Date: Tuesday 5th May 2026 Time: 14:00 - 15:00 GMT | 9:00 - 10:00 EST (US) | 15:00 - 16:00 CET Location: Online via Zoom Speakers: Nick Galwey, GSK (Retired)
Who is this event intended for?:
Statisticians interested in learning about False Discovery Rate and Shrunk Estimates
What is the benefit of attending?:
To gain a deeper understanding of tools used to address multiplicity, explore how these methods relate to one another, and see how they perform when applied to both real human gene‑expression data and simulated datasets.
Brief event overview:
This talk will explore the “replication crisis” in science, focusing on how testing large numbers of hypotheses can lead to false positive findings. It introduces key statistical approaches—False Discovery Rate (FDR) and shrinkage methods—to address this issue, and explains their conceptual foundations and connections. The session will also highlight how these tools can be understood within an empirical-Bayesian framework, linking significance testing with effect size estimation.
Cost:
This webinar is free to both Members of PSI and Non-Members.
Science is currently facing a ‘replication crisis’ – a concern that many scientific findings reported are difficult or impossible to reproduce. A major cause of this is the availability of technology that permits the exploration and testing of very large numbers of hypotheses, some of which will almost certainly show significant or large effects by chance, even when no real effects are present: this is the ‘multiplicity’ or ‘multiple testing’ problem. The statistical tools available to address this problem include:
the False Discovery Rate (FDR), which is specified in relation to the subset of the m hypotheses tested for which the discovery of an effect is reported, and which indicates the proportion of these ‘discoveries’ that is expected to be false; and
shrunk estimates, which reduce the estimated effect, in relation to every individual hypothesis, from the observed value towards the null value.
This talk will first examine the conceptual basis for each of these tools, then consider how they are connected. Though the FDR and shrunk estimates are both conventionally presented in the frequentist statistical framework, they can both also be presented in empirical-Bayesian terms, the prior probability distribution being calculated from the data relating to the m hypotheses tested, as follows:
in the case of the FDR, from the proportion of the m significance tests conducted that give a p-value at or below the specified significance threshold, and that are therefore announced as ‘discoveries’; and
in the case of shrunk estimates, from the distribution of the observed effect sizes over the m hypotheses.
Based on this connection, a formal relationship between FDR values and shrunk estimates will be presented, and it will be argued that these two tools can profitably be used in conjunction. Their combined application, both to real (human gene expression) data and to simulated data, will be illustrated.
Speaker details
Speaker
Biography
Nick Galwey, Former Statistics Leader, GSK (retired)
Nick was a demonstrator and lecturer in biometry and plant breeding at the University of Cambridge from 1979 to 1995, when he was appointed to a senior lectureship at the University of Western Australia, Perth. In 2001 he returned to the UK, working as a statistical geneticist at Oxagen Limited, Oxfordshire, until 2004. He then moved to GlaxoSmithKline, first as a statistical geneticist and later as a statistician, supporting pre clinical and clinical research and epidemiology, until he retired in September 2024. He is the author, or a co-author, of more than 100 scientific papers. His most recent book is:
Galwey, N.W. (2025) The False Discovery Rate: Its Meaning, Interpretation and Application in Data Science. Chichester, UK: Wiley. 266pp. ISBN 9781119889779
Scientific Meetings
Connecting the False Discovery Rate to Shrunk Estimates
Date: Tuesday 5th May 2026 Time: 14:00 - 15:00 GMT | 9:00 - 10:00 EST (US) | 15:00 - 16:00 CET Location: Online via Zoom Speakers: Nick Galwey, GSK (Retired)
Who is this event intended for?:
Statisticians interested in learning about False Discovery Rate and Shrunk Estimates
What is the benefit of attending?:
To gain a deeper understanding of tools used to address multiplicity, explore how these methods relate to one another, and see how they perform when applied to both real human gene‑expression data and simulated datasets.
Brief event overview:
This talk will explore the “replication crisis” in science, focusing on how testing large numbers of hypotheses can lead to false positive findings. It introduces key statistical approaches—False Discovery Rate (FDR) and shrinkage methods—to address this issue, and explains their conceptual foundations and connections. The session will also highlight how these tools can be understood within an empirical-Bayesian framework, linking significance testing with effect size estimation.
