Possibilities and Limitations of Using Respondent-Driven Sample

Sociological research on hard-to-reach populations (H2R) often faces serious obstacles due to the difficulty of 1) obtaining representative data, 2) reaching the respondents themselves, 3) estimating the sampling error itself due to the lack of quantitative data on the general population. In such cases, the use of traditional methods of respondent recruitment is difficult, since respondents may avoid contact with the researcher. For this reason, researchers use alternative recruitment methods, among which the most popular are the snowball method and the time-location sampling method, since they simplify the procedure for reaching the respondent and build trust between the respondent and the researcher. However, the results obtained using the snowball and time-location sampling methods are often hampered by the difficulty of estimating the response bias due to the lack of procedures for reducing selection bias. In this perspective, the respondent-driven sampling method appears to be an alternative that solves the problem of random selection of respondents, since it is based on the assumptions of 1) the first-order Markov process and 2) the biased chain theory. The article describes the possibilities and limitations associated with a new method of attracting respondents.