ESRA Special Interest Groups
This page contains information on ESRA Special Interest Groups (SIG). The general information on SIG, how to create one, expected research activities and others are provided separately here.
List of all ESRA SIGs:
Multi-Source Statistics
Chairs: Camilla Salvatore, Angelo Moretti
The Multi-Source Statistics Special Interest Group (MSS) brings together researchers and practitioners interested in the theory, methods, and applications of survey research that combine multiple data sources. As survey research increasingly operates in a complex data ecosystem, integrating probability-based surveys with administrative records, digital traces, non-probability samples, and other non-traditional data sources has become essential for improving inference and for studying phenomena that cannot be adequately measured by a single data source. The MSS-SIG provides a forum for exchange, collaboration, and methodological advancement within ESRA’s core mission.
Topics
The MSS-SIG focuses on, but is not limited to, the following topics:
- Integration of different data sources, including surveys, administrative data, registers, and digital trace data
- Inference using multiple data sources, with attention to bias, variance, and uncertainty propagation
- Inference with non-probability samples, including calibration, weighting, and model-based approaches
- Integrating surveys and digital traces, such as web data, social media, sensor data, and paradata
- Data integration methods, such asrecord linkage and statistical matching
- Quality assessment and validation in multi-source settings
- Ethical, legal, privacy and governance challenges in combining data sources
Plan of activities
- Establish and maintain a SIG mailing list to support communication, collaboration, and membership management.
- Organise three webinars per year, with at least one webinar addressing EDI-relevant topics (e.g. hard-to-reach populations, under-represented groups, or equity in data integration).
- Offer a short course or tutorial at the biennial ESRA conference focused on multi-source statistical methods.
- Organise and propose SIG-led sessions at international conferences, including ESRA, ITACOSM (Italian Conference on Survey Methodology), International Statistical Institute and International Association of Survey Statisticians related conferences.
- Hold regular online SIG meetings (once or twice per year) to update members on activities, review progress, and plan future initiatives.
Contacts
Camilla Salvatore (c.salvatore@uu.nl)
Angelo Moretti (a.moretti@uu.nl)
Artificial Intelligence in Survey Research
Chairs: Liam Wright (University College London), David Bann (University College London), Paulo Serodio (University of Essex), Gabi Durrant (University of Southampton)
Description
The Artificial Intelligence Special Interest Group (AI-SIG) brings together researchers and practitioners interested in applying artificial intelligence and machine learning (ML), including (but not limited to) large language models (LLMs), natural language processing (NLP), and deep learning, to any aspect of survey research, from questionnaire design and data collection to processing, analysis, and dissemination.
The AI-SIG is a forum for sharing developments within the fast-moving field of AI and for fostering ideas and collaborations.
Topics
The AI-SIG is interested in the use or integration of AI and machine learning into any stage of the survey lifecycle or the analysis of survey data. Topics include:
- AI for the design of surveys, including the identification of valid items or research topics a survey should cover.
- AI for survey data collection, such as the use of LLMs as interview agents or for automated coding of open-ended responses.
- AI for the production or navigation of survey metadata, including the use of natural language processing for variable discovery.
- AI for data cleaning and generating analytical code.
- Deep learning as a statistical modelling tool.
- Evaluating and validating AI/ML outputs in survey contexts, including the reliability and reproducibility of AI-assisted coding, classification, and analysis.
- Synthetic survey data and AI-simulated respondents (e.g., “silicon samples”), including their potential uses and limitations.
- Privacy, ethical, and legal considerations in the use of AI/ML for survey research, including data protection when using cloud-based AI services, informed consent in AI-mediated data collection, and algorithmic transparency and fairness.
Scope
The SIG’s scope encompasses the full range of AI and machine learning methods relevant to survey research, including but not limited to: large language models and generative AI; supervised and unsupervised machine learning; natural language processing; deep learning and neural networks; and computer vision.
The group welcomes contributions from researchers using any of these approaches, whether as the primary object of study or as tools within a broader research workflow.
Plan of Activities
- Run a mailing list supporting communication between researchers and practitioners interested in AI.
- Organise four webinars per year on latest developments, aimed at audiences with varied experience with AI.
- Organise at least one symposium per year at a relevant international conference, such as ESRA and the Royal Statistical Society.
- Organise SIG-led sessions, symposia, or invited panels at ESRA and other relevant conferences.
- Run one short course on using AI per year delivered at the chairs’ home institution, at a relevant international conference, or online. Courses will be open to external participants.
Contacts
Liam Wright (University College London) (liam.wright@ucl.ac.uk)
David Bann (University College London) (david.bann@ucl.ac.uk)
Paulo Serodio (University of Essex) (pamato@essex.ac.uk)
Gabi Durrant (University of Southampton) (g.durrant@soton.ac.uk)