Cost:
This webinar is free to both Members of PSI and Non-Members.
Science is currently facing a ‘replication crisis’ – a concern that many scientific findings reported are difficult or impossible to reproduce. A major cause of this is the availability of technology that permits the exploration and testing of very large numbers of hypotheses, some of which will almost certainly show significant or large effects by chance, even when no real effects are present: this is the ‘multiplicity’ or ‘multiple testing’ problem. The statistical tools available to address this problem include:
the False Discovery Rate (FDR), which is specified in relation to the subset of the m hypotheses tested for which the discovery of an effect is reported, and which indicates the proportion of these ‘discoveries’ that is expected to be false; and
shrunk estimates, which reduce the estimated effect, in relation to every individual hypothesis, from the observed value towards the null value.
This talk will first examine the conceptual basis for each of these tools, then consider how they are connected. Though the FDR and shrunk estimates are both conventionally presented in the frequentist statistical framework, they can both also be presented in empirical-Bayesian terms, the prior probability distribution being calculated from the data relating to the m hypotheses tested, as follows:
in the case of the FDR, from the proportion of the m significance tests conducted that give a p-value at or below the specified significance threshold, and that are therefore announced as ‘discoveries’; and
in the case of shrunk estimates, from the distribution of the observed effect sizes over the m hypotheses.
Based on this connection, a formal relationship between FDR values and shrunk estimates will be presented, and it will be argued that these two tools can profitably be used in conjunction. Their combined application, both to real (human gene expression) data and to simulated data, will be illustrated.
Speaker details
Speaker
Biography
Nick Galwey, Former Statistics Leader, GSK (retired)
Nick was a demonstrator and lecturer in biometry and plant breeding at the University of Cambridge from 1979 to 1995, when he was appointed to a senior lectureship at the University of Western Australia, Perth. In 2001 he returned to the UK, working as a statistical geneticist at Oxagen Limited, Oxfordshire, until 2004. He then moved to GlaxoSmithKline, first as a statistical geneticist and later as a statistician, supporting pre clinical and clinical research and epidemiology, until he retired in September 2024. He is the author, or a co-author, of more than 100 scientific papers. His most recent book is:
Galwey, N.W. (2025) The False Discovery Rate: Its Meaning, Interpretation and Application in Data Science. Chichester, UK: Wiley. 266pp. ISBN 9781119889779
Training Courses
Connecting the False Discovery Rate to Shrunk Estimates
Date: Tuesday 5th May 2026 Time: 14:00 - 15:00 GMT | 9:00 - 10:00 EST (US) | 15:00 - 16:00 CET Location: Online via Zoom Speakers: Nick Galwey, GSK (Retired)
Who is this event intended for?:
Statisticians interested in learning about False Discovery Rate and Shrunk Estimates
What is the benefit of attending?:
To gain a deeper understanding of tools used to address multiplicity, explore how these methods relate to one another, and see how they perform when applied to both real human gene‑expression data and simulated datasets.
Brief event overview:
This talk will explore the “replication crisis” in science, focusing on how testing large numbers of hypotheses can lead to false positive findings. It introduces key statistical approaches—False Discovery Rate (FDR) and shrinkage methods—to address this issue, and explains their conceptual foundations and connections. The session will also highlight how these tools can be understood within an empirical-Bayesian framework, linking significance testing with effect size estimation.
Cost:
This webinar is free to both Members of PSI and Non-Members.
Science is currently facing a ‘replication crisis’ – a concern that many scientific findings reported are difficult or impossible to reproduce. A major cause of this is the availability of technology that permits the exploration and testing of very large numbers of hypotheses, some of which will almost certainly show significant or large effects by chance, even when no real effects are present: this is the ‘multiplicity’ or ‘multiple testing’ problem. The statistical tools available to address this problem include:
the False Discovery Rate (FDR), which is specified in relation to the subset of the m hypotheses tested for which the discovery of an effect is reported, and which indicates the proportion of these ‘discoveries’ that is expected to be false; and
shrunk estimates, which reduce the estimated effect, in relation to every individual hypothesis, from the observed value towards the null value.
