An Official Statistics Reading List (2024)
?
1) Introduction
?
The purpose of this article is to share signposts to some specific pieces of writing from 2024 that I have found interesting and useful – in building my understanding and in contributing to my research related to the challenges and choices faced in the world of Official Statistics.
As for the 2023 Reading List, I have grouped the sources under the headings of (Official Statistics) Production, Communication, Use and Governance. The only format change this year is that I have created a new category for ‘favourite’ Corporate Documents (not a category historically seen on many end of year awards lists).
Also as before, the articles / papers included are (at least currently) available on an open-source basis and are constrained to those available in the English language.
??
2) Production
?
There are a lot of sources to choose from when looking at material on the various aspects of Official Statistics production. In my scanning I have tended to focus on sources that touch on broader strategic themes rather than, for example on detailed technical or methodological issues.
?
The issue of reducing survey response rates is much discussed in the world of Official Statistics and, from a UK context, has gained some level of public profile through some well documented challenges faced by the UK Labour Force Survey. I found this paper useful in providing a broader perspective on the problem. It analyses four major cross-Europe surveys over a 20-year period, seeking to assess whether there are specific factors that seem to influence the scale of reduction in survey response rates. The sobering conclusions confirm a consistent trend of reductions in survey response rates and offer little evidence that different technical strategies (in terms of different sampling frames or data collection modes) offer much ‘hope’ in terms of confidently constraining that trend.
?
We seem to be moving to a multi-source world for many Official Statistics, both for demand reasons (more complicated questions being asked) and for supply reasons (phenomena such as survey response rate reduction mean that complementary sources are often needed to get to a viable answer). One of the methodological implications of dealing with this multi-source (and multi-owner) context is the need to think very carefully about data privacy issues. This really useful paper gives a sense of the challenge and of some of the potential technical responses, describing work underway within the European Statistical System (ESS) on a potential way forward in the form of a shared system for ESS members that might securely embed Privacy Enhancing Technologies.
?
As this report notes, there is currently “a lack of harmonised, global and comprehensive statistics to properly measure the carbon content of economic activities”. This report provides a stock-take on the current state of play (for example in terms of different forms of voluntary and mandatory reporting) and helpfully sets out the key principles that need to underpin an effective carbon accounting system. There is honesty about the scale and complexity of the challenge and some optimism, for example in the potential of technologies such as remote sensing to help address measurement and consistency issues. Official Statisticians are clearly flagged as having a central role in this important endeavour.
?
It would be difficult to talk about recent developments in the world of Official Statistics without touching on the joys of Artificial Intelligence (AI). Clearly, anything written today will be at risk of being out of date tomorrow – but this report is, I think, a useful staging post in capturing a sense of where we are / were (as of mid 2024). The report captures the results of a survey of members of the Conference of European Statisticians (CES). As always there are a range of experiences and opinions but a clear overall sense of the likely breadth of opportunity / impact – for example in production activities such as data collection and processing - and of the areas of risk (including confidentiality and accuracy / validation mechanisms) and of challenge (skills).
?
3) Communication
?
In contrast to the topic of production, there is, traditionally, a less active stream of new material specifically focused on the communication of Official Statistics, whether touching on specific solutions / technologies, strategic choices or outcomes. Hence some of the helpful material below comes from broader examinations of the fast-changing communications landscape.
??
This paper offers an insightful exploration of what is involved in communicating uncertainty to audiences / potential users. Uniquely the paper looks at this question from both the production side (what graphical and other techniques might represent best practice) and the user side (what are the different factors that influence the way in which users process information about uncertainty).
Strategic Communications During the Inflation Crisis (UNECE HLG-MOS: Contributors - Emily Liddel, Janice Keenan, Maulana Faris, Aeidin Sheppard, Giulia Peci, Michela Troia, Margarida Rosa, Lukasz Augustyniak, Colin Stewart, Dean Vuletic, InKyung Choi, Stela Derivolcov, Tetyana Kolomiyets)
There are instances when regular (‘normal’) Official Statistics quickly acquire added prominence, and scrutiny. This was the case during the Covid pandemic and, more recently, during the period of increased inflation that started around mid 2021. This useful document takes stock of the experience (across many countries) of communicating inflation statistics during this period and offers some helpful lessons – for example the importance of tracking feedback and of investing in a range of formats as and when new users (and user needs) start to emerge. As the document outlines, in particular through five helpful case studies, these periods of enhanced focus do bring risks but also bring opportunities to bring value from Official Statistics to an expanded number of people.
?
