Maintaining and restoring quality

This section focuses on quality issues within the UK statistical system, concentrating on declining survey response rates, specific ONS economic statistics quality concerns, change programme considerations and the future role of administrative data.

Survey response rates

To produce high-quality reliable statistics, it is essential to have high-quality data inputs. Statisticians aim for data collections that provide representative coverage, whether sourced from surveys or administrative records. While many statistics across the system are produced to a high quality and meet the needs of users, declining responses to statistics household surveys are negatively affecting the quality of key UK statistics.

In the UK, and in other countries worldwide, there has been a consistent decline in household survey response rates. Although the pandemic has worsened this trend, issues with response rates existed prior to this crisis. This year, challenges with household survey response rates have become particularly significant, especially in ONS surveys.

It is important to note that, although survey response rates are an important factor in determining quality, they do not solely determine the overall quality of the resulting statistics. While high response rates contribute to extensive coverage and reliability, it is essential to understand them as just one proxy for overall quality. It is also important to consider the extent to which the sample remains representative of the general population.

We have set out in Annex A, for broader information, the trends in survey response rates for a number of key surveys. These data show that the picture is not uniform. As an example, surveys in Northern Ireland do not appear to have experienced the same reductions in response rates as have been experienced by other surveys. It should be noted that while ONS has experienced large drops in response rates, it appears that rates for some of its surveys may be stabilising. It is becoming critical, however, that the factors which affect variations in response rates to statistical surveys across the UK and topic areas are better understood by the whole statistical system.

Our systemic review of ONS’s economic statistics, which was published in April 2025, identified that ONS has faced significant challenges in consistently ensuring that the data feeding into economic statistics are of high quality. Most notably, our review noted the long-term reduction in response rates to household surveys, including the Labour Force Survey (LFS), has severely impacted data quality. The LFS, which is produced by ONS, is the source of the most important official statistics collected by social surveys in the UK.

We have been reviewing labour market statistics from the LFS and the development of the Transformed Labour Force Survey (TLFS) for a number of years. The quality issues with the LFS due to declining survey response rates have impacted statistical outputs across and beyond ONS. We removed the accreditation of ONS’s Great Britain LFS-based estimates and datasets in November 2023. To date, 13 other statistical outputs have had their accreditation temporarily removed due to issues with the quality of the LFS and associated Annual Population Survey (APS) produced by the ONS. This knock-on effect to key statistics produced by government departments and the governments of Scotland and Wales illustrates the complete dependence on ONS data that exists in some topic areas in the UK.

ONS’s Social Surveys Quality Recovery Project aims to improve household surveys by increasing the number of field interviewers and sample sizes. Communication campaigns, incentives and better pay and benefits are aimed at enhancing participation and retention and have led to an increase in responses for the Labour Force Survey. ONS has made progress in 2025 in securing improved response to the Labour Force Survey, but the long-term challenges have been associated with a decline in the quality and reliability of some of its key economic statistics. We found in our economic statistics review that action should have been taken sooner and at a greater pace, with the necessary funding put in place, and the emerging survey problems should have been better publicly communicated. As such, we set out a requirement for ONS to produce a fully resourced plan for its survey operations implementing a rolling programme of regular reviews of individual surveys and other data sources.

ONS published this plan on 26 June 2025 in response to the urgent need to address survey challenges and to support a wider strategic shift within ONS to focus on sustaining surveys into the future. ONS aims to restore confidence in its survey operations, enhance data quality and modernise systems and methodologies.

ONS has had notable success in moving its business survey collections online; around 2.1 million of the total 2.4 million questionnaires issued are now sent online. The response rates to ONS business surveys reflect substantial recovery towards pre-pandemic levels. For example, response rates for the Annual Business Survey averaged 77.2% for the three years prior to the pandemic, dropping to a low of 53.0% for the reporting year of 2019, but recovered to 77% for 2023.

Although business survey response rates have held up well, reflecting the mandatory nature of these surveys, there is a growing risk to their quality and representativeness. As highlighted in our review of ONS’s economic statistics, there are notable challenges associated with sample reviews, rotation and quality assurance. These challenges indicate that ONS needs to make additional efforts to prevent the associated risks becoming a serious quality issue.

