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?