Abstract

National Statistical Institutes are increasingly involved in producing official statistics from non traditional sources. Exploiting Big data, trusted smart statistics, non probabilistic samples and administrative data is a wonderful occasion for providing timely data with a very high level of detail, giving satisfaction to a wider audience. Often, such sources are used jointly, also with traditional surveys, so that the resulting statistics are multisource. These new input features gave also rise to a renewal of the methodologic machinery used in NSIs and a revision of the statistical processes. New approaches are needed to evaluate quality of results and to ensure that they are based on sound methods and worth to be included among official statistics. This session has the objective to gather different experiences and compare possibly different approaches.
The National Statistical Institutes of Italy, UK, Turkey, Mexico and Denmark (invited in this session), are working on different quality aspects in this new data ecosystem. Among the different quality principles, timeliness, coherence and relevance assume a higher importance than in the past and meed to be assessed accordingly. Appropriate measures should also take into account the possible benefical effects on lower response burden on respondents as well as a possible reduction in costs. First preliminary approaches tend to ensure the use of sound methodology, anyway this is one of the most critical aspects to assess, together with measures devoted to accuracy.
A lively discussion on these aspects from different NSIs perspectives is the objective of this proposal.