Data were collected in a total of 10 community pairs. Each pair consisted of members of an at-risk group in a particular geographic region (e.g., sex workers in Long Beach) who were assigned to be either an intervention or a comparison community. Individuals in both the intervention and comparison communities received referrals to community-based HIV prevention services. In addition, peer volunteer networks were established in the intervention communities to distribute role model stories, condoms, and bleach kits. The design is considered quasi-experimental because it was not always possible to randomly assign communities to either the intervention or comparison condition.

Interviewers conducted on-street surveys in the intervention and comparison communities to measure exposure to the ACDP intervention, condom carrying, sexual and drug injection practices, and intentions. These interviews were conducted prior to the implementation of the ACDP intervention and over the 32-month intervention period.

Two main types of data analyses were conducted. Community-level analyses were conducted to assess whether there were community-wide changes between the intervention and comparison groups as a whole. These analyses compared the responses of all of the people in the intervention communities to those in the comparison communities, regardless of whether or not they had seen any of the materials from the ACDP intervention. This type of analysis measures the direct and indirect effects of the intervention.

The other analyses were conducted at the individual level. These analyses compared the responses of people in the intervention community who had been exposed to the ACDP intervention with those who had not. This type of analysis measures only the direct effects of the intervention on those who had received role model stories or talked with a peer network volunteer.