A1252
Title: Achieving global and within-group spatial spreading in survey sampling
Authors: Maria Michela Dickson - University of Padova (Italy) [presenting]
Yves Tille - University of Neuchatel (Switzerland)
Flavio Santi - University of Trento (Italy)
Diego Giuliani - University of Trento (Italy)
Giuseppe Espa - University of Trento (Italy)
Abstract: In survey sampling, selecting a well-spread sample is essential for producing precise estimates when the variable of interest is spatially autocorrelated. In many practical applications, populations are naturally partitioned into meaningful domains, such as economic sectors, administrative regions, or activity groups, and survey objectives require ensuring spatial balance both across the whole population and within each subgroup. For instance, in business surveys it is often important to obtain a geographically well-spread sample of firms overall, while also guaranteeing adequate coverage within sectors such as manufacturing, services, or retail. While several methods exist for achieving global spatial balance, no existing method directly addresses the case where spatial balance is required simultaneously at both the global and within-group levels. To address this limitation, a Multispread sampling design is proposed, which simultaneously achieves this dual objective by extending the pivotal method to triplets of units and interpreting it as a special case of the cube method. The proposed procedure is governed by a tuning parameter between 0 and 1, which controls the trade-off between global and within-group spatial balance. Intermediate values allow flexible balancing of both objectives depending on survey requirements. The method is evaluated by assessing spatial balance and estimation efficiency.
