BACK TO SCHEDULE SPEEDTALK

Speed talk session 7

Strengthening Quality Control

4 June 2026
13:15 – 14:00
ŠIBENIK III

Presentation title
Managing Errors in Published Statistical Data and Preventing Recurrence
The goal of the Croatian Bureau of Statistics is to follow the European standards and provide high quality statistics for national and European purposes.

Read more Read less

To ensure the quality of both the European and national statistical systems, a common quality framework for official statistics is used, consisting of the European Statistics Code of Practice and the Quality Assurance Framework of the European Statistical System. Peer review is one of the instruments that ensures the implementation of the common quality framework and the quality of European Statistics, and its identified recommendations help National Statistical Institutes to further improve and develop their statistical systems. As a result of addressing the recommendation of the peer review from 2023, the Croatian Bureau of Statistics has developed Policy of the Management of Errors in Published Statistical Data. The Policy describes the basic framework for dealing with errors that appear in already published statistical data to maintain and improve the quality of the production and dissemination process of official statistics in order to eliminate the cause and prevent the recurrence of the same or similar errors and propose possible improvements. To achieve the objectives of this Policy, the internal Guidelines for the Management of Errors in Published Statistical Data are intended to be applied. The basic framework of action is based on the implementation and timely publication of the relevant corrections, as well as on the analysis of the identified errors, which serves to define corrective actions aimed at eliminating the root cause and preventing the recurrence of the same or similar errors, as well as make suggestion for improvements. By such procedures, error management activities are rounded off through a continuous loop of planning/Plan, implementation/Do, checking or analysis/Check and action/Act, which is applied in accordance with the Total Quality Management model implemented in the Croatian Bureau of Statistics. From the conducted error analyses carried out, it is also possible to identify further actions, which contribute to the continuous improvement of the quality of statistical processes and products. All activities arising from the Guidelines will be documented in a report, including error analyses, corrective actions undertaken, measures to address root causes and prevent the recurrence of errors, and implemented improvements. The implementation of these recommendations will contribute to greater transparency of the Croatian Bureau of Statistics’ work and will strengthen accountability within the established quality system, as well as users’ trust.

Main author / Presenter
Nives Španić Ciceli
Croatian Bureau of Statistics

Read more Read less

Since August 2023 I work at Croatian Bureau of Statistics, Statistical Methodologies, Quality and Customer Relations Directorate/ Sampling, Statistical Methods and Analyses Department, focusing on statistical methodology, sampling design, estimation, quality assessment, and data confidentiality. Prior to my current position, I gained extensive, long-standing experience in accreditation within a national accreditation institute, including roles as Lead Assessor and Quality Manager.

CO-AUTHOR:

Presentation title
Balancing efficiency, burden, and quality: insights from the analysis of response behaviours to the Italian census on PhD graduates’ employment
Although web-based surveys offer substancial operational advantages, they can suffer from low response rates and data quality issues.

Read more Read less

These limitations are particularly relevant in the current context of declining participation in official surveys, which complicates efforts to balance efficiency, respondent burden, and the reliability of collected information.

The present study addresses these aspects within the framework of the most recent Italian census on the employment outcomes of PhD graduates, through an analysis of the response behaviour of individuals involved. The survey was carried out exclusively via a self-administered web mode (Computer Assisted Web Interviewing) and investigated the timing and modes of entry into the labour market for two cohorts of PhD graduates several years after graduation.

A critical constraint was the absence of telephone numbers and email addresses in the survey frame, together with legal restrictions that precluded their scraping from the web. Consequently, the initial contact with potential respondents had to be established via the traditional postal service, a factor that significantly influenced the trade-off between survey efficiency and data quality.

The analysis of response behaviours is based on the final survey outcomes – completed, unanswered, and partially completed questionnaires – and paradata. For completed questionnaires, timestamp-based indicators – weekdays and time of day in which the compilation occurred, connection duration, and number of sessions – provide insights into the respondents’ burden. For partially completed questionnaires, the last answered item offers clues about the reasons for dropping out. Overall, the findings suggest strategies for improving the design and implementation of the survey questionnaire, optimising the reminder schedules, and reducing the respondent burden.

Furthermore, a logistic model is applied to profile the non-respondents and to predict the probability of non-response, using covariates from the survey frame. The following factors are identified as significant predictors: gender, age, nationality, geographical area of residence, disciplinary field of the doctoral degree, and type of university (public or private). The analysis of the characteristics associated with non-response assists in the early identification of groups at higher risk of exclusion. This supports the implementation of targeted interventions to foster participation and mitigate the non-response bias in future editions of the survey and in analogous research contexts.

