Populations with higher risk for HIV infection, including men who have sex with men (MSM), people who inject drugs (PWID), and female sex workers (FSWs), are difficult to be recruited as a representative sample of the underlying population, often because of stigma or legal issues. [1], [2], [3] However, accurate information about these hard-to-reach populations (i.e., subpopulations that are difficult to reach or involve in research or public health programs due to their physical and geographical location or their social and economic situation [2]) is necessary for HIV prevention interventions, resource allocation, and relevant policy making to control the spread of HIV.

Respondent-driven sampling (RDS) was proposed in 1997, as a new sampling method improved over snowball sampling to obtain statistically unbiased estimates of population characteristics. [4], [5] In RDS, eligible participants are given coupons to recruit their peers and those new participants are then given coupons to recruit further waves of participants. [6] Mathematical theory and simulation studies purport that statistically unbiased estimates of population characteristics can be calculated in an RDS-recruited study population. [4], [5], [7], [8], [9] However, unbiased estimates based on RDS requires many important assumptions, which cannot always be guaranteed in real world settings. [10], [11], [12], [13], [14] Assessing RDS by using empirical data in real world settings is helpful to measure the gaps between RDS methodology and RDS implementation.

One means of assessment is to evaluate the reliability of naïve RDS samples by using repeatedly collected surveys among the same population in the same location. This paper compares five RDS samples from National HIV Behavioral Surveillance (NHBS) – People Who Inject Drugs (PWID) Cycles in the greater Newark area of New Jersey (Essex, Hudson, Union, Sussex and Morris Counties). The objective of our analysis is to assess the reliability of repeated RDS samples, specifically to assess whether repeatedly collected naïve RDS samples are the same in time-insensitive demographic characteristics (e.g., network size, race/ethnicity, gender, and sexual orientation) and recruitment homophily statistics.