A design utility approach for preferentially sampled spatial data

Abstract Spatial preferential sampling occurs when the choice of locations depends stochastically on process interest. Ignoring this dependence leads to inaccurate inferences. Our framework models experimenter preferences jointly with spatial adjust for this. We dispense unrealistic assumption (required by existing methods) conditional independence defining a whole design distribution proportional utility function space designs. The proposed model likelihood is generally intractable. provide fitting techniques based noisy Markov chain Monte Carlo and demonstrate their usage data set spatially distributed ammonia concentrations.