Where Are Australia's Churches, and Who Lives Near Them?

Australia has about 11,000 local churches: Anglican, Catholic, Uniting, Baptist, Orthodox and many more. Where are they, and how do they line up with where people of different ancestry live today? We took every Christian place of worship that volunteers have mapped on OpenStreetMap (7,383 after removing duplicates) and the 2021 Census for 2,341 neighbourhoods (SA2 areas, 25.4 million residents), and asked the question with spatial statistics. The first two and a half minutes need no code. Then we open Python and R: one Overpass query and two ABS downloads; labelling churches by denomination (and why "St Mary's" is not a denomination); choosing the neighbour network with a spatial lag model; Moran's I, bivariate Moran and local clusters (LISA); geographically weighted regression (GWR); a self-organising map (SOM); maps with sf and ggplot2; and the checks. What the data say: • Both cluster: neighbourhoods of similar ancestry sit side by side strongly (Moran's I 0.38–0.66); churches of one kind cluster much more weakly (0.03–0.20). • Believers and buildings tell different stories. Where Orthodox believers live follows Middle Eastern ancestry closely (Spearman 0.78), Anglican believers follow European ancestry (0.48); church buildings per resident follow ancestry only weakly (at most 0.35). • Outside the capital cities there are about twice as many mapped churches per resident (4.5 vs 2.1 per 10,000); the fewest are in big-city suburbs with many Indian, Chinese and Middle Eastern residents. • The link changes from place to place: a separate regression for each area explains much more (R² Anglican 0.25 → 0.39, Catholic 0.20 → 0.38); for Baptist and Orthodox churches, ancestry explains almost nothing. • Local clusters: 116 areas, mostly in Melbourne's north and Sydney's south-west, where Middle Eastern ancestry sits next to many Orthodox churches. • 2,008 church names start with "St" — mostly Anglican (790) or Catholic (770): Saint is a name, not a denomination. • One pre-set check failed: we expected believers to follow ancestry more than buildings in 4 of 5 pairs; it was 16 of 25. It holds for Anglican, Uniting and Orthodox, not for Catholic and Baptist, and we say only that. Scope: OpenStreetMap covers about two thirds of local churches (NCLS estimate for 2016: about 11,400), Pentecostal churches only about one in ten (left out), and coverage differs by state. The Census asks about ancestry, not race; "Middle Eastern" is a combined definition (Lebanese ancestry + born in Iraq, Iran, Turkey or Egypt). These are patterns between areas, not between people, and associations, not causes. Educational example of spatial analysis.