This talk will first examine the conceptual basis for each of these tools, then consider how they are connected. Though the FDR and shrunk estimates are both conventionally presented in the frequentist statistical framework, they can both also be presented in empirical-Bayesian terms, the prior probability distribution being calculated from the data relating to the m hypotheses tested, as follows:
in the case of the FDR, from the proportion of the m significance tests conducted that give a p-value at or below the specified significance threshold, and that are therefore announced as ‘discoveries’; and
in the case of shrunk estimates, from the distribution of the observed effect sizes over the m hypotheses.
Based on this connection, a formal relationship between FDR values and shrunk estimates will be presented, and it will be argued that these two tools can profitably be used in conjunction. Their combined application, both to real (human gene expression) data and to simulated data, will be illustrated.
Speaker details
Speaker
Biography
Nick Galwey, Former Statistics Leader, GSK (retired)
Nick was a demonstrator and lecturer in biometry and plant breeding at the University of Cambridge from 1979 to 1995, when he was appointed to a senior lectureship at the University of Western Australia, Perth. In 2001 he returned to the UK, working as a statistical geneticist at Oxagen Limited, Oxfordshire, until 2004. He then moved to GlaxoSmithKline, first as a statistical geneticist and later as a statistician, supporting pre clinical and clinical research and epidemiology, until he retired in September 2024. He is the author, or a co-author, of more than 100 scientific papers. His most recent book is:
Galwey, N.W. (2025) The False Discovery Rate: Its Meaning, Interpretation and Application in Data Science. Chichester, UK: Wiley. 266pp. ISBN 9781119889779
Journal Club
Connecting the False Discovery Rate to Shrunk Estimates
Date: Tuesday 5th May 2026 Time: 14:00 - 15:00 GMT | 9:00 - 10:00 EST (US) | 15:00 - 16:00 CET Location: Online via Zoom Speakers: Nick Galwey, GSK (Retired)
Who is this event intended for?:
Statisticians interested in learning about False Discovery Rate and Shrunk Estimates
What is the benefit of attending?:
To gain a deeper understanding of tools used to address multiplicity, explore how these methods relate to one another, and see how they perform when applied to both real human gene‑expression data and simulated datasets.
Brief event overview:
This talk will explore the “replication crisis” in science, focusing on how testing large numbers of hypotheses can lead to false positive findings. It introduces key statistical approaches—False Discovery Rate (FDR) and shrinkage methods—to address this issue, and explains their conceptual foundations and connections. The session will also highlight how these tools can be understood within an empirical-Bayesian framework, linking significance testing with effect size estimation.
Cost:
This webinar is free to both Members of PSI and Non-Members.
Science is currently facing a ‘replication crisis’ – a concern that many scientific findings reported are difficult or impossible to reproduce. A major cause of this is the availability of technology that permits the exploration and testing of very large numbers of hypotheses, some of which will almost certainly show significant or large effects by chance, even when no real effects are present: this is the ‘multiplicity’ or ‘multiple testing’ problem. The statistical tools available to address this problem include:
the False Discovery Rate (FDR), which is specified in relation to the subset of the m hypotheses tested for which the discovery of an effect is reported, and which indicates the proportion of these ‘discoveries’ that is expected to be false; and
shrunk estimates, which reduce the estimated effect, in relation to every individual hypothesis, from the observed value towards the null value.
This talk will first examine the conceptual basis for each of these tools, then consider how they are connected. Though the FDR and shrunk estimates are both conventionally presented in the frequentist statistical framework, they can both also be presented in empirical-Bayesian terms, the prior probability distribution being calculated from the data relating to the m hypotheses tested, as follows:
in the case of the FDR, from the proportion of the m significance tests conducted that give a p-value at or below the specified significance threshold, and that are therefore announced as ‘discoveries’; and
in the case of shrunk estimates, from the distribution of the observed effect sizes over the m hypotheses.
Based on this connection, a formal relationship between FDR values and shrunk estimates will be presented, and it will be argued that these two tools can profitably be used in conjunction. Their combined application, both to real (human gene expression) data and to simulated data, will be illustrated.
