1 of 19

A Partial History of Partial Randomisation

Tom Stafford, AFIRE programme lead

16 December 2024

t.stafford@sheffield.ac.uk

2 of 19

“Details aside, the basic principle is clear; instead of dodging the fact that chance plays a big part in awarding money, the system will sanctify chance as the determining factor. After a few years, let's look back and evaluate the science that came out of this system.”

Greenberg, D. S. (1998). Chance and grants. The Lancet, 351(9103), 686.

Greenberg, 1998

3 of 19

Avin, S. (2019). Mavericks and lotteries. Studies in History and Philosophy of Science Part A, 76, 13-23.

Brezis, E. S. (2007). Focal randomisation: An optimal mechanism for the evaluation of R&D projects. Science and Public Policy, 34(10), 691-698.

Referees cannot recognise the true value of an invention, λi, and they all give the value λ*.

Uvi = α Di + γ B i + βλ* (4)

Moreover, we assume that, since referees have difficulty in recognising the future value of inventions, they tend to underestimate it, and give to λ* a lower value than the average of λi.

Brezis, 2007

4 of 19

Fang, F. C., & Casadevall, A. (2016). Grant funding: Playing the odds. Science, 352(6282), 158-158.

“Selection of the best of the best resembles a lottery in its unpredictability, but one that lacks the benefit of being truly random, due to bias”.

Fang & Casadevall, 2016

5 of 19

Fang, F. C., & Casadevall, A. (2016). Grant funding: Playing the odds. Science, 352(6282), 158-158.

“Selection of the best of the best resembles a lottery in its unpredictability, but one that lacks the benefit of being truly random, due to bias”.

6 of 19

The Health Research Council of New Zealand is the first major government funding agency to use a lottery to allocate research funding for their Explorer Grant scheme. … This paper presents the results of a survey of Health Research Council applicants from 2013 to 2019.

Health Research Council of New Zealand 2013 -

Liu, M., Choy, V., Clarke, P., Barnett, A., Blakely, T., & Pomeroy, L. (2020). The acceptability of using a lottery to allocate research funding: a survey of applicants. Research integrity and peer review, 5, 1-7.

7 of 19

Woods, H. B., & Wilsdon, J. (2021). Why draw lots? Funder motivations for using partial randomisation to allocate research grants. Research on Research Institute. https://doi.org/10.6084/m9.figshare.17102495.v2.

Research on Research

Partial Randomisation Trials Catalogue

8 of 19

These slides:

bit.ly/tomstafford

9 of 19

These slides:

bit.ly/tomstafford

https://researchonresearch.org/project/a-f-i-r-e/

https://researchonresearch.org/project/a-f-i-r-e/

RoRI’s Accelerator for Funder Experimentation - AFIRE

Partial Randomisation at The British Academy

19 February 2025 00:00 PDT / 08:00 BST / 09:00 CEST

Ken Emond, The British Academy & Adrian Barnett

Capacity building

Forum

Experiments

Sprint on AI/ML in reviewer selection (Feb -May 2025)

Distributed Peer Review, Core Outcome Sets for PR.

10 of 19

END

(reserve slides follow)

11 of 19

Wouldn’t this be gambling with scientists’ careers? ….These criticisms are without merit. The current system already gambles with scientific careers, just in a haphazard way.

Azoulay, P. (2012). Turn the scientific method on ourselves. Nature, 484(7392), 31-32.

12 of 19

AFIRE

  1. Awareness
  2. Motivation
  3. Capacity
  4. Scale and reach
  5. The culture

These slides:

bit.ly/tomstafford

https://researchonresearch.org/project/a-f-i-r-e/

https://researchonresearch.org/project/a-f-i-r-e/

13 of 19

Forum

Support

Experiments!

14 of 19

Experiment: Core Outcome Sets for Partial Randomisation

Sprint #1 : AI in reviewer selection

Nov

Dec

Jan

Feb

Mar

Apr

May

Jun

2024

2025

Sprint #2 : Designing RCTs

Experiment: Distributed Peer Review @ VolkswagenStiftung

Metascience2025

AFIRE Funders’ Forum (monthly)

AFIRE phase 2 experiments announced!

15 of 19

These slides:

bit.ly/tomstafford

Bendiscioli, Sandra; Firpo, Teo; Bravo-Biosca, Albert; Czibor, Eszter; Garfinkel, Michele; Stafford, Tom; et al.

(2022):

The experimental research funder’s handbook (Revised edition, June 2022, ISBN 978-1-7397102-0-0).

https://doi.org/10.6084/m9.figshare.19459328.v2

16 of 19

18 Core Partners

10 codesigned projects

17 of 19

Partial Randomisation at The British Academy

19 February 2025 00:00 PDT / 08:00 BST / 09:00 CEST

Ken Emond, The British Academy & Adrian Barnett

Do Grant Proposal Texts Matter for Funding Decisions? A Field Experiment at NWO

19 March 2025 07:00 PST / 15:00 GMT / 16:00 CET

Müge Şimşek, University of Amsterdam

An experiment with Distributed Peer Review

23 April 2025 00:00 PDT / 08:00 BST / 09:00 CEST

Hanna Denecke, The Volkswagen Foundation

Register for updates. Speaker suggestions welcome!

AFIRE

2024-25 Funder Forum series

18 of 19

AFIRE

2024-25 Funder Forum series

Villum Experiment – a grant scheme with double-blind evaluation

23 October 2024 00:00 PDT / 08:00 BST / 09:00 CEST

Anders Smith, Villum Foundation

Predicting Progress: A Pilot of Expected Utility Forecasting in Science Funding

20 November 2024 07:00 PST / 15:00 GMT / 16:00 CET

Alice Wu, Federation of American Scientists & Jordan Dworkin, Open Philanthropy

Enhancing EDI in application processes in the CES Transformation Fund, including evaluation of an anonymisation-deanonymisation review process

18 December 2024 07:00 PST / 15:00 GMT / 16:00 CET

Rachel Kendal, Principal Investigator CES Transformation Fund

19 of 19

AFIRE Experiments - phase 2

AFIRE exists to support funder use and generation of higher quality evidence, across all topics. Topics where funders can take advantage of active experiments, recent experience and strong interest from other funders interested in running new experiments are below. For more ideas, see The Research Funders Handbook!

Distributed Peer Review

applicants for a funding scheme become peer review college for that scheme

Partial Randomisation

reducing deliberation time and bias in the grey zone

Can AI help find reviewers?

Testing if language models can help match proposals to peer reviewers

Anonymisation

How and where to use anonymisation in proposal evaluation