Office for Statistics Regulation

Empowering statistical leaders

As part of our work on statistical leadership, we are hosting a series of guest blogs. This blog is from Stian Westlake, Chief Executive at the Royal Statistical Society.

From remote working to online shopping, the pandemic has been a great accelerator of long-term shifts. It has done much the same thing to the role of data and statistics within public life.

This was really brought home to me in late 2020 when the RSS’s panel of distinguished experts sat down to decide our Statistic of the Year. In past years, our choice had usually generated a quirky news piece, intended to highlight how statistics could make sense of the big stories of the year. But when we looked through the nominations for 2020, we realised things were different: the statistics before us were the big stories of the year. The news of the pandemic, its spread, and its impact on lives and on society were being understood through the medium of statistics.

In much the same way, statistics became a central tool of our collective efforts to understand and to tackle Covid. Crucial projects like the Coronavirus infection survey, the UK Coronavirus dashboard, and the RECOVERY trial were as central to the UK’s Covid response as Nightingale hospitals or the vaccine procurement programme, and each was, in its different way, an exemplary statistical undertaking. Statisticians were in demand across government, and proved their value time and again.

So it is extremely timely that earlier this year the Office for Statistics Regulation published its major review, Statistical leadership: making analytic leadership count. Others have written eloquently about several of the main themes of the report, such as the importance of statistical skills, and of transparency and trust. While these are dear to the RSS’s heart, there is another theme in the report that I think is especially important: empowering statisticians to provide leadership, and ensuring they have strong career prospects.

One way of thinking about the crucial importance of empowered statisticians is to consider the counterfactual. What happens when if the other conditions for statistical leadership – such as technical skills and transparency – are met, but if organisations fail to give statisticians the right organisational roles, access and opportunities?

When this happens, we see a very particular failure made. Statisticians are left out of the loop of strategic planning. Data is seen as a specialist function to be commissioned as an afterthought, often to justify rather than inform a decision. And the commissioning process breaks down: statistical projects are assigned by leaders with limited statistical background, sometimes with unrealistic objectives and little chance to iterate during the project. The near-term results are projects that are frustrating to work on and disappointing for users. The longer-term results is that skilled statisticians are demoralised and drift away. We’ve all seen organisations like this; we may even have worked at some. Sometimes statistics or statisticians get the blame, and we hear talk of mutant algorithms or statistical errors. But the root cause isn’t in the data or the methodology: it is a problem of organisation.

But the good practice of the past eighteen months have shown to the world at large that there is a better way. In our experience, this relies on a few elements.

First of all, putting statistics, data and evidence at the heart of the organisation’s strategy. This means those senior leaders who aren’t statisticians gaining the skills to be users of statistics and to work well with statisticians, and statisticians being supported and trained to take senior leadership roles, rather than existing as a permanent auxiliary function. This helps make statistics and data intrinsic to the organisation’s workings.

Secondly, it requires strong career development opportunities for statisticians. Technical skills are important, but for true statistical leadership these need to be complemented with opportunities to learn general management and other operational skills. Crucial to this is mentorship. (This is why the RSS runs a mentorship scheme for candidates for our Chartered Statistician designation.) One of the silver linings for some statisticians of the immense workload imposed by the pandemic has been the exceptional opportunities to try out new roles in other organisations, as statistical and data skills have been at such a premium. Wouldn’t it be good if the volume and quality of these opportunities continued once the burden of Covid-related work has subsided?

Thirdly, it requires managers and heads of profession to be mindful about the make-up of the profession and to ensure it is open, diverse and growing. Research has shown that lack of opportunity and diversity is a big barrier to society’s scientific potential; it is likely that the same holds true for our discipline. This means redoubling our efforts to increase the diversity of the statistical workforce when it comes to protected characteristics. It also means promoting non-traditional routes into the profession, building on the GSS’s apprenticeship and degree apprenticeship scheme, and making the most of in-work skills schemes like the RSS’s Data Analyst, Graduate Statistician and Chartered Statistician designations, and the competency framework we have designed for them.

Getting these vital human-level, organisational questions right is essential for a thriving statistical profession. And that in turn is indispensable for anyone who cares about rigorous, useful, trustworthy statistics.

The Code Pillars: Quality

When I joined OSR as a placement student last September, the Code of Practice for Statistics was unknown territory. It certainly sounded official and important. Was it password protected? Would I need to decipher something or solve a puzzle to get in?

