Thirty Years of The Network Scale-up Method

Estimating the size of hard-to-reach populations is an important problem for many fields. The network scale-up method (NSUM) a relatively new approach to estimate these by asking respondents question, “How X’s do you know,” where X population interest (e.g., female sex workers know?”). answers questions form aggregated relational data (ARD). NSUM has been used variety subpopulations, including workers, drug users, and even children who have hospitalized choking. Within methodology, there are multitude estimators hidden population, direct estimators, maximum likelihood Bayesian estimators. In this article, we first provide in-depth analysis ARD properties techniques collect data. Then, comprehensively review different estimation methods in terms assumptions behind each model, relationships between practical considerations implementing methods. We apply models discussed one canonical dataset compare their performance unique features, presented supplementary materials. Finally, summary dominant extensive list applications, discuss open problems potential research directions area.