The 2nd Survey Methodology Summer School 2026, jointly organised by the European Survey Research Association (ESRA) and the Centre for Social Informatics, Faculty of Social Sciences, University of Ljubljana will take place from Tuesday the 7th of July 2026 until Thursday the 9th of July 2026 at the University of Ljubljana, Ljubljana, Slovenia.
Programme
7 July 2026
Morning:
How much and how well can AI assist the survey data collection and processing processes? A primer of AI assisted surveys based on the latest research from the academic literature by Mario Callegaro (UK)
Afternoon:
Integrating Large Language Models in the Questionnaire Design Process by Caroline Roberts (Switzerland)
8 July 2026
Morning:
Web tracking: Augmenting web surveys with data on website visits, search terms, and app use by Joshua Claassen (Germany)
Afternoon:
Total Survey Error (TSE) in Cross-National Surveys by Vera Lomazzi (Italy)
9 July 2026
Morning:
Methods for Handling Mode Effects in Analyses of Mixed-Mode Survey Data by Liam Wright (UK)
Afternoon:
Introduction to Missing Data Imputation: Methods and Applications in R by Angelo Moretti (the Netherlands)
Interested parties can attend either individual courses or all courses. For individual courses the fees are £45, £70 for a full day (two courses taught on the same day), or £165 if you would like to attend all six courses. We will offer 50% discount for the applicants who are residents of Slovenia, Croatia, Serbia, Bosnia, Montenegro, Kosovo or North Macedonia. Places are limited and allocated on first come, first served basis. Registration is now open (just log into your ESRA account and click on “Book Short Course”). For suggestions on accommodation, see here.
Detailed Programme
How much and how well can AI assist the survey data collection and processing processes? A primer of AI assisted surveys based on the latest research from the academic literature
Integrating Large Language Models in the Questionnaire Design Process
Course Description:
Effective questionnaire design remains one of the greatest challenges in survey research, requiring a mix of scientific expertise and artistic skill, as well as evaluation and testing. Questionnaires – including how they are administered and how respondents interpret and respond to them – constitute a major source of survey error, but one that can be addressed at relatively low cost. An extensive literature on survey methodology provides guidance on the various pitfalls of poor question formulation, on optimal design choices to improve measurement quality, and on methods available for ensuring research objectives are met, while burden on respondents in minimised. Added to this, recent advances in the field of generative artificial intelligence (GenAI) – notably, increasingly powerful Large Language Models (LLM’s) and chatbots – now provide a new, and ever-expanding, range of tools that can be integrated into different phases of questionnaire development. These not only offer researchers opportunities to save time, but also the potential to optimise the formulation of survey questions. However, as research to validate the effectiveness of such tools remains in its infancy, their integration in the questionnaire design process should be handled in a critical way, based on background knowledge of both the scientific principles and craft of effective social measurement. This course, aimed primarily at beginners, aims to: 1) to present an overview of principal questionnaire design challenges, best-practice guidelines for writing effective questions, and frameworks for evaluating potential sources of error; and 2) to introduce available AI tools and ways they can be integrated at different stages of questionnaire development. Participants will work on practical examples of different types of survey question, to evaluate question problems and identify ways to improve them.
Learning objectives:
At the end of the course, students should be able to:
1. Describe the major challenges of writing effective survey questions, based on theoretical frameworks for identifying potential sources of error;
2. Complete steps involved in writing and evaluating survey questions drawing on best-practice guidelines aimed at minimising measurement error.
3. Integrate AI tools at different stages of questionnaire development and evaluate outputs critically.
Bio:
Dr. Caroline Roberts is a senior lecturer in survey methodology and quantitative research methods in the Institute of Social Sciences at the University of Lausanne (UNIL, Switzerland), and an affiliated survey methodologist at FORS, the Swiss Centre of Expertise in the Social Sciences. At UNIL, she teaches courses on survey research methods, questionnaire design, public opinion formation and quantitative methods for the measurement of social attitudes. She has taught a number of summer school and short courses on survey methods, questionnaire design, survey nonresponse, mixed mode surveys, and data literacy. At FORS, she conducts methodological research in collaboration with the teams responsible for carrying out large-scale academic surveys in Switzerland. Her research interests relate to the measurement and reduction of survey error. Her most recent research focuses on attitudinal and behavioural barriers to participation in digital data collection in surveys, and ways to leverage generative AI in questionnaire design, evaluation and testing. Caroline is currently Chair of the Methods Advisory Board of the European Social Survey and was President of the European Survey Research Association from 2019-2021.
Web Tracking: Augmenting Web Surveys with Data on Website Visits, Search Terms, and App Use
Course Description:
Web surveys frequently run short to accurately measure digital behavior because they are prone to recall error (i.e., biased recalling and reporting of past behavior), social desirability bias (i.e., misreporting of behavior to comply with social norms and values), and satisficing (i.e., providing non-optimal answers to reduce burden). New advances in the collection of digital trace (or web tracking) data make it possible to directly measure digital behavior in the form of browser logs (e.g., visited websites and search terms) and apps (e.g., duration and frequency of their use). Building on these advances, we will introduce participants to web surveys augmented with web tracking data. In this course, we provide a thorough overview of the manifold new measurement opportunities introduced by web tracking. In addition, participants obtain comprehensive insights into the collection, processing, analysis, and error sources of web tracking data as well as its application to substantive research (e.g., determining online behavior and life circumstances). Importantly, the course includes applied web tracking data exercises in which participants learn how to …
1) … operationalize and collect web tracking data,
2) … work with and process web tracking data,
3) … analyze and extract information from web tracking data.
