Responsible modelling - in-person only

Responsible modelling - in-person only

Date: Monday 23 January 2023, 11.00AM
Location: RSS, London
Royal Statistical Society, 12 Errol Street, London EC1Y 8LX
Section Group Meeting


Share this event

Modelling has been abused in many outcomes, to the point that some claim policy would be better off without its excesses. Professionals using models are responsible for redeeming their reputation, and sustaining their value to society.

This event will convene professionals to discuss how that can be done, identifying specific weaknesses, challenges in communication, and the value of codes in support of practice.

All welcome
 
The Blackett Review of Computational Modelling synthesised principles for the use of models in a wide range of public policy applications. Some of these applications are very specialised and technical, while managing significant risks relevant to large parts of society. So professionals have a responsibility to produce well-specified models that meet policy needs while communicating the assumptions, uncertainty and other limitations. More background on the notion of ‘responsible modelling’ is available in a blog post by one of the organisers.
 
Government departments and independent agencies have a particular concern for the policy suitability of their models. Examples of the challenges were seen during the recent pandemic, which showed the need for process robustness. To support the development, assurance and communication of models, a number of guidance documents have been produced. But there is also ongoing academic work on what it means to model real systems, how to extend our assurance processes, and how to get across responsibilities to our students.
 
We will hear perspectives from the National Audit Office, the Office for Statistics Regulation, the Office for Budget Responsibility and other UK agencies about their work on guidance and evaluating models. And we will also hear about some academic projects trying to bring ethics into mathematical work, where its absence is increasingly uncomfortable, but exact aims and concepts are not as clear as sometimes assumed. The speakers will all give examples of applications as well as some technical details, so the event is aimed at a broad technical audience of practitioners.
 
All Models are Normative: considering the value judgements contained in mathematical representations and statistical methods
Erica Thompson (LSE)
 
Applying Pseudo-Ethical Guidelines for Mathematical Practice to Modeling with Data Science
Victor Piercy (Ferris State)
 
How to build an ethical blockchain (or any other mathematical technology)
Maurice Chiodo (Centre for Study of Existential Risk, Cambridge)
 
Uncertainty in OBR economic and fiscal forecasts
James Ebdon (Office for Budget Responsibility)
 
What makes a super model? Lessons from the NAO’s work auditing government models
Ruth Kelly (National Audit Office)
 
A model’s journey to Trustworthiness, Quality and Value
Emily Barrington (Office for Statistics Regulation)
 
Other contributors TBC
 
Thomas King for RSS Data Ethics & Governance Section

Job de Roij for RSS Data Ethics & Governance Section

 
 
RSS Concessionary Fellows £15
RSS Professionally Qualified Fellows £20
RSS Fellows £25
Non-Fellows £50