ABSTRACT
Given the time- and resource-intense nature of human subjects research, we have developed a more intelligent approach to participant recruitment above and beyond random sampling that leverages pilot or preliminary results to reduce the overall number of participants needed for recruitment from an existing electronic cohort or database. Using open-access data from the General Social Survey (GSS) of the National Opinion Research Center, we generated pilot and validation datasets through a simulation to establish moderate and weak relationships based on linear regression. We then compared the performance of our residual-matching method against random sampling in their probabilities of achieving a given level of statistical power as well as their prediction accuracies. Results showed that the residual-matching method was superior to random sampling, yielding smaller sample sizes with equivalent mean square error. We therefore advocate the use of residual matching when scaling up pilot studies to conserve time and resources in larger follow-up studies.
Competing Interest Statement
The authors have declared no competing interest.