Main author / Presenter
Alessandra Nuccitelli, Gabriella D'Ambrosio
Istat

Read more Read less

The presenting author is to be defined

CO-AUTHORS:

Presentation title
Quantifying Burden in Official Statistics: Experimental Evidence on Questionnaire Length and the Validation of a Burden Monitoring Tool
Reducing reporting burden is a primary concern for Statistical Institutes seeking to maintain high-quality data while minimizing costs.

Read more Read less

However, effectively reducing burden requires precise measurement. This study evaluates the impact of questionnaire length on the perceived burden of establishments and validates a 6-item instrument designed for routine integration into official statistics production. We specifically test whether perceived burden can be measured consistently across varying levels of objective burden and over time.

We conducted a randomized controlled experiment embedded within three waves (2023–2024) of the IAB Job Vacancy Survey, a high-frequency German official establishment survey (n=3,888). Establishments were randomly assigned to a "short" (2-page) or "long" (4-page) follow-up questionnaire. To ensure the findings were not artifacts of measurement error, we applied multi-group Confirmatory Factor Analysis (CFA) to test for measurement invariance. This allowed us to confirm whether the scale maintains its meaning regardless of the actual length of the survey or the timing of the data collection.

The experimental results demonstrate a significant and measurable increase in perceived burden associated with questionnaire length. Establishments in the 4-page condition reported higher burden across all six dimensions. The analysis also confirmed full scalar invariance across both experimental groups (ΔCFI<0.01) and across longitudinal waves. These results indicate that the 6-item scale is a robust tool that accurately captures fluctuations in respondent burden.

The findings provide evidence that increases in questionnaire length (from 2 to 4 pages) significantly impact the perceived reporting burden in an establishment context. By validating a concise, binary-response scale, this study offers a cost-effective solution for other Statistical Institutes to monitor respondent burden. Implementing this measurement tool allows for a data-driven optimization of survey design, ensuring that efforts to reduce burden are based on empirical evidence rather than intuition. Ultimately, this supports the goal of maintaining high response quality while fostering long-term cooperation among reporting establishments.

Main author / Presenter
André Pirralha, Joseph Sakshaug
IAB - Institute for Employment Research

Read more Read less

André Pirralha is an expert in survey data and survey methodology at the Research Insitute of the German Employment Agency (IAB-BA). He earned his PhD in social sciences and survey methodology from the University Pompeu Fabra in Barcelona in 2016. Following this, he took on the role of leading the cross-cultural comparability project for the Data Quality Assessment team of the European Social Survey at the Research and Expertise Centre for Survey Methodology (RECSM) until 2019. From 2020 to 2023, André worked at the Leibniz Institute for Educational Trajectories (LIfBi) as a survey expert. Joseph Sakshaug is distinguished researcher, deputy head of research, and head of the Data Collection and Data Integration Unit in the Statistical Methods Research Department at the IAB. He is also university professor of statistics at the School of Mathematics, Computer Science, and Statistics at the LMU Munich (Ludwig-Maximilians-Universität München) and honorary full professor at the School of Social Sciences at the University of Mannheim. Previously, he was associate professor (senior lecturer) in Social Statistics at the University of Manchester (UK), and assistant professor (junior professor) of Statistics and Social Science Methodology at the University of Mannheim (Germany). Mr Sakshaug received his MSc and PhD in Survey Methodology from the University of Michigan-Ann Arbor, and his BA in Mathematics from the University of Washington-Seattle. From 2011 to 2013, he was an Alexander von Humboldt postdoctoral research fellow at the IAB and the LMU Munich. He is also a faculty member in the International Program in Survey and Data Science, and an adjunct research assistant professor at the Survey Research Center of the Institute for Social Research at the University of Michigan. His research focuses on data quality issues in complex surveys, the integration of multiple data sources, and empirical research methods.

CO-AUTHOR:

Presentation title
Automating Quality Assurance through GSBPM-Aligned Validation Rules in the Abu Dhabi Statistical Ecosystem
National statistical systems operating in decentralised production environments face increasing pressure to reduce reporting burden and improve cost-effectiveness, while maintaining high standards of statistical quality, consistency, and user trust.

Read more Read less

In Abu Dhabi, official statistics are produced across a distributed ecosystem comprising the Statistics Centre Abu Dhabi (SCAD) and twelve Other Producers of Official Statistics (OPOS). Ensuring consistent quality assurance across this diverse production landscape has traditionally relied on manual checks and ex post reviews, creating inefficiencies and duplication. To address these challenges, SCAD has implemented a comprehensive system of automated quality validation rules embedded within Tbyaan, an AI-powered producer platform used by all official statistics producers in the Abu Dhabi statistical ecosystem.

