A Gravity Model for Optimizing Random Samples for Populations with Low Incidence
Kai-Uwe Schnapp and Christian Hoops, Ipsos Public Affairs and Adrian-Schaefer-Rolffs , in Proceedings of the 67th World Association for Public Opinion Research Conference Nice, 09/2014
Over the past few years, interest in national, ethnic, cultural and other minorities has been increasing not only within politics, but also within the field of social research. However, survey-based studies are notoriously difficult to conduct among minority populations.
In this paper we present a procedure for sampling low-incidence populations based on gravitational analysis. The gravitational model applied uses the distances between people’s place of residence and a relevant reference location, as well as other attractiveness-related features, to determine the sample size per locality. This procedure is particularly cost-saving in its method of collecting data from low-incidence populations, while at the same time avoiding too much sampling bias.