We conducted 2 cross-sectional, observational surveys in 20 of patrons of the 1 sex club and 2 bathhouses located in Seattle, Washington. Our main intent was to describe the populations who visited these venues on any given week during these periods, report the frequency of HIV risk behaviors among these populations, and identify the correlates of high-risk behaviors. A secondary intent was to evaluate the comparability of the 2004 survey population, which achieved a low level of participation (30%, or 373 of 1252 eligible patrons), with the respondents to the much briefer 2006 survey, which enrolled 61% (199 of 20) of eligible patrons. The observed similarity of these populations would assuage concerns about selection bias in the in-depth 2004 survey. We used a time-venue probability-proportional-to-size cluster sampling design 22 in each survey to schedule sampling events across club hours of operations. The sampling events were conducted during randomly selected, 4-hour venue, day–time units (VDTUs), with the probability of random selection proportional to the size of the patron population in the specific VDTU.
For the 2004 survey, we first based probabilities for selecting each VDTU on estimates of patron volume at times throughout the week observed during a previous study of these 3 venues.
7 These probabilities were then revised to reflect observed patron volumes during the first weeks of the survey.