It soon became clear to me that this elusive ‘Code’ was at the heart of everything I would be doing at OSR. Not wanting to remain in the dark any longer, I dutifully dragged it to the front of my bookmarks bar and began to familiarise myself with its contents. (Thankfully no complicated code-cracking required).

The Trustworthiness and Value pillars appeared to be pretty straightforward. Yet, something about the Quality pillar didn’t seem quite so inviting. It sounded like the technical, ‘stats-y stuff’ pillar, that my degree background in economics and politics would surely leave me ill-equipped to understand.

*Spoiler alert* I was wrong.

It turns out that ensuring statistics are the highest quality they can be, isn’t as complicated and technical as I once feared. Quality simply means that statistics do what they set out to do and, crucially, that the best possible methods and sources are used to achieve that.

There are lots of ways that statistics producers can meet these aims. For example, quality can be achieved through collaboration. This can be with statistical experts and other producers, to arrive at the best methods for producing data. It can also be with the individuals and organisations involved in the various different stages of the production process – from collecting, to recording, supplying, linking, and publishing. Collaborating in these ways not only helps to ensure that statistics are accurate and reliable, but also that they are consistent over time and comparable across countries too.

There are lots of other important-sounding documents like our Code of Practice that set out national or international best practise and recognised standards and definitions for producing statistics and data such as the GSS harmonisation standards and the Quality Assurance Framework for the European Statistics System. These also help producers ensure that their statistics and data meet the highest possible standards of quality.

Quality is not only important at the producer-end of the equation, but at the user-end too. It is vital that producers are transparent with their users about how they are ensuring the quality of their statistics. This means telling users about the steps they take to achieve this, and being clear with them about the strengths and limitations of the statistics with respect to the various different ways in which they could be used.

For an indication of just how important quality is, the Quality Review of HMRC Statistics we conducted last year is a prime example. After identifying an error in its published Corporation Tax receipt statistics, HMRC asked us to assess its approach to managing quality and risk in the production of its official statistics. With the Code as our guide, we were able to review HMRC’s existing processes and identify potential improvements that could be made to reduce the risk of statistical errors in the future.

This is just one example of how high-quality data fulfils our vision of statistics that serve the public good. We have found many others across our work and we continue to support producers to consider quality when producing statistics. Last year, we published new guidance for producers on thinking about quality, which was inspired by the HMRC review and the questions we asked.

If you’re interested in finding out more about Quality and the other pillars of our Code, check out the Code of Practice website. I promise it’s not as scary or elusive as it sounds…

Did you know we have case studies on our Code website too? Here are some of our examples that highlight examples of good practice in applying the quality pillar of the Code.

Q1 – Ensuring source data is appropriate for intended uses
Q2 – Developing harmonised national indicators of loneliness
Q3 – Improving quality assurance and its communication to aid user interpretation

Leaving school during a global pandemic

What are the consequences for young people leaving school as a result of the pandemic?

How can more detailed statistics about school leavers help us understand and effect real change for our young people?

Last year we published our UK wide report – Exploring the public value of statistics about post-16 education and skills. This was an in-depth look at the post-16 education sector and covered statistics on workforce skills, universities and higher education, further education and colleges and apprenticeships. Doing a multi country, multi sector report of this nature was for me, a challenge in many ways, not only due to the fact that we were engaging with multiple producers and users all with diverse viewpoints, but also because there was a myriad of different statistics as well as data gaps to consider. We also wanted to ensure that areas of good practice and shared learning opportunities were highlighted across the four nations.

Our research highlighted the following areas as being of greatest importance to the sector:

the coherence of the available statistics, how they inform the bigger picture
the accessibility of the statistics to users
how well the current statistics fully meet the information needs of users and understanding where there may be information gaps

They say that timing is everything and of course by July last year we were in the midst of the global pandemic with the post-16 sector like many others facing immense challenges. We still felt however, that it was important to publish and share what we had found.

One year on…

Across the country this month, young people are leaving compulsory education and making decisions on their future career prospects. As both a parent and regulator in the children, education and skills domain, I think of the tough decisions young people are making, with the stakes seemingly higher than ever in a world of increased uncertainty during the pandemic. We need to ensure that the data available to help them is timely, relevant and accessible to those that need it.

We have been encouraged that, even with the challenges faced by the post-16 education sector, we have seen many of the recommendations we made in our report progress, but there is more to do.

Statistics to make a real difference

Leading the user engagement of the Scotland statistics, I remember how passionately some researchers spoke about the need for good quality statistics to track individuals from their early years in the education system, through to the choices they make in their post-16 years and beyond. They felt this could make a difference, building an evidence base to support targeted interventions at the right time.

