3. How do we currently think official statistics can serve the public good?

This think piece is about creating a shared understanding and collaboratively answering the question how can official statistics serve the public good? This section of our think piece describes the evolution of our overall thoughts on answering the question, and proposes an initial answer.

3.1 How to answer this question

In answering the question ‘how can official statistics serve the public good?’, we focus on what allows official statistics to serve the public good, rather than defining ‘the public good’. We are aiming to collate information and add depth to known definitions rather than replacing existing understandings. Even the decision to address this specific question and from this perspective took time to develop.

The OSR research programme began our exploration Into statistics that serve the public good with a literature review on the topic of the public good of statistics. This literature review notes that the economic definition of a public good is ‘a resource which is nonexcludable, meaning everyone can use it, and non-rivalrous, meaning one person using it does not take it away from anyone else’. Statistics produced by government could widely be considered public goods in that they meet this criteria, however some explorations of this in greater detail (such as Asle Rolland’s paper on the concept and commodity of official statistics or section seven of Steve MacFeely’s paper in search of the data revolution) show there is more nuance that could be debated here. A further (unpublished) consideration in this respect came from discussions with independent researcher Ken Roy, who challenged how closely official statistics align with the colloquial definition of public goods as well (produced by government for citizens). Here, he highlighted that as well as being produced for the public, official statistics are also produced for, and can sometimes prioritise, government use.

If we followed the public goods route, then perhaps ‘statistics serving the public good’ might mean statistics are produced by government, and can be accessed and utilised by everyone. However, we see a distinction between the public good and these public goods, which makes such a definition overly simplistic. The distinction is described in a book sharing perspectives from the Census of India, with authors Abhishek Jain and Varinder Kaur stating ‘Census is both a public good and for public good. Census collects data, but this data is not an end in itself, but a means to promote human welfare’. This statement about censuses can be extrapolated out to official statistics more widely – official statistics themselves may be public goods, but they must benefit society if we are to say they serve the public good.

Noting the distinction between public goods and the public good means we can move away from existing definitions of public goods and focus on understanding the concept at hand. To define the public good, Neil Walker described in a paper on European public good and public goods that one requires ‘a prior sense of who that particular public is; which public, once identified, provides the reference point for what is good for it’. This statement might suggest that to understand statistics and the public good we ought to be identifying who the public is and what good is for them. Doing so would follow the approach of the National Data Guardian’s explanation of what public benefit means for the context of public benefit evaluations, which set out who the public is and what benefits are in this context. In fact, we did explore this when we spoke to members of the public as part of a series of public dialogues. From this we heard that ‘the public’ could be a current or future public, and could include either everyone or a subset of the population. We also heard that the ‘good’ could be wide ranging, including tangible and less tangible, but overall ought to be about meeting real world needs.

We might have explored this further as our primary research question, because knowing who the public is and what the good is would allow us to define ‘the public good’ in the context of statistics. However, in OSR our remit is not to dictate what is ‘good’ or ‘bad’ for the public. Our role is to support public confidence in, and appropriate use of, statistics rather than to direct the ends to which they are used. As such, we chose to remain neutral with respect to conceptions of what is in the public good and instead focus on what the role of statistics may be in achieving it.

When exploring the public good of statistics from this angle, the most obvious piece of evidence comes from legislation. The Statistics and Registration Service Act (2007) states that ‘serving the public good includes in particular informing the public about social and economic matters, and assisting in the development and evaluation of public policy’. This statement begins to explain what it means for statistics to serve the public good. However, it can be difficult to know what it means in practice: the high-level legislative definition still leaves statistics producers and OSR as a regulator with little information on how they can make decisions that support statistics to serve the public good.

This is the space where the present think piece sits – focussing on how official statistics can serve the public good, rather than looking at statistics as ‘public goods’ or stating what is ‘good’ for ‘the public’, and aiming to get to a point where understanding allows for practical decision making. We already provide practical guidance in the form of our Code of Practice for Statistics (the Code), which sets out a framework made up of pillars, principles and practices and states ‘statistics will serve the public good if producers follow the principles and practices set out in the Code’. However, knowing that the public good is being served by following the Code is not the same as understanding how to serve it, in the same way that driving a car is not the same as understanding the underlying mechanics. Although most of the time this deeper understanding may not be needed, in new or complex situations it can be beneficial.

To provide the strongest possible support for our statistical system, we therefore have chosen to focus on the question how can official statistics serve the public good? While our question of interest is specifically about official statistics, much of the content within this think piece could be applied more widely as well; given our remit is tied to official statistics, however, they remain our primary focus.

3.2 An initial answer

The evidence we have identified and discussions we have held to date have informed our thinking and supported the development of an initial understanding on how official statistics can serve the public good. To communicate our thinking in a succinct manner, we have captured our initial answer in a statement:

This statement compliments the Statistics and Registration Service Act (2007), which notes ‘serving the public good includes in particular informing the public about social and economic matters, and assisting in the development and evaluation of public policy’. In this legal definition, ‘informing the public’ relates to being a public asset that provides insight, ‘development of policy’ sits under shaping action, and ‘evaluation of policy’ is one example of statistics informing understanding.

The most substantial difference between our current understanding (reflected in our initial statement) and legislation is that we propose that actions in service of the public good go beyond only the development and evaluation of public policy. For example, during the COVID-19 pandemic businesses and individuals were using statistics to make a variety of decisions that could be seen to benefit society. Despite this difference, we assert that proposing a broader definition in this way does not contradict legislation, as the Act describes the public good as including the specific public policy examples, rather than being limited to only them. In addition, statistics use beyond public policy would still fit with the legal definition, as for use of statistics beyond policy to happen the public must be informed, which the legal definition includes as serving the public good.

In the next section of this think piece, we will set out how we arrived at our current understanding, exploring why we include certain terms in our statement and what we see them as meaning. Before that, however, is the first discussion box of this think piece. These discussion boxes include specific questions we would like to hear your views on.