Large-scale datasets are generally used for the creation of global indicators within an area or territory. In statistics there is a variety of statistical methods appropriated for this purpose. However, during the years there has been an accelerated demand for the creation of local indicators. A relevant problem arise when the number of available information within the areas is too small (called here small area). In this case, the use of “direct estimates” (that is, estimates based only on the available information within the areas) might be unacceptable due to its large standard errors. In order to avoid this problem we can use Small Area Estimation (SAE) methods. In SAE we “borrow strength” information from related areas to find more accurate estimates for a given area or, simultaneously, for several areas. SAE can bedivided broadly into: design-based and model-based methods. A common feature in both approaches is the use of auxiliary information which is obtained from large surveys and/or administrative records such as censuses, registers, etc. During this presentation I will present a general overview of SAE methods and particularly some techniques related with the model-based approach.
2016. október 19.
Open lecture by Betsabé Garrido Perez
Small Area Estimation methods: A general overview
Date:19th October (Wednesday) from 10 to 11 hrs.
Venue: Sgraffito-House - iASK-Library