As noted earlier, if a survey sample remains representative of the general population despite a reduction in response rates, it can still give users confidence in the quality of the data. In October 2024, we reaccredited the estimates from the Crime Survey for England and Wales, which had been temporarily suspended since July 2022 due to quality concerns. ONS has demonstrated that the representativeness of the Crime Survey sample remained broadly consistent before and after the pandemic. On that basis, we were satisfied with the quality of the current estimates, and that ONS provides sufficient assurance to users about the limited impact of the lower response rate.

Northern Ireland collects data for its own separate quarterly Labour Force Survey (LFS), which is unaffected by the ONS LFS accreditation decision and therefore remains an accredited official statistic. Since July 2021, a ‘knock to nudge’ strategy of visiting addresses, securing cooperation and conducting phone interviews later has boosted response rates close to pre-pandemic levels. This strategy to try to maintain household response rates is being implemented whilst work to move to an online-first transformed Labour Force Survey called the Northern Ireland Labour Market Survey (NI LMS) is also underway.

The independent report on the Statistical Assembly suggested that more proactive research was needed on what influences response to surveys and what is an acceptable level of response. We fully support this view. Research on this topic currently includes work being conducted by ONS and also through the UKRI-ESRC-funded Survey Futures, which aims to deliver a step change in approaches to collecting population survey data in the UK. Currently, however, there are no UK-statistics-system-wide solutions for the challenges with survey collections. We consider that this is a critical area for the statistical system to address.

Quality of ONS economic statistics

The UK’s economic statistics are evolving, with new and innovative data becoming available. The economic statistics produced by ONS, the UK’s national statistical institute, are vital for understanding the economy and a key source of information and intelligence for those making important decisions on tax, spending and interest rates. These statistics need to be of good quality, use data sources that reflect the modern economy and clearly communicate any uncertainties.

Since the UK’s departure from the EU, Eurostat no longer verifies UK statistics. As the UK’s independent regulator of statistics, we have developed a Spotlight on Quality framework based on the Code of Practice for Statistics and incorporating frameworks like the ESS Quality Assurance framework and the IMF Data Quality Assessment Framework.

Our Spotlight on Quality assessment programme provides a detailed review of many of the data sources and components that feed into the production of gross domestic product (GDP) and the broader National Accounts. This regulatory work has been vital to highlight issues and areas for improvement in economic statistics, which are produced primarily by ONS.

Over the last few years, a number of economic shocks have brought increased interest in, and scrutiny of, the economic statistics produced by ONS. Over the last year in particular, there has been growing external criticism of ONS.

Common factors identified in our regulatory work of ONS economic statistics are the limitations and potential impact to quality posed by software systems and ‘legacy system’ issues:

• The performance of supporting IT systems was a key source of concern in our October 2024 assessment of ONS’s business investment statistics – we found that many of the systems used to produce business investment statistics are outdated. For example, Common Software (a legacy system) cannot be updated and so increases the risk of errors in the statistics.
• In May 2025 ONS wrote to us to advise that, during an investigation for future improvements, an error was found in the existing production system for Producer Price Indices (PPI). ONS advised that the system was outdated and needed to be uplifted as a matter of urgency. This error had resulted from the chain-linking approach introduced in November 2020 being incorrectly coded. At ONS’s request we have suspended the accreditation of PPI outputs while ONS undertakes further analysis of their quality.
• Our May 2025 assessment of ONS’s trade statistics found that ONS is working to move away from the legacy system and is developing requirements for a new system as the legacy system used to process trade in services data is inefficient, no longer fit for purpose, putting strain on resources and restricting capacity for continuous improvement.

In the light of our emerging common concerns about the quality of economic statistics arising through our Spotlight on Quality programme, we initiated a Systemic Review of ONS Economic Statistics in 2024 and published the first report in April 2025. We concluded that it is critical that ONS takes decisive action to restore confidence, and we set out a number of requirements for ONS to meet by July 2025.

Our requirements included that ONS publish a resourced plan that sets out the risks that continued data quality challenges pose to economic statistics; be more explicit on how resources are aligned with its core purposes and outputs as a national statistical institute; publish a regularly updated vision and strategy for the data sources used to compile its economic statistics; and take a more strategic and systematic approach to quality reviews of its data sources.