The quality validation rules are explicitly designed around the GSBPM 5.2 framework to ensure that quality assurance is systematically applied across all phases of statistical production. Quantitative indicators are defined for input, process, and output stages, enabling objective and evidence-based validation throughout the statistical lifecycle. Input validation rules address the quality of administrative data sources, survey data, and other data inputs, including coverage, completeness, and structural integrity. Process validation rules monitor transformation steps, calculation logic, and methodological consistency, while output validation rules assess statistical values against coherence, comparability, plausibility, timeliness, and accessibility criteria prior to dissemination. These validation rules are applied uniformly to all statistical values within the Abu Dhabi Statistical Portfolio, ensuring cross-domain consistency and traceability.

By embedding quality validation directly into the production environment, quality assurance becomes an integral part of routine statistical workflows rather than an ex post reporting exercise. Producers receive immediate feedback on quality indicators, enabling early identification and resolution of quality risks. This approach significantly reduces fragmented documentation, manual reconciliation, and separate quality reporting, thereby lowering reporting burden while strengthening governance and accountability across the ecosystem.

The validation framework is implemented within an AI-powered producer platform that supports automated rule execution, structured approval workflows, and consistent application across SCAD and OPOS. To the knowledge of the authors, this represents the first producer platform that integrates automated statistical production with embedded, GSBPM-aligned quality assurance for official statistics at scale. The Abu Dhabi experience demonstrates how rule-based quality validation can enhance efficiency, consistency, and trust in decentralised statistical systems without imposing additional reporting demands on producers.

Main author / Presenter
Aysha Almarzooqi
Statistics Center Abu Dhabi

Read more Read less

Dr. Haifa Alhamdani is the Executive Director of the Strategy & Planning Sector at the Statistics Centre – Abu Dhabi, a position she has held for over four years. She leads the Centre’s strategic direction, planning frameworks, and transformation initiatives, ensuring alignment with Abu Dhabi’s broader agenda. Dr. Hifa holds a PhD in Economics and has extensive experience in public-sector strategy, economic policy, and statistical system development. Her work focuses on strengthening evidence-based decision-making, advancing institutional performance, and embedding a culture of quality and innovation across the statistical ecosystem.

CO-AUTHORS:

Presentation title
Apples-to-apples: using comparative tables to enhance statistical quality and consistency
Ensuring the production of high-quality statistics is a core mandate of the Statistics Department of Banco de Portugal and a key driver of users trust in official data.

Read more Read less

A fundamental question in this context is whether similar economic phenomena should yield identical values across different statistical outputs. Conceptually, the answer is yes. In practice, however, differences may arise while statistics remain fully consistent, reflecting the use of distinct data sources, production calendars, and methodological frameworks. Over the last years, the Statistics Department has invested in the development of centralized shared repositories, structured around common dimensions and concepts. This centralization, combined with the standardization of statistical language, has made compilation processes more linear, flexible and transparent. It has also created the conditions for the development of an innovative system, supported by interactive dashboards, known as the Comparative Tables. The Comparative Tables system is designed to systematically compare related statistical outputs by cross-checking data for comparable economic phenomena, instruments, and institutional sectors, whenever available. Its primary objective is to strengthen external consistency across statistical domains and, ultimately, enhance the overall quality and coherence of the information disseminated by Banco de Portugal. At present, most tables focus on reconciling financial accounts with a wide range of underlying primary statistics, including balance of payments, monetary and financial statistics, and securities statistics. Beyond these current applications, the system is evolving towards more advanced and automated forms of validation. Ongoing developments aim to refine cross-checking and reconciliation processes at the microdata level, improving the ability to distinguish genuine methodological differences from data errors. In parallel, the planned implementation of a warning system will enable real-time monitoring during the production cycle, allowing potential inconsistencies to be detected and addressed at an earlier stage. In summary, the Comparative Tables initiative is a concrete expression of the Statistics Departments strategy to safeguard statistical consistency through innovation. By embedding systematic cross-domain validation into the production process, it supports more efficient workflows and reinforces the reliability and credibility of Banco de Portugals statistical outputs.

Main author / Presenter
Paula A. Silva
Banco de Portugal

Read more Read less

Paula Silva holds a Bachelor’s degree in Economics and an Executive Master’s in Corporate Finance. She has worked at the Statistics Department of the Banco de Portugal since 2013, where she currently holds the position of Coordinator of the Statistical Quality Unit. She has also worked in the areas of securities issues statistics and central balance sheet database statistics. Prior to joining the central bank, she worked in investment banking and at Deloitte as an auditor.

CO-AUTHORS: