Office for Statistics Regulation – Enhancing the public value of UK official statistics through a new Code of Practice
What is the Office for Statistics Regulation?
It might be helpful to explain briefly the structure of the UK official statistics system is structured and how the Office for Statistics Regulation (OSR) fits into that.
The UK Statistics Authority has two functions:
-One for producing statistics, which is carried out by the Office for National Statistics
-And one for regulating statistics, which is carried out by us, the Office for Statistics Regulation. Our Director General is Ed Humpherson.
OSR was actually launched in November 2016. Before, we were known as the Monitoring and Assessment team of the Authority. One of the main drivers launching a separate regulatory office was a review of the state of ONS’s economic statistics by Sir Charles Bean. The Bean review included some criticism of the governance of the UKSA and its regulatory activities. There was a sense that we have ‘marked our own homework’, i.e. saying that ONS has had undue influence on regulator.
But, we have been publicly critical of ONS’s statistics, and economic statistics in particular on a number of occasions.Including removing the designation of NS from UK trade stats, RPI, construction price and cost indices, overseas travel & tourism (as well as crime statistics).But, perception is essential to credibility. We clearly have not been seen as separate from statistics production. OSR creates visible separation, and allows us to continue to regulate effectively and independently.This is crucial to ensuring that the public has confidence in our decisions as a regulator.
What is OSR trying to achieve?
At the heart of everything is our belief that statistics are an essential public asset that should meet society’s needs. Statistics provide a window on society, the economy and on the work and performance of government. They are fundamental to the judgements made by the public, government and a huge range of other organisations.
We aim to do three things:
-Enhance public confidence in the trustworthiness, quality and value of statistics
-Challenge when standards are not maintained
-And also champion relevant statistics
The way that we ensure that users of statistics can have confidence in the trustworthiness, quality and value of statistics is by having a Code of Practice that outlines the standards that producers of statistics must meet.The current Code of Practice (Code 1.0) has served us extremely well and been an excellent tool for maintaining standards of official statistics. But statistics and data look different now than they did when we launched Code 1.0 back in 2009.We are in the middle of refreshing the Code of Practice – to ensure that it remains relevant.
I’d like to spend the rest of talk telling about the new Code – what it will look like and how it will help enhance the public value of statistics.While the scope of the new Code is official statistics, we aim to have a wider reach – I’ll come back to this later.
How did we get here?
It’s been a long journey and conversation to get to the stage we’re at now – with a draft of the refreshed Code. It’s been a very iterative and collaborative process.
It started with the Stocktake of the Code of Practice, which was launched in late 2015.There were two key drivers for the stocktake.The first is the Bean review of Economic Statistics, which I’ve already mentioned.The second is the changing nature of information production and dissemination – it has changed a lot since the first edition of the Code was launched almost 10 years ago.
We produced a report based on our findings which we published in December 2016It included a number of recommendations, included the need for a new Code (that was more transformational than the existing one – less of a checklist and more enabling.)
Following this initial report we did a huge amount of engagement with statistics producers to gather views on what the new Code should include and what it should look like, through focus groups and workshops. This was extremely helpful for developing our thinking. Based on all these comments and our work, we ended up with a draft of a new Code. We are currently running a consultation on the draft of the refreshed Code.
What will the new Code look like and how will it enhance the public value of statistics?
The whole of the new code is structured around three pillars: trustworthiness, quality and value – three concepts that mentioned at the beginning of the talk.
Trustworthiness is about trusted people, systems and processes. In other words, it’s about the environment in which the statistics are produced. Quality is about robust data, method and statistics. And Value is about statistics serving the public good.
It’s important to consider these three pillars together, because they are closely linked. Trustworthiness is the bedrock for Quality. Without trusted people to produce the statistics, and trusted systems and processes to collect,analyse and disseminate the statistics, it is difficult to produce statistics of high Quality. And both Trustworthiness and Quality are needed before statistics can add Value, i.e. serve the public good. At the same time, there’s no point producing statistics that are Trustworthy and of high Quality if they don’t answer society’s important questions. The barriers to public trust in statistics could come from any one of these pillars. The lack of impartiality of statistics would reflect low Trustworthiness, issues with data quality would reflect low Quality, and lack of engagement with users of the statistics would reflect low Value.
Examples: trade statistics (Quality), student migration statistics (Trustworthiness, pre-release access)
Underneath each of the three pillars are a number of principles that set out what we mean by T, Q and V. So, for example, for Trustworthiness, we have integrity, good governance, statistical capability and protected confidentiality. And to have Value, statistics must be accessible, insightful, reflect the range of users and uses, be innovative, and be the result of efficient data collection and use.
The new Code will promote a stronger sense of the need for government statisticians to innovate and to work collaboratively. It also encourages maximum possible release of low level data – useful for those who want better and quicker access to microdata, for example.
Wider scope of the new Code
There is greater use of other types of statistics and data now than there was when the original Code was launched back in 2009.Now, digital information is abundant and much easier to collect and produced, and often, data will come from multiple sources rather than just one source.
Therefore, we think it is important to advocate the Code’s pillars and principles as best practice to a wider group. In developing the new Code, we’ve thought more broadly about the production of numerical information. We see trustworthiness, quality and value as a framework that has wider relevance to different types of numerical information. By different types of numerical information I meanthings like Management Information, adhoc releases, and datasets, produced by both official bodies like government departments, and non-official bodies, like wider public bodies, charities and think tanks.In recognition of this wider reach, the new Code will have a different name – it will be called the Code of Practice for Statistics (we’ve dropped the Official bit).
In spite of this wider reach, the scope of OSR remains official statistics.The bulk of the work that we do will still be assessing official statistics against the new Code of Practice and looking across the system of official statistics to identify themes and issues.The aim of widening the scope is not designation – we’re not planning to put the badge on more stats.The aim is to raise awareness of the standards in the Code and their wider relevance. By encouraging their adoption, we will help inform the quality of public debate, no matter what type of organisation is producing the numbers, and ultimately, enhance the public value of wider numerical information.
To encourage the use of the Code’s pillars and principles beyond producers of official statistics, the new Code will be structured differently. The Code will be split into two parts. Part 1 is written with different producers and users in mind, for example non-statistician analysts, or policy makers. It will include a high-level summary of Trustworthiness, Quality and Value and why these pillars are important. The aim is to have a Code on one page – and by doing it in the form of a narrative, it will give people a better understanding of the Code. Part 2 contains the detailed Principles and Practices. This section is aimed largely at those producing official statistics.
We are also developing an interactive version of the new Code for our website, which will include content tailored to different types of users of the Code – again, thinking beyond just producers of official statistics. So, the way that we communicate the information is central to the wider reach of the Code. Finally,we’re spending a lot of time engaging with producers about the new Code and how they could use it in a proportionate way when producing numerical information.
Code consultation
To gather views on the draft Code we’re running a Consultation. It runs until 5th October. I think your conference pack contains a card about it, so I won’t go into too much detail. Just to say that all responses will be very welcome. There is an online survey and a Word version. The consultation pack is available on our website, including the Draft Code 2.0. The final Code 2.0 is likely to be launched early 2018.
1