|
Abstract:
|
Throughout his career, Graham Kalton has contributed in a major way toward how we, as survey statisticians, think about and apply probability sampling and estimation methods to studies of rare and hard to survey populations. This presentation will seek to honor those contributions through a brief review of the general approaches and a more in-depth look at recent developments in several approaches to probability sampling of populations with rare or hard to detect attributes.
The talk will open with a general review of stratification, disproportionate allocation, and multiple frames as tools to improve sample coverage and screening efficiency followed by a more in-depth discussion of multi-phase designs that utilize two or more screening phases to identify population elements that are potentially eligible for the study. Indirect sampling methods will be reviewed next including time and location sampling and adaptive cluster sampling. The talk will conclude with coverage of methods that exploit social networks to identify samples of hard to survey populations. In this last segment, the emphasis will be on recent developments in multiplicity and respondent driven sampling.
|
