Background:
Sampling and recruiting ethnically/racially diverse populations and populations at risk for HIV/AIDS and other STDs poses many difficulties.

Objective:
The Phoenix Project surveyed women about factors hypothesized to be related to willingness to use microbicides, experimental products that may reduce the chance of HIV and other STD transmission during sexual intercourse. A non-proportional quota sampling approach is discussed, along with the utility of recruitment sources.

Method:
A cross-sectional questionnaire was administered using an audio computer-assisted self-interview to 531 sexually active, 18-55 year old Latina/Hispanic (Latina), Black/African American (Black), and White women of negative or unknown HIV status from 4 states in the northeast United States. There was a two-step eligibility process: eligibility based on demographic screening questions and eligibility based on quotas for race/ethnicity and number of male vaginal sexual partners (1 versus 2+) in the past year.

Result:
Approximately equal numbers of Latina (n=166), Black (n=193), and White participants (n=172) were enrolled. In the 12 months prior to the interview, 31% had 3-9 partners, 18% had 10+ partners, and 9% had used injection drugs. Thirty-nine percent reported ever having had an STD. Seventeen percent of those screened were ineligible based on screening questions; an additional 16% were ineligible based on quota closures. Approximately 76% of women interviewed were recruited through word-of-mouth (39%), community-based organizations (19%), and media (19%). Women recruited through word-of-mouth had the highest screen-to-interview completion percentage (67%). Hypothetically, approximately 138 additional women would have been interviewed had quota sampling not been used.

Conclusion:
Non-proportional quota sampling is a feasible option to obtain ethnically/racially diverse and at-risk samples. Community partnership and trust were found to be important factors in recruitment.

Implications:
Quota sampling ensures adequate representation of characteristics of interest, but this goal should be weighed against study staff burden and cost.