It was also an eye opener for me to find out about the complexities around linking this data using a common unique identifier between schools, colleges and universities as well as other post-16 options. Again, the real value comes when the linked datasets tell the stories and thus allow progress and change within the education system. This has benefits beyond those who have been linked in the data as it enables researchers understand issues and develop appropriate solutions for the future.

As we continue our engagement with the relevant statistics producers, we will encourage them to address issues around data granularity, quality and linkage so those working within this sector can understand and effect real change for our young people. As the effects of COVID-19 may affect their outcomes for decades to come – they deserve it now, more than ever.

If you wish to discuss user views for post-16 education and skills statistics please get in touch with us.

Letting the good data shine: The state of the UK system of statistics 2021

At OSR we’ve long been concerned about the risks that a world of abundant information and misinformation could lead to a catastrophic loss of confidence in statistics and data – and that our public conversation, cut loose from any solid foundations of evidence and impartial measurement, would be immeasurably weakened as a result. That is, at root, what we exist to prevent. I have written about this before as a form of statistical Gresham’s Law – how the risk is that the bad data drive out the good, causing people to lose confidence in all the evidence that’s presented to them in the media and social media.

I’ve also said that this is not inevitable, and indeed we can easily envisage a reverse effect: the bad data being driven out by the good data – that is, the trustworthy, high quality, high value data.

What it takes to secure this positive outcome is a public sector statistical system focused on serving the public good. A system that does not regard official statistics as just a Number, shorn of context, calculated in the way it always has been done, some kind of dusty relic. Instead a system that regards the production of statistics as a social endeavour: engaging with users of statistics, finding out what they want and need to know, and responding in a flexible and agile way to meet those needs.

The pandemic has really tested the public sector statistical system and its ability to provide the good data, the trustworthy, high quality, high value data. The pandemic could have seen us being overwhelmed with data from a wide range of sources, some less reliable than others. It could also have seen Government statistics retreating to the Just a Number mindset – “we just count cases, it’s up to others to decide what the numbers mean”. But the system has not done this. Instead, as our report published today shows, the statistical system has passed the test with flying colours.

Statistical producers (producers) across the UK nations and system-wide have responded brilliantly. They have shown five attributes. It’s easy to see these attributes in the work of public health statisticians and ONS’s work on health and social care statistics. They have done great things. But what’s clear to us is that these attributes are system-wide – appearing in lots of producers of statistics and across lots of statistical areas.

Responsive and proactive

Producers across the UK governments have been responsive, proactive and agile in producing data and statistics to support policy and to provide wider information to the public which really adds value. For example the ONS launched the Coronavirus (COVID-19) Infection Survey in swift response to the pandemic. The survey provides high-quality estimates of the percentage of people testing positive for coronavirus and antibodies against coronavirus. These statistics provide vital insights into the pandemic for a wide range of users, including government decision-makers, scientists, the media and the public that are essential for understanding the spread of the virus, including the new variants.

Timely

Producers have responded impressively to the need for very timely data to ensure that decisions around responses to the pandemic are based on the most up-to-date evidence. For example, the ONS published the first of its weekly Economic activity and social change in the UK, real-time indicators (previously called Coronavirus and the latest indicators for the UK economy and society) publications in April 2020, one month after the UK first went into lockdown and has continued to do so ever since. The publication contains a series of economic and social indicators (for example, card spend, road traffic and footfall), which come from a variety of different data sources. These assist policymakers with understanding the impact of the pandemic and gauging the level of overall economic activity. During the early weeks of the Covid-19 pandemic, the Department for Transport rapidly developed near-to-real-time statistics about Transport use during the coronavirus (COVID-19) pandemic. The statistics were regularly used by No10 press conferences (example in slide 2) to show the change in transport trends across Great Britain and gave an indication of compliance with social distancing rules.

Collaborative

Collaboration and data sharing and linkage have been a key strength of both the UK statistical system and the wider analytical community over the past year. This more joined-up approach has improved our understanding of the impact of the pandemic both on public health and on wider areas such as employment and the economy. For example, during the pandemic, ONS and HMRC accelerated their plans to develop Pay as You Earn (PAYE) Real Time Information (RTI) estimates of employment and earnings. The Earnings and employment from PAYE RTI is now a joint monthly experimental release that draws from HMRC’s PAYE RTI system which covers all payrolled employees and therefore allows for more detailed estimates of employees, rather than a sample based approach, as well as information on pay, sector, age and geographic location.

