In a "snowball sample" each respondent is asked to suggest other persons for inclusion in the research. These persons are then contacted to see if they wish to serve as research participants. This is a valid procedure often used by investigators who seek to recruit from populations for which adequate sample frames are not available. For example, a researcher seeking to study patterns of informal leadership in a community may ask individuals to name others who are influential in a community.
Snowball samples, in and of themselves do not necessarily pose a risk for human subjects. Researchers assess the project for risks of harm commensurate with normal life. Each respondent is normally given the opportunity to participate or to decline participation.
For research projects studying sensitive topics, investigators generally should ask persons to inform others about the research and ask them to contact the investigators to express an interest in participation. Study protocols should adhere to the recommendations for confidentiality. For example, studies of networks of drug users or tracking sex partners of HIV+ cases require extreme caution with information gathered from one subject about another. All information should be treated confidentially. (See Certificate of Confidentiality)