Speaker details
Speaker
Biography
Nick Galwey, Former Statistics Leader, GSK (retired)
Nick was a demonstrator and lecturer in biometry and plant breeding at the University of Cambridge from 1979 to 1995, when he was appointed to a senior lectureship at the University of Western Australia, Perth. In 2001 he returned to the UK, working as a statistical geneticist at Oxagen Limited, Oxfordshire, until 2004. He then moved to GlaxoSmithKline, first as a statistical geneticist and later as a statistician, supporting pre clinical and clinical research and epidemiology, until he retired in September 2024. He is the author, or a co-author, of more than 100 scientific papers. His most recent book is:
Galwey, N.W. (2025) The False Discovery Rate: Its Meaning, Interpretation and Application in Data Science. Chichester, UK: Wiley. 266pp. ISBN 9781119889779
Webinars
Connecting the False Discovery Rate to Shrunk Estimates
Date: Tuesday 5th May 2026 Time: 14:00 - 15:00 GMT | 9:00 - 10:00 EST (US) | 15:00 - 16:00 CET Location: Online via Zoom Speakers: Nick Galwey, GSK (Retired)
Who is this event intended for?:
Statisticians interested in learning about False Discovery Rate and Shrunk Estimates
What is the benefit of attending?:
To gain a deeper understanding of tools used to address multiplicity, explore how these methods relate to one another, and see how they perform when applied to both real human gene‑expression data and simulated datasets.
Brief event overview:
This talk will explore the “replication crisis” in science, focusing on how testing large numbers of hypotheses can lead to false positive findings. It introduces key statistical approaches—False Discovery Rate (FDR) and shrinkage methods—to address this issue, and explains their conceptual foundations and connections. The session will also highlight how these tools can be understood within an empirical-Bayesian framework, linking significance testing with effect size estimation.
Cost:
This webinar is free to both Members of PSI and Non-Members.
Science is currently facing a ‘replication crisis’ – a concern that many scientific findings reported are difficult or impossible to reproduce. A major cause of this is the availability of technology that permits the exploration and testing of very large numbers of hypotheses, some of which will almost certainly show significant or large effects by chance, even when no real effects are present: this is the ‘multiplicity’ or ‘multiple testing’ problem. The statistical tools available to address this problem include:
the False Discovery Rate (FDR), which is specified in relation to the subset of the m hypotheses tested for which the discovery of an effect is reported, and which indicates the proportion of these ‘discoveries’ that is expected to be false; and
shrunk estimates, which reduce the estimated effect, in relation to every individual hypothesis, from the observed value towards the null value.
This talk will first examine the conceptual basis for each of these tools, then consider how they are connected. Though the FDR and shrunk estimates are both conventionally presented in the frequentist statistical framework, they can both also be presented in empirical-Bayesian terms, the prior probability distribution being calculated from the data relating to the m hypotheses tested, as follows:
in the case of the FDR, from the proportion of the m significance tests conducted that give a p-value at or below the specified significance threshold, and that are therefore announced as ‘discoveries’; and
in the case of shrunk estimates, from the distribution of the observed effect sizes over the m hypotheses.
Based on this connection, a formal relationship between FDR values and shrunk estimates will be presented, and it will be argued that these two tools can profitably be used in conjunction. Their combined application, both to real (human gene expression) data and to simulated data, will be illustrated.
Speaker details
Speaker
Biography
Nick Galwey, Former Statistics Leader, GSK (retired)
Nick was a demonstrator and lecturer in biometry and plant breeding at the University of Cambridge from 1979 to 1995, when he was appointed to a senior lectureship at the University of Western Australia, Perth. In 2001 he returned to the UK, working as a statistical geneticist at Oxagen Limited, Oxfordshire, until 2004. He then moved to GlaxoSmithKline, first as a statistical geneticist and later as a statistician, supporting pre clinical and clinical research and epidemiology, until he retired in September 2024. He is the author, or a co-author, of more than 100 scientific papers. His most recent book is:
Galwey, N.W. (2025) The False Discovery Rate: Its Meaning, Interpretation and Application in Data Science. Chichester, UK: Wiley. 266pp. ISBN 9781119889779
Careers Meetings
Connecting the False Discovery Rate to Shrunk Estimates
Date: Tuesday 5th May 2026 Time: 14:00 - 15:00 GMT | 9:00 - 10:00 EST (US) | 15:00 - 16:00 CET Location: Online via Zoom Speakers: Nick Galwey, GSK (Retired)
Who is this event intended for?:
Statisticians interested in learning about False Discovery Rate and Shrunk Estimates
What is the benefit of attending?:
To gain a deeper understanding of tools used to address multiplicity, explore how these methods relate to one another, and see how they perform when applied to both real human gene‑expression data and simulated datasets.
