Starting with adolescents residing in an inner-city neighborhood, a ra
ndom-walk sampling design was implemented with the aim of examining th
e socioecological context of behaviors pertinent to HIV transmission.
A three-step, four-node random walk was attempted for each initial res
pondent, n = 43, of a probability sample of 12-15-year-old boys and gi
rls who resided in a census-tract delimited area. The success rate in
identifying and locating named associates was approximately 62%. The r
esponse rate of locatable associates exceeded 90%. No systematic item
non-response or refusal was apparent in the interview process. The maj
or obstacle to implementation was difficulty locating named associates
and not, as anticipated, nonresponse or refusal to disclose sensitive
information.