Maria Dickson

Department of Economics and Management, University of Trento Via Inama 5, 38122, Trento, Italy


The use of spatial sampling methods in presence of locational errors

In recent times, business registers have been provided with geographical information about units, making possible to conduct spatial studies on firms. Sometimes data result perturbed respect to their real geographical location owing to privacy problems and technical problems in phase of geo-coding of units. In both cases, they constitute non-sampling errors particularly relevant in different fields of research, such as, among others, health (Boulos, 2004; Rushton et al., 2006), political science (Haspel and Knotts, 2005) and economy. About privacy problems, sometimes is not possible disseminate information about geographic location of units, due to the nature of these sensitive data (VanWey et al., 2005). The second kind of errors, suffered rather than induced, has a notable impact on business registers. In particular, causing by the specific geo-coding process (Cozzi and Filipponi, 2012), it could happen that some units are not locate with precision on the territory or they are allocated to the centroids of the areas which include them (Jacquez and Rommel, 2009; Jacquez, 2012).
In the present work, the focus is on incomplete and erroneous geo-coding of business registers, and on consequences of it in spatial sampling of firms. By means of a simulation study, the aim is to investigate the effect of locational errors in the selection of units for different spatial sampling designs (Grafström et al., 2011; Grafström, 2012; Grafström and Tillé, 2013). The population used has been deliberately perturbed based on the known percentage of geo-coding errors in business registers. Samples of different sizes will be drawn and the impact of locational errors on the efficiency of sampling designs will be evaluated comparing the estimation of population totals between and after the perturbation.