Estimating uncertainty in respondent-driven sampling using a tree bootstrap method
Abstract
Some hidden or hard-to-reach populations of interest to researchers are difficult to study with standard statistical methods because there is not a reliable list of members from which samples can be drawn. Respondent-driven sampling (RDS) is a common way to reach members of these populations by allowing a small number of respondents to recruit further respondents in the target population from their personal contacts. However, estimates derived from RDS are known to have a high degree of uncertainty, for which current methods do not fully account. We present a method that overcomes this problem and allows for better statistical inference from RDS data.
Document Details
- Document Type
- Pub Defense Publication
- Publication Date
- Dec 07, 2016
- Source ID
- 10.1073/pnas.1617258113
Entities
People
- Aaron J. Baraff
- Adrian Raftery
- Tyler H. McCormick
Organizations
- Army Research Office
- National Center for Medical Rehabilitation Research
- National Heart, Lung, and Blood Institute
- University of Washington