Map Intersection Based Merging Schemes for Administrative Data Sources and an Application to Germany

In many situations the applied researcher wants to combine different data sources without knowing the exact link and merging rule. This paper considers different interpolation methods for interpolating attributes from German labor office districts to German counties and vice versa. In particular, we apply dasymetric weighting as an alternative to simple area weighting both of which are based on estimated intersection areas. Since these estimates can be spurious, our theoretical framework extends the well-known Goodchild and Lam (1980) approach to the presence of measurement error in the underlying maps. We also present conditions under which the choice of interpolation method does not matter and confirm the theoretical results with a simulation study. Our application to German administrative data suggests robustness of estimation results of interpolated attributes with respect to the choice of interpolation method. We deliver weighting matrices for regional data sources of the two largest German data producers