ESRA 2025 Preliminary Program

Monday 14 July    Tuesday 15 July    Wednesday 16 July    Thursday 17 July    Friday 18 July   

All time references are in CEST

Register-based sample surveys in practice
Session Organisers	Dr Hafsteinn Einarsson (University of Iceland) Mr Kim Backström (Åbo Akademi)
Time	Friday 18 July, 09:00 - 10:30
Room	Ruppert 040

Interest in utilising administrative data sources in survey practice has grown in recent years. Many countries are providing greater access to administrative register data which may contain valuable information for improving survey outcomes. In many European countries, data from various registers can also be linked with survey data at the individual level, which holds the potential for producing rich sample frame data. We encourage submissions of papers which focus on utilising administrative register data to increase quality in any aspect of survey practice, including in:

1. Informing survey design and fieldwork efforts. For instance, how can register data inform sampling strategies, questionnaire design, when and how to contact sampled units, survey modes offered, and how to reduce survey costs?

2. Evaluating and minimising total survey error. For instance, are register variables associated with unit nonresponse and measurement errors? Are there noticeable trends in these associations over time and are adjustments informed by these variables effective in error minimisation?

3. Under which circumstances should researchers rely on register data rather than survey data and vice versa? For instance, can register information replace survey items (e.g. receipt of benefits) and thus reduce response burden? Can survey responses be used to correct register errors?

4. Any other topics relating to utilising register data in survey practice, such as linkage, privacy, statistical disclosure control, how can national statistical institutes increase access to register data or enrich survey data with register information, etc.

Keywords: Registers, Administrative data, National statistical institutes, Augmentatation, Total survey error

Papers

Combining register data and stratified random samples to account for nonresponse bias: Findings from a decade of recruitments to a probability-based online panel

Mr Felix Cassel (University of Gothenburg) - Presenting Author
Dr Sebastian Lundmark (University of Gothenburg)

Utilizing register data may be one fruitful way to offset the type of nonresponse bias stemming from some people being both harder to recruit and more likely to attrite from the panel. This study presents efforts by an online access panel to decrease their nonresponse bias by combining register data with information about a panel’s demographic nonresponse bias when recruiting new panelists. For over a decade, the Swedish Citizen Panel (SCP) has recruited panelists using simple random samples of the Swedish population. However, doing so led to a heavily skewed panel composed of those both more easily recruited and those less likely to leave the panel. To reduce this skewness, a collaboration with the authority in charge of the Swedish registries (Statistics Sweden) was initiated, where several registries were combined to facilitate stratified random sampling of the Swedish population. These stratified random samples were estimated to account for the likelihood of specific groups signing up to the panel in previous recruitment efforts and the overall skewness of those groups in the already existing panel. The invited sample in each recruitment was divided into 48 demographic groups (Sex * Age * Education * Birth Country), where the size of each group in the sample depended on; 1) the group’s share of the Swedish population, 2) the group’s share of active participants in the SCP, and 3) the number of invited and participated persons from each group in the latest recruitment. Our presentation demonstrates how these recruitment efforts affected the skewness of the panel, the panel’s ability to draw stratified demographic samples of panelists, and the recruitment’s effect on reducing nonresponse bias in terms of reported attitudes and beliefs. The recruitment approach presented will provide insights to survey practitioners on how register data can be utilized in online probability-based

Do I Catch them All? Evaluation of a Strategy to Identify Mothers in Register Data

Mrs Dana Müller (Institute for Employment Research) - Presenting Author

The Research Data Centre of the German Pension Insurance (RV) and the Research Data Centre of the Federal Employment Agency (BA) at the Institute for Employment Research (IAB) provide sensitive register data derived from identical data sources, and additionally offer different data with respect to the legal mandate of each authority. For instance, information on the date of childbirth is only available in the register data of the RV. To overcome this lack of information in the IAB data, Müller and Strauch (2017) developed a strategy to identify mothers and childbirths by using different reasons for deregistration in the register data. The stata script is published and can be adapted to different data available at the IAB. However, using non-direct information to identify childbirths could lead to unintended misspecification or possible selection bias. There are two obvious issues. On the one hand, the identification strategy processes information on employment interruption with compensation from the statutory health insurance. This compensation is paid during maternity leave but also during long-term sickness and could lead to women being misclassified as mothers. On the other hand, childbirths could go unobserved due to unregistered breaks in employment. The aim of this paper is to evaluate the reliability of the identification strategy by using uniquely linked data from the RV and BA, the "Biographical data of selected insurance agencies in Germany" (BASiD) from 1975 to 2007, and to deviate recommendation for which research questions the identification strategy can be used. Results show that 82 percent of childbirths can be found independent from women being employed or unemployed five months before childbirth, and that the probability of finding childbirths is higher for the first child than for all subsequent siblings.