Clear and insightful

We have seen some good examples of clearly presented and insightful statistics which serve the public good. For example, Public Health England (PHE) developed and maintain the coronavirus (COVID-19) UK dashboard. This dashboard is the official UK government website for epidemiological data and insights on coronavirus (COVID-19). The dashboard was developed at the start of the pandemic to bring together information on the virus into one place to make it more accessible. Initially it presented information for the UK as a whole and for the four UK countries individually. Over time it has developed so that data are now available at local levels. We have seen the increasing use of blogs to explain to users how the pandemic has affected data collection, changes to methodologies and bring together information available about the pandemic. For example, the Scottish Government has blogged about analysis and data around COVID-19 available for Scotland. We have also seen examples of statisticians engaging openly about data and statistics and their limitations, both within and outside government, helping the wider understanding of this data and statistics. For example, Northern Ireland Statistics and Research Agency (NISRA) statisticians have introduced press briefings to explain their statistics on weekly deaths due to COVID-19. The Welsh Government Chief Statistician’s blog is a regular platform for the Chief Statistician for Wales to speak on statistical matters, including providing guidance on the correct interpretation of a range of statistics about Wales.

Transparent and trustworthy

For statistics to serve the public good they must be trustworthy, and this includes statistics being used and published in an open and transparent way. We have seen efforts to put information in the public domain and producers voluntarily applying the Code of Practice for Statistics (‘the Code’) to their outputs. For example, the Department of Health and Social Care (DHSC) publishes weekly statistics about the coronavirus (COVID-19) NHS test and trace programme in England. DHSC has published a description about how the pillars of the Code have been applied in a proportionate way to these statistics. However, inevitably the increased volume of and demand for data has placed a greater burden on producers and led to selected figures being quoted publicly when the underlying data are not in the public domain.

But our report also shows how the system cannot take these 5 attributes for granted. What has been achieved in the high pressure environment of a pandemic must be sustained as we ease out of being a pandemic society. New challenges – like addressing regional imbalances, or moving to a greener economy or addressing issues like loneliness and inequality – cannot be understood using objective statistics if the system retreats into the Just a Number mentality.

So, our report sets a number of recommendations. The recommendations aim to make sure that the statistical system we have seen in the pandemic is not an aberration, but is – in the classic pandemic phrase – the new normal. A system that can harness these five attributes is one that serves the public good. It is the best way to ensure that the bad data do not thrive and the good data shine out.

Glimmers of light for adult social care statistics

I was very interested in a recent Social Finance report on how to secure impact at scale. One of their points is that, if you want to see impact at scale, you need to be willing to seize the moment. Change arises when supportive policies and legislation fall into place, and when a new public conversation starts.

This idea – new policies, and new public conversations – made me think of social care statistics. It’s very tragic that it has taken the disastrous impact of the pandemic in care homes to focus attention on this issue, but there seems to be some potential for progress on the statistics now.

The background is that we’ve been shouting about the need for better statistics for the last couple of years. We’ve done so through reports on social care statistics in England , Scotland and Wales . We’ve done it through presentations and I’ve taken the opportunity to highlight it when I’ve given evidence at the Public Administration Committee in the House of Commons.

Certainly, we found some excellent allies in Future Care Capital and the Nuffield Trust, yet it has sometimes felt like we’re in the minority, shouting in the wilderness.

What were our concerns? Our research in 2020 highlighted that there were several challenges and frustrations related to adult social care data that were common to England, Scotland and Wales. Our report summarising the common features of the statistics across Great Britain highlighted four key needs to help both policymakers and individuals make better informed decisions about social care:

Adult social care has not been measured or managed as closely as healthcare, and a lack of funding has led to under investment and resourcing in data and analysis.
There is an unknown volume and value of privately funded provision of adult social care. Although data is collected from local authorities, this only covers activities that they commission and fund, which constitute a smaller proportion of total adult social care activity.
Robust, harmonised data supply to ensure comparable statistics from both public and private providers is problematic, as data collection processes are not always standardised. Furthermore, data definitions might not always be consistent across local authorities and other providers.
Data quality is variable within and across local authorities, with inconsistent interpretation of data reporting guidance by local authorities. This means that data isn’t always reliable and so users have difficulty trusting it.

As data issues go, as the pandemic has highlighted, there is not so much a gap as a chasm, with consequences to our understanding of social care delivery and outcomes.