Brief event overview:
This talk will explore the “replication crisis” in science, focusing on how testing large numbers of hypotheses can lead to false positive findings. It introduces key statistical approaches—False Discovery Rate (FDR) and shrinkage methods—to address this issue, and explains their conceptual foundations and connections. The session will also highlight how these tools can be understood within an empirical-Bayesian framework, linking significance testing with effect size estimation.
Cost:
This webinar is free to both Members of PSI and Non-Members.
Science is currently facing a ‘replication crisis’ – a concern that many scientific findings reported are difficult or impossible to reproduce. A major cause of this is the availability of technology that permits the exploration and testing of very large numbers of hypotheses, some of which will almost certainly show significant or large effects by chance, even when no real effects are present: this is the ‘multiplicity’ or ‘multiple testing’ problem. The statistical tools available to address this problem include:
the False Discovery Rate (FDR), which is specified in relation to the subset of the m hypotheses tested for which the discovery of an effect is reported, and which indicates the proportion of these ‘discoveries’ that is expected to be false; and
shrunk estimates, which reduce the estimated effect, in relation to every individual hypothesis, from the observed value towards the null value.
This talk will first examine the conceptual basis for each of these tools, then consider how they are connected. Though the FDR and shrunk estimates are both conventionally presented in the frequentist statistical framework, they can both also be presented in empirical-Bayesian terms, the prior probability distribution being calculated from the data relating to the m hypotheses tested, as follows:
in the case of the FDR, from the proportion of the m significance tests conducted that give a p-value at or below the specified significance threshold, and that are therefore announced as ‘discoveries’; and
in the case of shrunk estimates, from the distribution of the observed effect sizes over the m hypotheses.
Based on this connection, a formal relationship between FDR values and shrunk estimates will be presented, and it will be argued that these two tools can profitably be used in conjunction. Their combined application, both to real (human gene expression) data and to simulated data, will be illustrated.
Speaker details
Speaker
Biography
Nick Galwey, Former Statistics Leader, GSK (retired)
Nick was a demonstrator and lecturer in biometry and plant breeding at the University of Cambridge from 1979 to 1995, when he was appointed to a senior lectureship at the University of Western Australia, Perth. In 2001 he returned to the UK, working as a statistical geneticist at Oxagen Limited, Oxfordshire, until 2004. He then moved to GlaxoSmithKline, first as a statistical geneticist and later as a statistician, supporting pre clinical and clinical research and epidemiology, until he retired in September 2024. He is the author, or a co-author, of more than 100 scientific papers. His most recent book is:
Galwey, N.W. (2025) The False Discovery Rate: Its Meaning, Interpretation and Application in Data Science. Chichester, UK: Wiley. 266pp. ISBN 9781119889779
Upcoming Events
PSI Book Club: Change: How organisations achieve hard-to-imagine results in uncertain and volatile times
Organizations have to adapt to the transforming landscape of our industry to ensure they continue to be successful in the future. Many of us are feeling the impact of organizational change. By reading John P Kotter’s book we can understand about organizational change and learn how to thrive, rather than just survive, through change.
Change, by John P Kotter (and his team), is a summary of all that he has learned over his decades of research and leading change. His book describes why many current approaches to change are inadequate and explains why new solutions need to give people a voice and a role in a new, change-embracing organization.
Develop your understanding of organisational change and become empowered to be part of your organisation’s change, by reading Change by John P Kotter and joining the Sept-Dec 2025 book club. You will be invited to join facilitated discussions of the concepts and ideas and apply knowledge from the book in-between sessions.
