Open Conference Systems, 50th Scientific meeting of the Italian Statistical Society

Font Size: 
Post-sampling crowdsourced data for reliable statistical inference: the case of market price index in Nigeria
Giuseppe Arbia

Last modified: 2018-05-31


Pure crowdsourcing is in contrast with formal sample design. Indeed, in the latter, the choice of units is suggested by a precise mechanism which allows the calculation of the probability of inclusion while crowdsourced data units are self-selected and no probabilities of inclusion can be calculated. In our method, by exploiting the location of the collectors, the map of the crowdsourced data-points is compared with the map of points selected according to a formal spatial design with equivalent sample size. The observations are then reweighted so as to resemble the optimal plan. In this paper, we considered, in particular, the Local Pivotal Method 2 (Grafström, 2012) and we report some results of application of the suggested method to price data recently crowdsourced by FAO in 16 local markets of Kaduna State (Nigeria).



Grafström A, Lundström NL, Schelin L. (2012) Spatially balanced sampling through the pivotal method, Biometrics, 68(2):514-20.

Full Text: PDF