Most people we’ve talked to, inside and outside the UK’s governments, recognise these issues. But to date there hasn’t been much evidence of a sustained desire to inject energy into the system to effect change.

Maybe, though, there are glimmers of light. Whilst this list is not meant to be exhaustive, I would like to draw attention to some initiatives that have caught my eye.

The first comes from an extremely negative space. That is the pandemic’s impact on those in care homes. Not only has the pandemic highlighted the importance of care and care workers, it has also led to much more interest in data about the care home sector. The Care Quality Commission and the Office for National Statistics (ONS) collaborated to publish timely information on the numbers of deaths in care homes , to shine a light on the impact of the pandemic for this vulnerable population. And DHSC has commenced the publication of a monthly statistics report on Adult social care in England to fill a need for information on the social care sector itself. This means that COVID-19 has resulted in people listening to analysts and statisticians when we raise the problem with social care data now. Of course, the questions people are interested in go well beyond COVID-19.
The Department for Health and Social Care’s draft data strategy for England makes a significant commitment to improving data on adult social care.
The Goldacre Review for data in England may present a further opportunity for improvement.
I was pleased to restore the National Statistics designation to the Ministry of Housing, Communities and Local Government’s statistics report about local authority revenue.
Beyond the pandemic, ONS is working in collaboration with Future Care Capital to shine a light on one of the biggest data gaps here: the private costs borne by individuals and families for care. And ONS has recently published estimates of life expectancy in care homes prior to the pandemic.
Adult social care remains high on the political agenda in Scotland, with the recently published independent review of adult social care by the Scottish Government and the inquiry by Scotland’s Health and Sport Committee.
The Welsh Government remains committed to improving the data it captures on social care .

It’s far too early to declare “problem solved”, but we ought to be optimistic about improvements to data as a consequence. We’ll be reviewing the actions currently underway as statistics producers react to the gaps in social care statistics highlighted by the pandemic and publishing a comprehensive report of our findings in the autumn.

What I do think is that there is an opportunity here – if statistics producers across the UK are willing to take it, we can anticipate much better statistics on this sector. And a much better understanding of the lives and experiences of citizens who receive, and provide, social care.

The Code Pillars: Trustworthiness is about doing things differently

Trust can seem a no-brainer. It may seem so obvious, that of course it matters. It has often featured as the guiding aim of many a strategy for statistics.

I spend much of my time explaining about the Code of Practice for Statistics and our three pillars. I think of Quality as being the numbers bit – getting the right data, using the right methods. I think of Value as being why it all matters, making sure to serve the public and society. And Trustworthiness? Well, Trustworthiness for me is a lot like reputation – as Warren Buffett once said:

“It takes 20 years to build a reputation and five minutes to ruin it. If you think about that, you’ll do things differently.”

So, the Trustworthiness pillar is about ‘doing things differently’ – for analysts and their organisations. You can’t expect just to be trusted – you must earn it.

You have to behave honestly and with integrity – you can show that in the way that you use numbers. Anyone who spins data to make themselves look good, or cherry picks the numbers that tell the best story, will reveal themselves to be untrustworthy. But those that set out the facts plainly and objectively will be seen as someone to be trusted.

How you handle data and show respect to people and organisations giving their personal information can also prove that you are a safe pair of hands. But if you are seen to lose people’s data, or share it inappropriately, you’ll probably find people are not willing to share their information with you again.

And the way you release information matters – if you give any sense of being under the sway of political opportunism, the credibility of your statistics will be harmed.

So why isn’t the pillar called ‘Trust’ if that is what we are after?

Well, the answer is thanks to the seminal work of philosopher Baroness Onora O’Neill. She said that focusing on trust is fruitless – instead, you need to demonstrate that you are worthy of trust.

Basically, you can’t force someone to trust you. You can only show through the way you behave, not just once, but repeatedly, that you are honest, reliable, and competent:

tell the truth
do what you do well
and keep on doing these

Being reliable in these ways will mean that people will come to trust you. But the only bit you can do is show you are worthy of trust.

So, if you reflect on your reputation for being trustworthy and you want to be sure to keep it, do things differently.

Here are some case studies on our Code website that illustrate some ways that statistics producers show their Trustworthiness:

Office of Rail and Road showing independent production and the managed handling of statistics and data
Department for Transport demonstrating leadership in the development of new statistics on transport use during the pandemic
Welsh Government being transparent about its user engagement and quality management approaches
Department for Transport developing statisticians’ coding skills to meet future organisational needs
Scottish Government demonstrating transparency when linking and publishing data

How Alan Turing’s legacy is inspiring our work today

To coincide with his birthday, on 23 June 2021, the UK honoured the life and work of Alan Turing, one of its most famous mathematicians, by featuring his image on the design of its latest £50 note.