Those seeking to communicate Official Statistics now face a range of options in terms of the most appropriate format (or formats) to use to wrap or present the core information. Traditional approaches, notably the print-like pdf report, have been augmented, or in some cases replaced, by dashboards, APIs and videos. This paper explores a topic that has similar characteristics to the communication of Official Statistics, in looking at the potential impact of different approaches to the communication of central bank inflation expectations. The paper seeks, amongst other things, to assess the different impacts of using ‘static’ mediums (for example written text) and more ‘dynamic’ mediums (audio and video clips) – and in addition explores which approaches might have most impact on different population groups. Spoiler alert – the medium matters and, if the conclusions of this exercise are at some level transferable to Official Statistics, then it might argue for increased investment in audio/video outputs (in particular to connect to audiences that are less pre-informed).
?
This paper explores the developing trend of personalisation in the way that (digital) news is provided to users. The perspective, as suggested by the title, is a very much a journalistic one. The Official Statistics interest, I think, is twofold. Firstly, there is the basic need to understand this trend as it will impact on the extent to which stories based on Official Statistics might appear in front of (different) users. Secondly it raises interesting questions for the direct channels that are used to communicate Official Statistics – notably National Statistical Office (NSO) websites. Is there a way that personalisation approaches can and should be used in that context – for example, when you as a user visit your NSO’s website, some of the information that you are offered could reflect what you were interested in during your last visit? As this paper suggests, navigating such choices would require consideration of the trade-offs between the benefits in terms of potentially enhanced reach and engagement and the wider public interest (or public good) considerations.
?
4) Use
?
It is good to see an increased interest in the use (or potential use) and the misuse of Official Statistics. This feels essential in building a knowledge base that can be applied in seeking to increase the value that Official Statistics can generate.
?
领英推荐
Policymakers have always been seen as an important (if not the central) set of users for Official Statistics. This paper notes the “step change in the availability of data and methods of analysis” and asserts that “It is no longer acceptable to rely on custom-built statistics and consultants’ reports to inform high level policy discussions.” There are, as with any paper with this scope and ambition, questions that could be posed about some of the assumptions. Nonetheless it does present some important challenges which, I think, merit close reflection – not least about the future distinctiveness of the ‘offer’ to policymakers from Official Statistics (and how this offer may differ from that of other data-driven resources).
?
This paper develops a proposed typology of the different ways in which the use or the value of data in organisations is described or conceptualised. I thought that the final four themes proposed offered an elegant framing of different types of value that might be equally applicable in the related world of Official Statistics – perhaps playing both into the ways that we develop narratives about the value of Official Statistics and the ways in which we seek to improve the value that is delivered. I was particularly interested in the concept of data as a ‘practice’ which relates to its role and value in terms of enabling innovation and learning – which is more about continuous improvement within an organisation (or perhaps within a society) than about discrete decision-making.
?
Moving from use to (potential) misuse, this paper draws on different disciplines (including philosophy) to help the UK’s statistics regulator evolve its guidance and practice – in terms of its role to challenge misuse of statistics. Specifically, this paper offers a view of what it means for use of statistics / data to be misleading.
?
Still in the land of misuse, this paper analyses the potential consequences of using different approaches to combat misinformation. The paper suggests that specific actions (for example fact checking) may well have beneficial impacts – but may also lead to wider, unintended consequences. Building on this analysis, the paper suggests some alternative / adjusted strategies that might be trialled – which may be of interest to those in the world of Official Statistics (from National Statisticians through to regulatory bodies such as OSR – see above) who need to make decisions on when and how to effectively respond to potential examples of misinformation.
?
5) Governance
?
These papers / articles move beyond the day-to-day production, communication and use of Official Statistics and look at wider questions (and choices) about how Official Statistics systems are designed and operated.
?
This paper forms the first chapter of an academic book on governance. I found it really useful in providing wider context to the current challenges and choices faced by Official Statistics in the 21st Century. Cosmo’s paper provides an accessible overview of the way that, over recent centuries, Official Statistics systems have evolved in parallel to the evolution of states. The paper then summarises an intriguing framework (looking at ideas, interests, institutions and identities) that might help characterise the four ‘eras’ of Official Statistics – using the example of the Australian Official Statistics system to map the journey (to date) between these different stages.
?
From my occasional skimming of corporate documents produced by NSOs and others, I am conscious the narratives used about the value of Official Statistics often touch on the contribution to a healthy democracy. That association, between democracy and Official Statistics, is also one that is referenced in a range of international documents including the Fundamental Principles of Official Statistics. Luca’s paper does something that I do not believe has been attempted before which is to attempt a high-level global analysis of the relationship between the different indices used to assess / score countries in terms of both the level (or quality) of democracy and the quality of the national Official Statistics system. As Luca notes, the noted correlation does not necessarily signify causation, and there is much further work to be done on this important topic.
