Following provision of informed consent, each individual participated in an in-depth interview that lasted for between 1.5 and 2.5 h. These semi-structured interviews followed a standardized guide with broad questions followed by open-ended probes to elicit further information.
The interview guide was developed by a multidisciplinary team that included sociologists, psychologists, ethnographers, and NGO staff with backgrounds in HIV risk issues and substantial experience working in the same Roma communities where participants were recruited.
Topics covered in the guide included friendship and sexual relationships, sexual practices and circumstances when they occurred, views and perceptions concerning condoms and sexual protection, beliefs concerning HIV/AIDS and STDs, and other topics that would elicit data in the domains of interest.
The interviews were conducted by three sociologists and psychologists experienced in using in-depth interview techniques and Roma cultural issues. All had worked previously in the Roma community. They were trained in the use of the study's open-ended interview guide, and each conducted one pilot interview before beginning the formal data collection phase.
Interviews took place either in a neighbourhood field office setting or in participants' own homes. Interviews were conducted individually with only the participant and the interviewer present, and anonymity of participant identity and information was assured.
All interviews were conducted in Hungarian or Bulgarian and were tape-recorded.
This study has several limitations. Participants in this research were recruited from a limited number of neighbourhoods in Hungary and Bulgaria, and the findings may not necessarily generalise to other Roma communities. There were no injection drug users in the sample, and risk patterns related to drug injection were not assessed. In-depth interviews primarily serve to identify themes and dynamics, and not necessarily the prevalence of behaviours within a population. Additional research with larger samples is needed to determine the prevalence of phenomena found in this study. While the study's interviewers had extensive background in in-depth interviewing, sensitive issues were discussed and it is possible that some participants did not disclose about stigmatised aspects of their behaviour.