Although best known for his codebreaking work at Bletchley Park, Turing’s legacy goes far beyond his contributions during the war. Recognised by many as an early pioneer of modern computing, his work on algorithms, computing machinery and artificial intelligence have changed the way we live today.

The early 1900s was the start of a data revolution in the UK. Punch cards were being used to input data into early computers, and statistics and science were opening the door to a world of technological possibility. This time of rapid discovery and progress was of immense inspiration to Turing, who foresaw that automation and ‘computing machinery and intelligence’ would have a huge impact on the world and the way we live in it. In fact, the new £50 note features a quote from Turing saying, ‘This is only a foretaste of what is to come, and only the shadow of what is going to be.’

We at OSR use modern advancements in Turing’s work by applying Machine Learning methods to gather data on statistics from places like Twitter, Government websites, parliament reports and the media in order to support us in our regulatory work. This automation helps us to gather large amounts of important data, from a multitude of sources that we wouldn’t have been able to capture before. These techniques will allow us to find where and how statistics are being used in the public domain and help shape our future work.

But although we can see direct impacts of Turing’s legacy in how we conduct some of our own work, perhaps it is the attitude he had to his work that should inspire us more.

Turing was a pioneer in his field, not just because of his keen mind, but because of the way he approached problems with both intellectual bravery and pragmatism. He was not afraid of huge radical ideas, but he was also able to think of how they be used practically, for the betterment of humanity. The idea of public good was at the forefront of his work, helping him to approach old problems with a new, novel perspective.

Making the world a better place has always been a driver to mathematical and scientific discovery, and Turing was a great proponent of applying maths and statistics to solve real world problems. At Bletchley park and beyond, he was able to show the value of accumulating data and how important statistics are to making informed decisions. This type of thinking; calculating probabilities and using plausible reasoning to make decisions, has been a major influence on the way governments around the world have used data to tackle the coronavirus pandemic.

He spent a lot of his time theorising about the concept of intelligence and how it applied to both humans and machines, but even he knew his own intellectual limitations. In his pursuit of knowledge and answers, he often spoke with people from differing fields to his own, discussing problems with philosophers for example. He knew that collaboration only added strength to problem solving and that working together with others would lead to better outcomes.

It is impossible to discuss the life of Alan Turing without remembering that he was persecuted for being gay. Although applauded for his intelligence and work during the war, he was arrested because of his sexuality and forced to take experimental medications. He ended his own life soon after his conviction.

It would not be a huge stretch of the imagination to think that, had Alan Turing’s life not ended prematurely, he would have continued to make intellectual discoveries that would have further positively impacted the world today. It is also not hard to imagine that there are many other diverse, intelligent minds that should have equal chance to contribute to solving the world’s problems. Turing’s life should inspire us not only to new intellectual heights, but to stronger commitments to equality and diversity as well.

As a regulator of statistics, the links from our work to Turing’s are many. Not only do we use automation to help us gather and analyse large data sources, but we question the methodology and fairness behind algorithms and are also passionate collaborators, seeking input from others as an integral part of our processes rather than just an afterthought.

Just as Turing was driven to use mathematics to tackle real world problems and benefit humanity, the accurate use of data and statistics to make decisions for the public good is at the heart of everything we do. Going forward, not only will we continue to explore new ways in which automation can aid our work, but we will strive to collaborate with diverse minds, continue to teach the importance of transparency, quality and value and above all protect the public’s interest by making sure they have statistics they can trust.

Thank you to the Alan Turing institute whose talk, Breaking the code: Alan Turing’s legacy in 2021, helped inform the content of this blog.

OSR in the pandemic and beyond: Our year so far

The first half of 2021 has seen further lockdowns, an impressive vaccination rollout and as we ease into the Summer, some easing of restrictions across the UK. It’s also been a busy time for us, as we continue to push for the production of official statistics, and other forms of data, that serve the public good. We really feel that public good matters more than ever.