?
Although this paper does not touch directly on the world of Official Statistics, it does cover a topic of strong relevance to the design of Official Statistics systems. Perceptions (and realities) of the trustworthiness of governments are one reason why so much emphasis is placed on institutional arrangements that establish some level of separation of Official Statistics organisations from government influence. Nonetheless previous analytical work by STATEC suggest that there is some association between tendencies to trust governments and to trust Official Statistics. The specific concept that I found of most interest in this paper was the concept of ‘benevolence’ (the sense that governments are trying to act in citizens’ interests) and the suggestion that this might be one of the most important attributes driving levels of trust.? The question I was left with was whether, if asked, members of the public would see Official Statistics organisations as benevolent.
?
Finally, I wanted to include the latest independent review of the primary Official Statistics organisation in the UK. Obviously, there is a lot of insightful analysis and recommendations focused on the specific characteristics (and many idiosyncrasies) of the UK’s Official Statistics System. I thought, however, it was worth including here as it touches on one specific governance issue of potential wider interest. At the centre of many governance discussions is the question of ‘who decides’ – and the Lievesley Review touches on the question of who decides about what the priorities are – notably the priorities for which Official Statistics are produced. This takes us to how the Official Statistics System delivers (and demonstrates) the right balance between the needs of governmental users and the needs of citizens – and the Review offers an innovative solution which will be tested over the coming year.
?
6) Corporate Documents
?
As touched on elsewhere, I have a specific interest in the corporate documents produced by governments or individual public bodies (often NSOs) concerned with Official Statistics. I think that they can be an interesting, and perhaps an untapped, resource to track the challenges and opportunities that Official Statistics organisations are facing and to note the choices that they are making.
I have therefore included some examples below that I thought were helpful in signalling the context, the ambitions and the choices for Official Statistics in different countries.
?
On a presentational level, I did want to include this one to recognise the innovation (and elegance) in the format that the CSO have used for a few years now. In addition, I think that there are some helpful signals about the different choices being made by the CSO, not least in the way that they seek to live up to the organisational vision (“Independent Insight for All”) in the development of more citizen-focused products. In the context of a wider debate for many NSOs, it is also interesting to see that one of the four Strategic Priorities relates to the CSO’s growing role as the (or ‘a’) Data Steward within the wider Irish Civil and Public Service.
?
Norway is a country that has updated / modernised its legislation regarding Official Statistics. One requirement of the 2019 Act is that Statistics Norway publishes an annual report on the quality of official statistics. This is the most recent example. One specific element of this report that I found both interesting and compelling is the part of the assessment process that involves the completion of a questionnaire by those involved in, or close to, the production of each of the country’s Official Statistics products. This breadth of regular analysis (requiring over 300 responses in the latest exercise) feels helpful in maintaining commitment to quality at the operational level and in establishing a richer picture of the diversity of experience that inevitably exists when producing an Official Statistics portfolio of any scale.
?
Most, if not all, NSOs are currently engaged in exploring the potential and the risks of the emerging family of Artificial Intelligence (AI) technologies. That work will, I suspect, continue apace as the scope and the ambition of new AI tools continues to expand. Noting the centrality of trust and transparency to so much of Official Statistics work, one essential element of the ongoing development activity is to continue to clearly signal to users and stakeholders (and staff) what is being done and what guardrails are being defined to manage risk. This policy statement from Statistics Sweden does, I think, offer a good example of how to do this – with clarity and with a structure that will enable ongoing development of the policy as the list of possibilities and the risks evolves. It is also noteworthy (and commendable) that this specific iteration of Statistics Sweden’s AI policy has an explicit and short lifespan (one year).
?
?End
Professor, Oxford Internet Institute, University of Oxford; Programme Director for Public Policy, Alan Turing Institute;
2 个月Proud to have a mention with Omar A Guerrero on the Official Statistics reading list, thank you!
Innovation Diplomacy and Management; Senior Scientist, Pravasi Bharatiya Samman, Fellow America Physical Society
2 个月ROHIT BHARDWAJ you should enjoy the holidays. Reading can wait but this time with your family and kids will not come again.
Director General, Office for Statistics Regulation
2 个月Great stuff! Now I know what to do in those quiet days between Christmas and New Year.?
Based in Scotland, Trustee of Loch Lomond & the Trossachs Countryside Trust
2 个月The Ken Roy annual end of year reading list is much anticipated and becoming almost as famous as Barack Obamas summer music list. As the Gen Zs would say it’s “dank” !