Skip directly to content

Minimize RSR Award Detail

Research Spending & Results

Award Detail

Doing Business As Name:University of Washington
  • Corinne Mar
  • (206) 543-4043
Award Date:08/24/2009
Estimated Total Award Amount: $ 269,318
Funds Obligated to Date: $ 269,318
  • FY 2009=$269,318
Start Date:09/01/2009
End Date:08/31/2012
Transaction Type:Grant
Awarding Agency Code:4900
Funding Agency Code:4900
CFDA Number:47.075
Primary Program Source:040101 RRA RECOVERY ACT
Award Title or Description:Statistical Methods for Respondent Driven Sampling Data
Federal Award ID Number:0851555
DUNS ID:605799469
Parent DUNS ID:042803536
Program:Methodology, Measuremt & Stats
Program Officer:
  • Cheryl Eavey
  • (703) 292-7269

Awardee Location

Street:4333 Brooklyn Ave NE
Awardee Cong. District:07

Primary Place of Performance

Organization Name:University of Washington
Street:4333 Brooklyn Ave NE
Cong. District:07

Abstract at Time of Award

This award is funded under the American Recovery and Reinvestment Act of 2009 (Public Law 111-5). This project will develop probability models and inferential methods for the analysis of data collected using Respondent Driven Sampling (RDS). RDS is an innovative sampling technique for studying hidden and hard-to-reach populations for which no sampling frame can be obtained. RDS has been widely used to sample populations at high risk of HIV infection and has also been used to survey undocumented workers and migrants. RDS solves the problem of sampling from hidden populations by replacing independent random sampling from a sampling frame by a referral chain of dependent observations: starting with a small group of seed respondents chosen by the researcher, the study participants themselves recruit additional survey respondents by referring their friends into the study. As an alternative to frame-based sampling, the chain-referral approach employed by RDS can be extremely successful as a means of recruiting respondents. Yet despite the promise of being able to recruit study participants from communities that are otherwise intractable to study, there are very few statistically informed methods available to analyze the data. In particular, extant methods for estimating population means from RDS data rely upon qualitative assumptions about unobserved aspects of the underlying social network connecting survey respondents; and the standard practice for regression analysis of RDS data is to either ignore dependency between observations or to limit analysis to variables where dependency is assumed to be absent. Thus this project has two key aims: 1) replace the qualitative network assumptions with a well-defined probability model for the underlying social network and 2) develop regression models that explicitly incorporate the dependencies between observations. In addition to carrying out the basic science required to fulfill these aims this proposal will create freely available software that will make these new analytical techniques available to field researchers using RDS.

For specific questions or comments about this information including the NSF Project Outcomes Report, contact us.