We recently published our business plan for 2021/22 in which we outline our focus for the statistical system over the coming year and how it can consolidate the huge gains made in data collection and publication. We have also made progress we have made on our role in data. Our review and findings for developing statistical models to award 2020 exam results may well be the most comprehensive review of the 2020 exam story. It’s comprehensive in two senses. It covers all four parts of the UK, unlike other reviews. And it goes beyond technical issues about algorithms to identify lessons for all public bodies that want to use statistical models in a way that supports public confidence. We have also published an insightful review on Reproducible Analytical Pipelines and our research programme.

The use of statistics during the pandemic

Statistics have and will continue to play an important and extremely visible role in all our lives. I recently provided evidence to the inquiry run by the House of Commons Public Administration and Constitutional Affairs Committee on the use of data during the pandemic. Since the start of the pandemic, Governments across the UK have maintained a flow of data which has been quite remarkable. We continue to push for further progress, for example on vaccination data.

Statistical Leadership

One thing that the pandemic has highlighted is how important it is for leaders to be analytical, and this is something that our Statistical Leadership report published recently highlights.

Good analytical leadership will be crucial to answering the many questions that have arisen over the course of the pandemic and continue to come to light including the importance of transparency. We are currently planning an in-depth discussion on these issues and more for our second OSR annual conference, which we aim to host later this year, focusing on high quality data, statistics and evidence.

Looking Forward

There are lots of good things happening for statistics at present. I was delighted to see changes to pre-release access in Scotland because equality of access to official statistics is a fundamental principle of statistical good practice.

I am also really looking forward to announcing the results of our 2021 annual award for Statistical Excellence in Trustworthiness, Quality and Value in July. This is the second year we have worked in partnership with the Royal Statistical Society to offer the award.

Keep up to date with our latest work and news by following us on Twitter, and sign up to our monthly newsletter.

The productivity puzzle: Looking at how productivity in the UK is measured

This is a guest blog from Stuart McIntyre, Head of Research at the Fraser of Allander Institute.

Productivity is a term that economists intuitively understand, but it can often be difficult for non-specialists to track and understand productivity changes.

Post the global financial crisis of 2008/09 (GFC), productivity in the UK has both underperformed relative to other developed economies and underperformed historic growth in productivity in the UK. This has come to be known as the ‘productivity puzzle’.

Why has UK productivity performance seemingly slumped? And why should people care?

UK productivity performance might seem like it doesn’t matter to the average worker – but this would be to misunderstand what it represents.

Improvements in productivity, producing more with the same or fewer inputs like hours worked, is a key element in unlocking improvements in wages for workers and competitiveness for businesses.

As the Nobel Prize winner Paul Krugman once said, “Productivity isn’t everything, but, in the long run, it is almost everything. A country’s ability to improve its standard of living over time depends almost entirely on its ability to raise its output per worker”. Indeed, one of the key reasons why living standards in the UK have stagnated over the last few years has been the weak performance of UK productivity.

Productivity matters, and for people trying to understand how the economy is performing productivity is a key indicator, and not just as a number, but also in comparison to other parts of the country and indeed other countries.

How is productivity measured in the UK?

The Office for National Statistics produce regular updates on UK productivity. This includes measures of labour productivity, output per hour worked and per job, as well as multi-factor productivity– which measures the change in economic output that cannot be explained by changes in inputs like labour and capital.

One challenge remains that many people take labour productivity measures to be the same as overall productivity, which ignores broader measures of productivity. In particular, our understanding of changes in productivity require us to look beyond what is happening to labour productivity, and look at whether businesses are investing. We also need to consider changes in innovation, regulation, and the business environment.

Since the GFC we’ve seen significant improvements in the quality and range of productivity statistics, and also significant investment by researchers inside and outside of government to understand why the UK’s productivity performance has been so poor.

Even among experts, there are competing assessments of the root cause and it seems like a number of factors might be important. Explanations range from weak investment in skills and training, as well as in research and development, through to factors causing weak demand in the economy. Another interesting explanation centres on whether part of the explanation lies in how we measure productivity.

All of which invites significant investment in our economic data – and since the GFC we have seen a lot of this take place.

We now have more and better productivity statistics in the UK. For example, we know much more about how productivity differs between firms, between regions, and between the UK and its international competitors.

How could we improve how productivity is measured?

We’ve seen some important measurement issues identified in the academic literature being addressed, as well as the testing of key ideas from the academic research about the role of management practice in explaining productivity performance, through the collection of new survey data from businesses.

Yet there remains lots still to do.

We’ve seen increasing use of so—called administrative data, like VAT returns, to measure what is happening in the economy in a timelier manner. These data developments are vital – and have an important role to play in measuring productivity.