ONS’s plan for economic statistics, published on 26 June 2025, aims to restore confidence and improve the quality of its core statistics. ONS is open about the current situation and where it needs to do better. The publication of the plan is a crucial part of its response to actioning the immediate requirements in our review of its economic statistics.

When considered alongside its Surveys Improvement and Enhancement Plan for Economic Statistics, which seeks similar improvements in ONS’s household and business surveys and was also published on 26 June 2025, and its forthcoming revised strategic business plan, we conclude that ONS’s actions meet the immediate requirements identified in our review. But there is clearly much more to do.

Change programmes and administrative data

By focusing resources and funding on specific programmes of change, statistics producers can reduce risk and use modern methods and data sources to deliver improved statistics more efficiently. We consider that the importance of producers understanding and mitigating any risks emerging during any transitions to new methods and data sources has become more pronounced, as the risks associated with change work have become clearer.

All transformation work is a form of change management. The distinction between a change and a transformation can often be unclear and may vary depending on the context of the change. Transformation is a fundamental and strategic shift in how things are done. In the context of data and statistics, it can mean rethinking and redesigning the way data are collected, processed and used. This might involve moving away from traditional surveys towards more-integrated systems that use administrative and other big datasets (sometimes integrated alongside survey data), new digital tools and new statistical models to process these data. Such work can also serve to reduce production risk, providing an opportunity to review methods, systems and processes to ensure they are fit for purpose.

ONS is undertaking a significant amount of change work. Specifically, it is transforming how it produces labour market statistics for Great Britian and population statistics for England and Wales. It is also undertaking work in other areas, for example UK long-term international migration and GB travel and tourism statistics.

ONS has been developing a transformed version of the Labour Force Survey (LFS), known as the Transformed Labour Force Survey (TLFS), using a digital collection approach. ONS has been developing and dual-running the TLFS alongside its recent work to improve the LFS. The TLFS was intended to address many of the concerns and shortcomings of the LFS, but this work has also faced significant delivery challenges.

On 3 December 2024, ONS published an interim action plan for the TLFS based on the results of its lessons learnt review conducted in summer 2024 and published the detail of an independent methodological review. In our February 2025 update, we noted that ONS has increased its engagement, communication and openness on the programme and set recommendations for ONS to set out detailed plans for transitioning to the TLFS and for regular reporting on the progress of the interim action plan from its ‘lessons learnt’ exercise. ONS shared an update in April 2025 on the improvements it has made to the TLFS and on its plans to carry out a readiness assessment in collaboration with its main users in July 2026.

The Devereux Review of the Office for National Statistics performance and culture, published on 26 June 2025, highlights several areas where change is needed in how ONS manages its programmes. In its plan for economic statistics, ONS has recognised the need for a revised approach to implementing change, which places greater emphasis on continuous improvement. This approach includes dedicating the right mix of skills to each production area, while continuing to use change programmes where appropriate.

NISRA is also improving labour market statistics through survey transformation and increasing the use of administrative data and other data sources. NISRA intends to move to a new online-first Northern Ireland Labour Market Survey (NI LMS) later this year. NISRA’s LMS is being developed in collaboration with ONS, and ONS and NISRA are working together to enable UK-wide key labour market statistics, derived from a combination of the TLFS and NI LMS, to be available.

Balancing innovation with the need to sustain essential statistical outputs is a significant challenge that requires careful resource allocation and strategic planning. It is necessary to balance the ambitions for research and development required for any change programmes with the need for consistent operational delivery that meets user requirements and resource constraints. This challenge applies to the entire UK, not just the ONS.

Administrative data sources play a key role in many change projects, particularly as data sharing and data access capabilities increase. Administrative data are collected for operational purposes, often when people interact with public services, with statistical use being a secondary purpose.