The course has three overarching learning objectives: Participants will learn to a) independently plan and conceptualize the collection of web tracking data, b) decide on best practices when it comes to data handling and analysis of data on website visits, search terms, and app use, and c) critically reflect upon the opportunities and challenges of web tracking data and its suitability for empirical-based research in the context of social and behavioral science. Previous knowledge on web tracking data or programming skills are not mandatory (beginner level). Participants should bring a laptop PC for the data-driven exercises.
Bio:
Joshua Claassen is doctoral candidate and research associate at Leibniz University Hannover in association with the German Centre for Higher Education Research and Science Studies (DZHW). His research focuses on computational survey and social science with an emphasis on digital trace data.
Total Survey Error (TSE) in Cross-National Surveys
Course Description:
When conducting or using data from cross-sectional surveys several elements can compromise the quality of the survey data. This short course introduces key aspects of Total Survey Error in cross-national surveys.
The Total Survey Error framework is the most meaningful approach to understand and address biases that can arise in any phase of the survey cycle (design, data collection, processing, analysis, etc.). Cultural differences may intervene and further impact on the total survey quality.
Using examples of cross-national surveys, the course presents challenges and potential errors involved, their consequences for meaningful comparative research and possible mitigation strategies.
Level: introductory
Bio:
Prof. Vera Lomazzi is associate professor of Sociology at the University of Bergamo (IT). She is board member of the European Survey Research Association. Her research covers comparability issues in cross-national surveys, European values, with a focus on gender equality, solidarity and democracy.
Methods for Handling Mode Effects in Analyses of Mixed-Mode Survey Data
Course Description:
Surveys are increasingly adopting mixed-mode methodologies. Due to differences in how items are presented, responses can differ systematically between modes, a phenomenon referred to as a mode effect. Unaccounted for, mode effects can introduce bias in analyses of mixed-mode survey data. Several methods for handling mode effects have been developed but these have mainly appeared in the technical literature and vary in their ease of implementation. Further, the assumptions these methods make (typically, no unmodelled selection into mode) can be implausible. To improve adoption of methods for handling mode effects, in this interactive short course we will provide background on the problem of mode effects by placing it within a simple and intuitive Causal Directed Acyclic Graphs (DAGs) framework. Using this framework, we will then describe the main methods for handling mode effects (e.g., regression adjustment, instrumental variables, and multiple imputation) and introduce a promising but underutilised approach, sensitivity analysis, which uses simulation and does not assume no unmodelled selection into mode. Finally, we will show users how to implement sensitivity analysis with a hands-on R tutorial using real-world mixed-mode data from the Centre for Longitudinal Studies’ (CLS) birth cohort studies. By the end of the session attendees will:
• Understand why mode effects can cause bias in analyses of mixed-mode data.
• Be able to draw DAGs that represent assumptions about mode effects.
• Use DAGs to design an analysis of mixed-mode data and to identify the biases that may appear in such an analysis.
• Understand methods for handling mode effects, including sensitivity analysis.
• Be able to implement sensitivity analysis within the software package R.
Activities will include:
1. Exercises drawing and interpreting DAGs that illustrate the issue of mode effects.
2. An R practical on implementing methods for handling mode effects using CLS cohort data.
Bio:
Liam Wright: Liam is Lecturer in Statistics and Survey Methodology at the Centre for Longitudinal Studies (CLS), University College London. Liam is Principal Investigator on the Survey Futures project Assessing and Disseminating Methods for Handling Mode Effects. Liam has experience creating tutorials on methods for handling mode effects, as well teaching programming skills. Most recently he has co-authored user-friendly guidance (with Richard Silverwood) on accounting for mixed-mode data collection for users of CLS’ cohort data.
Introduction to Missing Data Imputation: Methods and Applications in R
Course description:
This course provides an introduction to missing data and imputation and does not assume any prior knowledge of the topic. It covers the fundamental terminology related to missing data, with particular attention to missing data mechanisms and missing data patterns, as well as how to identify and distinguish between these mechanisms in practice.
In addition, the course introduces multiple imputation as a principled approach for handling missing data and presents the most widely used R package for imputation and analysis, mice. Participants will gain an overview of how to implement multiple imputation using this package and how to work with imputed datasets in applied statistical analyses.
At the end of the course, students should be able to:
- Evaluate missing data mechanisms
- Perform multiple imputation in R
- Conduct a simple analysis on an imputed file in R.
Prerequisites
- Basic knowledge of R
- Regression models.
Level: introductory
Bio:
Dr. Angelo Moretti is a survey statistician and works at Utrecht University in the Department of Methodology and Statistics. He is an elected member of the International Statistical Institute (ISI). He is part of the European Survey Research Association (ESRA) board, where he is currently serving as Conference Chair.
He is the Principal Investigator of the 5-year ERC funded project ‘Small Area Estimation Methods to Monitor the Progress towards the Sustainable Development Goals at the Subnational Level in the European Union (SAESDGs-EU).
He has conducted research in small area estimation under multivariate mixed-models, survey calibration, mean squared error estimation based on bootstrap approaches, data integration methods (statistical matching and probabilistic record linkage), and weighting of non-probability samples. His work has been published in leading journals such as the Journal of the Royal Statistical Society: Series C, International Statistical Review, and the Journal of Survey Statistics and Methodology.
Suggested Accommodation