Müller, Dana and Katharina Strauch (2017): Identifying mothers in administrative data. FDZ-Methodenreport

Reporting Survey Fieldwork Outcomes when Sampling from Register Systems

Dr Hafsteinn Einarsson (University of Iceland) - Presenting Author
Professor Joseph Sakshaug (Ludwig Maximilian University of Munich / University of Mannheim / Institute for Employment Research)

Conformity in reporting fieldwork outcomes is essential for ensuring cross-survey comparability. To this end, many leading journals and cross-national survey projects have advocated for more widespread adoption of AAPOR standard definitions among survey researchers. However, as sampling strategies may differ between countries, such comparisons can be complicated. Many European countries rely on register systems to form sampling frames, where administrative data from multiple sources can be deterministically linked to produce detailed information about units prior to the start of fieldwork. Here, we focus on the classification of non-contacts in such surveys, differentiating between fielded and non-fielded non-contacts. Non-fielded non-contacts are cases which have known sampling probabilities, as they are included in the sample frame, but are not contactable by the mode(s) specified by the survey design. We argue that fielded and non-fielded non-contacts can arise from different mechanisms and call for different responses for survey designers wishing to reduce nonresponse rates units associated with either category. We consider practical examples of why non-fielded non-contacts might occur and discuss how to report these cases using the AAPOR standard definitions.

Advancing Register-Based Surveys in CEE: Legal Barriers, Data Accessibility, and Academic Applications

Mrs Anna Micał-Čujová (Jagiellonian University in Cracow) - Presenting Author

Register-based surveys are increasingly recognized as a key method for data collection, offering enhanced accuracy and cost reduction for researchers. However, in Central and Eastern Europe (CEE), their implementation remains in the early stages, presenting specific challenges and opportunities. This study examines the legal and accessibility frameworks for utilizing register-based surveys in three CEE countries: Poland, Slovakia, and the Czech Republic.
Using comparative analysis, the study assesses the legal frameworks, data accessibility, and practical applications of register-based surveys in these three countries. It evaluates the availability and accessibility of administrative registers to both the public and the academic community, as well as the quality and interoperability of data.
The analysis investigates the legislative environments that regulate the use of administrative registers for statistical and scientific purposes. It identifies barriers along with cross-country variations in data protection laws, particularly GDPR. Additionally, the study addresses the technical and infrastructural challenges that impede the effective utilization of these data for research purposes. Special attention is given to the balance between data accessibility and privacy protection, a critical issue for ensuring public trust and ethical research practices. Additionally, the practical experiences of academic researchers in these countries will be presented, serving as real-world examples of end users.
The findings highlight differences in GDPR implementation and data access policies, with Poland and the Czech Republic offering better data accessibility compared to Slovakia. The study suggests the need for harmonizing access rules, improving data interoperability, and balancing data protection and efficient registry usage. Overall, the research emphasizes the importance of aligning legislative frameworks, enhancing infrastructure, and fostering collaboration between government institutions and the academic community.

The feasibility of routinely using administrative data sources in UK survey practice.

Ms Gerry Nicolaas (The National Centre for Social Research, UK) - Presenting Author
Mr Andrew Phelps (The Office for National Statistics, UK)
Ms Véronique Siegler (The Office for National Statistics, UK)

In the absence of a population register, high quality social surveys in the UK rely on address-based sampling. This approach provides almost complete coverage of the population, but information about residents is lacking. Consequently, we are restricted in how to optimise the design and implementation of social surveys to improve the quality of estimates and/or reduce costs.

Although it’s unlikely that the UK will have a population register in the foreseeable future, the Office for National Statistics is developing a Registration Data Management Framework (RDMF) which is built from five administrative data sources that link and match data on addresses, businesses, classifications, demographics and location. Although this framework is being designed for data linkage purposes within the Government Statistical System, it could also potentially be used in the design and implementation of surveys, providing opportunities to significantly transform the end-to-end survey process and to produce higher quality statistics more quickly and more cheaply.

In this presentation, we will present the design of a study which will explore (a) how the RDMF could be used to improve the end-to-end survey data collection process and reduce costs and (b) what the constraints are and possible solutions for using the RDMF in the design and implementation of social surveys (including data protection, legal requirements and public acceptability). At the time of the 2025 ESRA conference, we will be able to present some initial findings from the study.