Our review of ONS’s economics statistics looked in depth at the role and potential role that administrative data can play. We found the following:

• There have been good examples of the use of administrative and big data sources in core ONS economic statistics, including the introduction of rail transaction data in price statistics, but our review found that progress on the use of administrative data has been much slower than expected at the time of the 2016 Bean review.
• Overall progress in utilising administrative data has been slow due to practical and cultural barriers within government departments. These barriers include issues related to data privacy, data quality and the willingness of departments to share data. Overcoming these barriers is crucial for the successful integration of administrative data into statistical processes.
• There is a growing recognition that administrative data cannot always replace survey data directly. A more nuanced approach is needed to assess how these data sources can complement each other, which involves understanding the strengths and limitations of both administrative and survey data and finding ways to integrate them effectively to enhance the overall quality of statistical outputs.

ONS is actively transitioning towards using admin-based population estimates (ABPEs) as the official population estimates for England and Wales. ABPEs have potential to provide more timely, and potentially more accurate, population data compared to estimates derived using the traditional cohort-component method. Following its recent evaluation, ONS intends for the ABPEs to become the official population estimates for England and Wales in 2026. Given the extensive use of population statistics, it is vital that this new methodology has appropriate oversight and scrutiny.

We are undertaking a phased assessment approach to provide reassurance to users on ONS’s new methods for producing population estimates in England and Wales. ONS has used the findings of our first assessment to help shape and steer its development work for the ABPEs. In October 2024, ONS published an action plan for how it will develop population statistics. We will continue to engage with ONS and consider any learning from this transformation work for the wider statistical system.

More broadly, ONS aims to make more use of administrative data as part of its Future of Population and Migration Statistics (FPMS) programme. This programme of work includes considering the future of the census.

The National Records of Scotland (NRS) and NISRA are also engaged in research and are developing methods to make more use of administrative data in their population statistics. In December 2024, NISRA published a research paper exploring the creation of a statistical population dataset for Northern Ireland using a range of administrative datasets. In February 2025, NRS published its latest statistical research, presenting a revised version of Scotland’s Administrative Data Based Population Estimates. This research includes administrative-based estimates of population, household and ethnicity for the period 2016 to 2022, created by acquiring and linking data from a range of sources.

The UK Statistics Authority has recommended that the UK government commission a census of England and Wales in 2031, to be delivered by ONS. The Authority recommendation sets out an approach that delivers a high-quality census for England and Wales in 2031, while further developing population estimates based on administrative data – such as tax records, NHS data and driving licences, which have the potential to deliver more-frequent and higher-quality population estimates. It considers that this approach will strengthen the statistical system so that it is ready to adapt to the technological and social changes of the next decade.

In Scotland, the National Records of Scotland has recommended a mandatory questionnaire-based census approach in 2031 which will incorporate learnings from the 2022 census, including the effective use of administrative data. In Northern Ireland, NISRA has recommended conducting a full Census of Population and Housing in 2031, emphasising the need for high-quality data on various demographics and advocating for government support in accessing administrative data.

Case study – data sources in the criminal justice system in England and Wales

Building strong relationships with data suppliers and service areas helps statisticians to better understand administrative data and improve their outputs. It also helps data providers appreciate the importance of accurate reporting.

In spring 2024, HM Courts and Tribunals Service (HMCTS) and the Ministry of Justice (MoJ) identified inaccuracies in the Crown Court data that are used to produce HMCTS’s monthly management information and MoJ’s quarterly accredited official statistics on the criminal court system in England and Wales. Statistics releases and use of the data were paused and, upon investigation, a range of causes for the Crown Court data inaccuracies were identified by the MoJ. These included human error, technical issues which allowed the use of free-text fields and data coding issues. In our March 2025 review, in which we confirmed the continued accreditation of the statistics, we commended both organisations on their collaborative approach – known as ‘One Crown’ – to investigating and addressing the Crown Court data quality issues and to creating a more transparent and reliable dataset that benefits all users. Better collaborative working between statisticians and data providers has enhanced the understanding or the administrative data sources and improved the quality of these statistics.

In summary, while the use of administrative data is undoubtedly advancing, the maturity of administrative data sources and systems as statistical data sources has not progressed as quickly as hoped. Concerns about progress on the use of administrative data in statistics were highlighted in the Independent report on the 2025 UK Statistics Assembly. Based on engagement with stakeholders during the Assembly, the report identified as a priority the need for the Statistics Authority and Government Statistical Service to take a leadership position in a significant scaling up of the use of administrative data across the system.

Recommendations