Assisting total estimation in spatial populations

In the last decade, we witnessed a reappraisal of design-based tech- niques for inference on synthetic values (i.e., mean or total) for spa- tially distributed populations. Some authors use splines to incorpo- rate the geographical information in estimation, while others treat the sampling design as a covariate. Our proposal is able to exploit the geographical relationships between the locations in the domain to en- rich estimation. Weights are dened as standardized functions of the Euclidean distances. Some statistical properties of the estimator are derived. Results of a Monte Carlo simulation are presented in order to better appreciate the performances of our proposal when compared to other alternatives not able to exploit the spatial information of the data.