Considering Public Good in Research and Statistics: Ethics Guidance

Public good project checklist

To help you reflect on the relative risks and benefits of your proposed work and identify key aspects that may be relevant to the public good statement for your project, work through the questions and points in the checklist below.

This checklist is designed to encourage consideration of key issues and provide a basis for discussion and engagement with colleagues directly involved in the project, as well as other interested parties, such as the public, representatives of relevant organisations and data owners.

When completing the checklist, we encourage you to not only think about what the potential benefits of your work may be, but also how likely those benefits are to occur.
This checklist supports the existing UK Statistics Authority’s ethics self-assessment tool and can be used as an additional aid by researchers to consider public good aspects. It is recommended that this checklist be used as early as possible in the project design phase to ensure that a consideration of data ethics is effectively built into the project approach.

Support is also available from the data ethics service and associated committees to assist with this if needed (National Statistician’s Data Ethics Advisory Committee and Research Accreditation Panel).

Download a PDF version of the checklist

Checklist

Potential Benefits

What are the specific envisaged public benefits of your work? And how will you achieve these benefits? (revisit the ‘What is public good’ section if needed)
What is the evidence-base behind your justification of potential benefits? Is it peer-reviewed? How confident are you that these benefits will be realised?
Are there any limitations in your project approach that may limit the impact of potential benefits? What are these and how have they been minimised?
How many people are likely to be affected by the potential benefits arising from the work? And who/what sections of society do they represent? Focusing on a small section of the population is fine, but make sure that you have a clear justification for doing this.
Has the rationale behind, and benefits of, your work been clearly articulated and documented? Is this easily accessible and understandable to relevant groups?
Is the work focused on enhancing trust in statistics or statistics producers (e.g., challenging or validating official statistics)? By what means will it do this?
Is the work addressing a topic that requires urgent or timely data to aid decision-making? If so, what is the rationale for this?
Is the work addressing data gaps in statistics? If so, which ones?
Will your work effectively communicate findings so that public benefit can be maximised across different audiences who may engage with your project results? What communication methods and channels will you use to ensure this?
Does your project approach uphold the principles of trustworthiness, quality and value in statistics? In what way?
Now make sure that you check the key points to remember and what to avoid sections below.

Potential Risks

Is there potential for your work to be used to make decisions about individuals (e.g., as may be the case with predictive modelling projects) or to identify individuals? What ramifications may this have for these individuals?
Is there potential for your work to be used to make decisions about, or to identify, particular groups or communities within society? What ramifications may this have for them?
Are there any potential data gaps in your work that could lead to harm, stigmatisation or distress for individuals or groups who are under-represented in your analysis (i.e., those who may be missing from your data)? What are the impacts on public good and can you justify these? How could this be mitigated?
Is there potential for bias in the data sources and/or data collection methods used? Are any assumptions and constraints in the data and methods clearly identified and communicated?
Is there potential for harm, stigmatisation or distress for individuals or groups who are (a) included as data subjects in your project or (b) may be impacted as a result of the findings of the research (including social, environmental, economic, physical or mental health impacts)? If so, how can these risks be minimised?
Is there potential for negative impacts for organisations who are (a) included as data subjects in your project or (b) may be impacted as a result of the findings of the research (including reputational impacts)? If so, how can these risks be minimised?
Is there potential for harm or distress to members of the research team, research facilitators, or other individuals involved in activities related to conducting the project? If so, how can these risks be minimised?
Does your work involve using data related to sensitive or personal information or protected characteristics, sensitive cultural or social contexts, or engaging with vulnerable groups? How are these aspects being managed?

Key points to remember

Potential harms are not solely related to data collection and primary research, the use of secondary data also has the potential to cause harm!

When considering potential harms, these may not solely relate to impacts on individuals – think about wider social networks and social structures that this may relate to.

A project collecting and/or analysing data in one country or cultural context may present substantially different risks than if conducted in a different country or context, due to differing data governance, cultural sensitivities, or potential harms related to identification of data subjects. It is important to consider the wider societal context that your project relates to and the impact that this may have on potential harms and the sensitivity of data.

When considering public good, it should be remembered that this may be viewed differently by different people. It may be useful to involve subject matter experts or community / group representatives to help understand how public good is viewed from their perspective.

Assessing public good is an inherently subjective process and the relative benefits and risks of any project are likely to differ according to the particular methods and data sources used, the rationale for the work, and the particular context in which the work is undertaken. It is therefore very important to document your process for defining and evidencing the public good of your project.

Ensure that your work is as transparent as possible to help maintain public confidence in the use of data for research and statistics. Reflecting on, and capturing, the extent to which public good impact was achieved at the end of your project, as well as assessing this throughout the project timeline, is an important aspect of transparency.

Final Questions

Are there any other risks and benefits related to this particular project that you can think of that have not been considered? This list is unlikely to be exhaustive!
For each risk identified, think through any possible mitigations that could be applied to minimise this. For example, if researchers are dealing with sensitive or unpleasant information arising from a study on children’s mental health, how can you ensure that the appropriate support services are in place for researchers, as well as participants?
For each benefit identified, think about how likely it is that the benefit will be realised (i.e., the probability of the benefit occurring).
Can you justify all of the remaining risks that are associated with your work? Are these proportionate to the proposed public benefits of the project?

What to avoid when considering public good

Assuming that research is beneficial due to the reputation of the organisation or nature of the work.
Articulating public good in a way that is not understandable to a non-specialist or assuming that the reader will know what certain data or methods are or why research is important, regardless of how common the research approach is considered to be.
Assuming that harm is only applicable to surveys or primary data, i.e., a justification that there is no potential harm because there is no direct interaction with data subjects.
Claiming that there is no bias identified without a justification of why this is the case.
Identifying an issue (e.g., limited public good, limited sample or the identification of potential harm or bias) without providing any mitigations for this.

« Previous

Download PDF version (131.69 KB)