|
Abstract:
|
Respondent driven sampling (RDS) is applied to studies targeting population subgroups that are difficult to sample using traditional methods. Using the connectedness within the target group, RDS traces the link between one node and its altars in waves for recruitment. A key premise in this process is pre-existing social networks within the target group. Another key premise is participants’ cooperation to recruitment requests. If participants do not recruit their peers, RDS may not materialize. As recruitment noncooperation in RDS is not well understood, assessing viability of RDS implementation is difficult. This study uses 11 independent RDS studies from 2003 to 2017 to improve our ability to explain the recruitment success by identifying its covariates. Further, we will attempt to build a prediction model. For doing so, we will sort the data chronologically and develop a baseline model using the earliest data. When applied to the next data, we will update the baseline model through Bayesian as well as non-Bayesian approaches and compare predicted recruitment success against the actual outcome in order to ascertain the effectiveness of different modeling approaches.
|
