What is it about?

Modern official statistics requires a combination of statistics, data science, engineering, programming, literacy, ethics, and domain knowledge—not a single technical skill set. National Statistical Offices are increasingly working with big data, machine learning, remote data collection, web technologies, and new digital sources. These changes create new opportunities for producing official statistics, but they also require skills that go beyond traditional statistical training. This study develops a Data Science Model for Official Statistics (DSMOS) that brings together statistical engineering, data engineering, data analysis, software engineering, statistical thinking, statistical literacy, and knowledge specific to official statistics. The model was validated through a survey of experts working in European statistical systems. The results identify key competencies for modern official statistics, including machine learning, programming, data visualisation, high-frequency and spatial data, big data, microdata, statistical literacy, ethics, and new approaches to data quality.

Featured Image

Why is it important?

Official statistics is changing rapidly as statistical agencies adopt new data sources and emerging technologies. To use these tools effectively, National Statistical Offices need professionals who can combine statistical expertise with programming, data engineering, machine learning, data visualisation, ethics, and modern data-quality methods. The DSMOS framework provides a structured way to identify these competencies and connect them to the statistical production process. It can therefore support training strategies, curriculum design, workforce development, and institutional planning within national statistical systems. The broader goal is to help statistical organizations modernize while maintaining the quality, reliability, security, and trust expected from official statistics.

Perspectives

The digital transformation of official statistics is not only a technology challenge; it is also a skills and education challenge. Traditional statistical knowledge remains essential, but modern statistical production increasingly requires interdisciplinary capabilities. Data scientists working in official statistics need to understand data acquisition and processing, machine learning, programming, visualization, statistical quality, ethics, privacy, and the institutional context in which official statistics are produced. Our proposed DSMOS framework brings these competencies together and provides a basis for designing education and professional-development programs that reflect the evolving needs of National Statistical Offices.

Prof. Afshin Ashofteh
Universidade Nova de Lisboa

Read the Original

This page is a summary of: Data science training for official statistics: A new scientific paradigm of information and knowledge development in national statistical systems, Statistical Journal of the IAOS, September 2021, IOS Press,
DOI: 10.3233/sji-210841.
You can read the full text:

Read

Resources

Contributors

The following have contributed to this page