A Quality Framework for Statistical Algorithms

Abstract

The aim of national statistical offices (NSOs) is to develop, produce and disseminate highquality official statistics that can be considered a reliable portrayal of reality. In this context, quality is the degree to which a statistic’s set of inherent characteristics fulfills certain requirements [9]. These requirements are typically set out in a quality framework, which is a set of procedures and processes that support quality assurance within an organisation and is meant to cover the statistical outputs, the processes by which they are produced, and the organisational environment within which the processes are conducted. Many widely accepted quality frameworks related to official statistics exist; for example, see the Australian Bureau of Statistics’ Data Quality Framework [10], the United Nations’ National Quality Assurance Framework [11], Eurostat’s European Statistics Code of Practice [12] and Statistics Canada’s Quality Assurance Framework [13].