Innovative Sampling Methods
In many countries, alternative sampling approaches are needed which do not depend on up-to-date census data or interviewer involvement.
I have been involved in two field experiments to develop and test new methods:
Ethiopia: As a World Bank consultant, I helped test Random Geographic Cluster Sampling. This method involves selecting random points and interviewing all persons within a given radius of a points. The survey successfully found and interview Pastoralists in the Afar region. Our experiences were [published] (https://content.sciendo.com/view/journals/jos/30/2/article-p191.xml) in the special issue of the Journal of Official Statistics on hard to reach populations.
India: RTI conducted an experiment in Uttar Pradesh to compare random walk (a commonly-used but non-probability method of household sampling) with our alternative, called geosampling. We will compare the response rate, cost, bias and variance of the two approaches. Results will be ready in 2020.
In addition, I have published several papers and book chapters on related topics.