Measuring Network Sizes in the Context of Respondent Driven Sampling: Evidence from Two Independent Surveys

Area of Science:

• Social Sciences
• Epidemiology
• Network Analysis

Background:

• Respondent-driven sampling (RDS) is a key method for surveying hidden populations.
• Network size, or
• degrees
• is crucial for RDS point estimators, adjusting for selection probabilities.
• Current measures of degrees may not accurately reflect recruitability.

Purpose of the Study:

• To investigate various degree measures for their accuracy in respondent-driven sampling.
• To assess if degree measures considering recruitability improve RDS inference.
• To examine the influence of recruitment priming and social relationship context on degree reporting.

Main Methods:

• Conducted two independent RDS surveys: one with people who inject drugs (PWID) and another with Korean immigrants.
• Compared standard degree measures with those incorporating recruitment priming and social relationship specifics.
• Analyzed degree reports' sensitivity to interview language (English vs. Korean).

Main Results:

• Standard degree measures showed more variability ('noise') compared to other methods.
• Subtle recruitment cues reduced reported degrees, indicating a mismatch with standard questions.
• Interview language significantly impacted degree reports and their ability to explain recruitment.
• Measures reflecting close social relationships improved RDS inference, unlike standard degrees.

Conclusions:

• Network measures in RDS should prioritize recruitability over simple connectedness.
• The standard degree question may not adequately capture the network dynamics relevant to RDS.
• Further research is needed to refine network measures for improved RDS accuracy.