From Start to Finish: Specify - Analyse/Adapt - Evaluate
Small Area Estimation (SAE) addresses the problem of producing estimates for domains where the direct estimators used in the classic survey sampling do not work, due to small sample sizes. SAE has attracted much attention in the last two decades because of its broad applicability. Nowadays, a considerable set of statistical techniques, some quite complex, is available for practitioners. Even though SAE projects share some of the characteristics of any statistical consultancy project, some processes, such as deciding the level of geography, or the evaluation of the final estimates are quite specific. Unfortunately, up to our knowledge, a set of guidelines to help practitioners in this path is not available. In this presentation, we address the issue of how to implement a SAE project in practice, not with a normative perspective but with the main aim of opening the floor for a necessary discussion about this. In order to keep the focus on the practice, the presentation runs around a real case study: the measurement of multidimensional poverty in Mexico.