First, I am not sure we can talk about MAUP if units are nested. Results might vary across units as more fine-grained spatial measures provide more precise information about the social environment. Consider this classic on the neighborhood effect:
Comments
Log in with your Bluesky account to leave a comment
One of the replications by Lee et al. replacing NUTS3 data from the Colantone & Stanig import shock study with more highly aggregated NUTS2 level data (using our EU-NED data) might not be the best evidence for the relevance of MAUP. Maybe NUTS2 areas are just too heterogenous for effects to show?
The authors recommend to run robustness checks for alternative spatial units. We do something in our recent paper on local ethnic diversity and rally effects. Results hold on the postcode and municipality level. But what if latter would not hold? Is it MAUP or internal heterogeneity?
Great paper by Lee et al. and an important discussion. But maybe we need to differentiate between issues arising from the availability of similarly sized spatial units vs. choice of units nested at different levels of aggregation...?
Comments