Our monthly webinar series allows attendees to gain practical knowledge and skills in open-source coding and tools, with a focus on applications in the pharmaceutical industry.
This networking event is aimed at statisticians that are new to the pharmaceutical industry who wish to meet colleagues from different companies and backgrounds.
PSI Book Club: Another Door Opens – Book Club Special Event
This is a Book Club Special Event in response to the changes in our industry and as a supportive move to create community and connection for those navigating redundancy and uncertainty. Read the book in advance of the book club session then join the zoom call to discuss ideas. There will be breakout groups to connect with others, exchange experiences of how the book has helped, and offer support.
Joint PSI/EFSPI Visualisation SIG 'Wonderful Wednesday' Webinars
Our monthly webinar explores examples of innovative data visualisations relevant to our day to day work. Each month a new dataset is provided from a clinical trial or other relevant example, and participants are invited to submit a graphic that communicates interesting and relevant characteristics of the data.
PSI Training Course: Propensity Scores: Practical Application in Non-randomised Studies
The course will introduce the topic of propensity scores and the use of external data. Covering the topics of matching and weighting as well as more advance topics of high dimension propensity scores, multi-valued treatments, double robustness and time-varying scenarios. There will be the opportunity to participate in some hands on practical exercises in R.
Making Estimands Work - Practical Approaches with Implementation in Data Standards
This event will discuss the impact of the estimand framework on data flow and data handling for analysis datasets (ADaM). What are the considerations to produce data sets suitable for estimation of estimands and traceability of the occurrence of intercurrent events. This webinar will include tips for best practice, based on the White paper Implementation of ICH E9(R1) Estimands Framework using Data Standards.
PSI/EFSPI Biomarker SIG Webinar: Statistical Methods for ctDNA Data
This webinar will describe what ctDNA is and what the key challenges are for analysing it. We will then summarise and review methods that were published over the last few years for analysing ctDNA data.
This two-afternoon virtual course provides a practical introduction to adaptive clinical trials, focusing on the concepts, applications, and regulatory principles outlined in ICH E20 through real-world examples and case studies.
PSI Careers - MEDMathS: Medicine Empowered by Data, Maths and Statistics
Date: Wednesday 4th November 2026
A careers talk about medical statistics and how it plays a crucial role in developing new medicines. Learn about the field of medical statistics and how it plays a crucial role in developing groundbreaking new medicines, vaccines and healthcare products.
Date: 18 November 2026
This is an excellent opportunity for students to find out more about the field of medical statistics, talk to people from different organisations and make contacts for the future. All students currently studying for a mathematics or statistics-related BSc, MSc or PhD are invited to attend, and we welcome interest from exhibitors too.
This networking event is aimed at statisticians that are new to the pharmaceutical industry who wish to meet colleagues from different companies and backgrounds.
In this vital role you will generate evidence to meet Health Technology Assessment (HTA) requirements, while ensuring that the analyses performed meet required standards and are robust and valid.
Bristol Myers Squibb - Director, Statistical Methodology and Innovation
Lead the development of innovative statistical methods, provides expert consulting, oversees tools and software, and mentors team members while collaborating cross-functionally to address complex drug development challenges.
Phastar – Associate Director, Statistics – Real World Evidence (RWE)
As an Associate Director, Statistics (RWE) you will focus on supporting one of our key pharmaceutical clients within our FSP function. This will be a hands-on technical role, with oversight responsibilities and cross-functional exposure.
Phastar – Principal Statistician – Real World Evidence (RWE)
We are seeking an experienced Statistician to join our FSP team to support one of our fantastic and well-known global pharmaceutical clients, at Principal level.
We are seeking an experienced Statistician to join our FSP team to support one of our fantastic and well-known global pharmaceutical clients, at either Senior or Principal levels.
This position is deal for a statistician who values ownership, collaboration, and using data to enable confident development decisions and to support regulatory submissions.
.jpg?sfvrsn=22c9a9db_1 "N.W. Galwey, Essex, 31 March 2026 (20260331_144133)")
