page group constituencies with a similar socio-economic profile, on z-standardised indicators, so that figures measured in different units become comparable.
Map colours. The election map uses the real party colours (a lighter shade → full colour by vote share); the structure map uses sequential scales. The time slider ties both maps to a shared year.
Definitions & breaks. The unemployment rate for the reunification years uses the definition “dependent civilian labour force” (continuous from 1991); the district series from 2001 uses the more modern “all civilian labour force”. Visible jumps mark real changes in method, Hartz IV in 2005 (a broadened definition of unemployment) and the census revisions of 2011 and 2022 (population, share of foreign nationals, density). They are not genuine trends.
- Correlation is not causation. Patterns that sit side by side prove no causal link.
- Some data is national only. Trust in media, the elite monitor and voting behaviour by age are available only nationwide or by sector, not per district. Inferring from the federal level down to individual districts risks an ecological fallacy.
- Reporting periods differ. Not every indicator reaches equally far back or is equally up to date; the relevant year is shown next to each value.
- GDP is nominal. Economic output at current prices, not adjusted for inflation.
Questions, data tips or found an error? Write to info@ck71.de.
See also Legal notice and Privacy policy.