Spatial econometric models become increasingly popular in various subfields of political science. However, the necessity to specify the underlying network of dependencies, denoted W, prior to estimation is a prevalent source of criticism since the true dependence structure is rarely known and theories provide only insufficient guidance. The present study advocates Bayesian model averaging (BMA) as a superior approach to this problem, located at the intersection of theory and empirics, as it conditions inferences on a set of feasible networks. Unlike model selection procedures, BMA directly accounts for network uncertainy, which is a special case of model uncertainty that arises from uncertainty about the specification of W, in a coherent framework. Three replication studies compare BMA to alternative techniques and demonstrate the benefits of this approach for practical research. In contrast to the commonly held belief, the results show that, while model uncertainty is a valid concern, network uncertainty rarely changes the substantive conclusions derived from spatial econometric models. Instead of solely focusing on the specification of W, researchers should reflect more thoroughly on temporal dynamics, common shocks, and unit heterogeneity since these specification issues can distort the estimates of the spatial effect.