The emphasis in the 2030 Agenda for Sustainable Development on leaving no one behind (No one left behind) and reaching the most vulnerable presents the statistical community with many challenges and opportunities. Disaggregating data is essential to understanding whether the fruits of development are benefiting the entire spectrum of society, including the most vulnerable and the most backward. It will be a challenge for national statistical systems, including those in developed countries, to produce data disaggregated by various population characteristics and by different levels of geography. New and improved statistical methodologies, innovative techniques, and increased use of administrative data are among the essential initiatives needed to respond to the daunting challenges presented by the need for more detailed data. The 2030 Agenda provides opportunities to enhance and enhance the statistical capacity of countries to produce relevant, up-to-date, and detailed information on the population.
In this way, capacity building in new estimation techniques becomes crucial to provide social statistics for those subgroups that are not well represented in a survey. The Statistics Division of the Economic Commission for Latin America and the Caribbean and the Regional Office for Latin America and the Caribbean of the United Nations Population Fund are focusing their attention on the problem of data disaggregation and the need to provide guidance to countries on this issue. Therefore, this course aims improve the capacity of countries to adopt new methodologies to produce disaggregated SDG indicators, through the use of household survey data that can be analyzed using appropriate statistical techniques.
Requirements
Knowledge of statistics, survey processing and basic use of “R”.
General objectives
The course will allow technicians from national statistical systems to discuss the usefulness of household surveys together with other data sources to produce social statistics for small areas (specific population subgroups). Likewise, current methodologies for combining data sources will be reviewed to achieve higher levels of accuracy and share challenges and best practices to fulfill the call to leave no one behind.
This course aims to present a practical guide for the Bayesian integration of information from the analysis of household surveys, censuses, administrative records and satellite images, through the use of R statistical software and its STAN interface.
Specific objectives
1. Introduce attendees to the Bayesian paradigm as an approach to modeling phenomena in household surveys on the finite population.
2. Present the estimation methodologies of Fay-Herriot (area models) and Battese-Harter-Fuller (unit models) with the Bayesian approach.
3. Present the methodologies associated with complex parameters that deviate from the assumption of normality.
4. Carry out computational practices in R and STAN for all estimation methodologies in small areas.