At the same time, we know how important measurement issues can be in tracking productivity in particular parts of the economy – this is increasingly important as the economy undergoes significant change with digitalisation and growth of new technologies. It must remain a key area of focus.

One area of economic statistics where there is increasing investment and focus is sub-UK statistics – as the UK Government embarks upon its ‘Levelling Up’ agenda this will be increasingly important. In particular understanding what is happening in the regions and nations of the UK. This is an area where productivity statistics could also use further investment.

In particular, providing more timely estimates of regional productivity, but also by ensuring that analysis and statistics at a UK level are paralleled at a sub-national level. This also goes for measurement of factors that are important in explaining movements in productivity, for example what is happening to capital investment.

There is also scope to help non-expert users engage with these statistics. The past few years have seen growing use of infographics and social media – which has helped the latest data reach a broader audience.

More generally, we’ve seen increasing focus on setting out the uncertainty around economic statistics. It is clear from recent evidence that users of economic statistics benefit – in the form of improved understanding of what the statistics show – from having the uncertainty inherent in economic statistics explained to them. In my view, this is an area where the productivity statistics could also be enhanced.

In summary, the post GFC period has seen renewed interest and investment in understanding UK productivity. In particular investing in more local statistics, making greater use of administrative data, and in the communicating and explanation of these statistics to non-expert users. We’ve come a long way, but there remains much still to do.

Making an impact: How can statisticians help journalists to answer key questions with data?

As part of our work on statistical leadership, we are hosting a series of guest blogs. This blog is from Robert Cuffe, Head of Statistics at the BBC.

It’s not all rush, rush, rush in news. One colleague, helping me with my transition from the months-away-deadlines of clinical research to the five-minute-deadlines of breaking news, revealed that when he worked on the evening TV news he had often had the luxury to go away and “just think about a story for ages, like, easily 15 minutes”. It didn’t help much.

Does it need to be so frenetic? And, if so, what does that mean for departments or statisticians trying to get their data covered broadly and correctly?

We can debate about five minutes versus fifteen, but the news machine does need to work very quickly. Editors are the last line of defence between everything that happened in the world recently and your news feed.

Every select committee, every ministerial pronouncement or opposition line, every report from a charity, think tank or government statistician, every freak goal from any team anywhere in the world, every man who bit a dog, every new collection, every sleb indiscretion and for quite some time almost every major number about the pandemic gets reviewed and someone has to decide “does this make”?

If it does, do we need to send a video crew to capture footage and interview bystanders, how quickly can we get 400 words up online to summarise this, who are good voices to put this into context, does the final script or writeup need to be run by the duty lawyer? And so on.

And, if not, what else can we put in the paper or in the bulletin? Because the program is happening at 6pm (or the print run starts at 7pm) whether or not we’re ready.

So decisions have to be made pretty quickly. And fast, good decisions about data need a prominent, clear and succinct summary of what the data can or can’t say.

As statisticians, we understand better than anyone else what the data can say. But our deep understanding of and intuition for the principles of our specialism do not always help with the prominent, simple summary. There are professionals who specialise in that translation: journalists or communications officers.

Here are some good guidelines given to me by an excellent journalist, who wanted me to help editors understand (in reverse order)

What are the caveats?
How prominently do we need to alert the reader/listener/viewer to them?
Can we run the story with this top line (or a different top line that the caveats don’t undermine)?

I’ve learnt to choose my caveats carefully: pushing hard on some and forgetting completely about others. Almost every number comes with a margin of error, but not every number story needs to include a confidence interval at the top. If the CI contains the null, why run the story? And if the bottom of the CI is miles away from the null, who cares that the truth could be a teeny bit higher or lower?

Equally, getting a key caveat, say “the time series jumps suddenly because the data definitions have changed” on page 28 won’t do when decisions or press releases are based on the executive summary.

So getting in the room for the painful discussion of final sign-off on a report or a press release, when much of the meeting feels like it’s about where the semi-colon goes, can be the most valuable contribution a statistician can make to the public understanding of their data. The day before the rush, there’s more time for discussion. Prepping the ground by explaining in advance to journalists or trusted intermediaries which questions the data can and can’t answer (rather than talking about specifically what answers the data contain) is also enormously valuable. Communication should be right first time. Statisticians can help make that happen.

PS –Delivering data during the pandemic is no mean feat. Trying out improvements to the service or data collection, as many have done, is remarkable. Thank you to all those producing statistics for all the work you have done in what has been an awful year for so many.