Prof. Peter Green — statistical methodology and base rates

Why he matters in this case

Prof. Peter Green is one of the UK’s most senior statisticians. As a past president of the Royal Statistical Society (2002–2003), he has direct institutional standing with the body whose post-Sally-Clark guidance on statistical evidence in criminal trials is the canonical English reference point. His public commentary on the Letby shift-rota chart carries specific weight for that reason: he is not an outside critic applying foreign statistical methodology; he is a former head of the UK professional body whose post-Clark framework is directly applicable.

Green authored (jointly with Rebecca Steventon and Prof. Jane Hutton) the October 2022 Royal Statistical Society report Healthcare serial killer or coincidence? Statistical issues in investigation of suspected medical misconduct. The report was published approximately a month before the Letby jury retired. Its content is a direct methodological critique of the specific statistical approaches the Crown’s case at Manchester Crown Court then used. That the RSS would issue such a framework publication at that timing is itself load-bearing.

What the RSS 2022 report says about cluster charts

The 2022 RSS report sets out, in plain English, the specific errors characteristic of cluster-chart evidence in healthcare-serial-killer prosecutions:

Selection effects. Events are selected because the suspect was there. Once the events are selected, the suspect’s presence at them is arithmetic, not evidence.
Ignored base rates. Sudden neonatal deaths and unexplained collapses occur at documented baseline frequencies. The probability of cluster patterns arising by chance when base rates are correctly applied is much higher than prosecution experts typically represent.
The Texas sharpshooter fallacy. Drawing the target around where the bullets landed. Clusters are defined only after their members are identified, then the same members are used as evidence that the cluster is unusual.
The prosecutor’s fallacy. Inversion of conditional probability: the probability of the evidence given innocence is confused with the probability of innocence given the evidence.
The independence assumption. Events are treated as statistically independent when they clearly are not (they occur on the same unit, with shared staff, shared equipment, shared infection-control context).

The Letby shift-rota chart presented to the jury exhibits all five problems. This is not Green’s opinion; it is the RSS’s published institutional guidance applied to a specific evidential instrument.

The Sally Clark continuity

The RSS’s framework on statistical evidence in criminal trials was developed specifically in response to the 1999 R v Clark conviction and its subsequent 2003 quashing. In that case, Sir Roy Meadow’s “one in 73 million” cot-death calculation was wrong — a squaring of a single-event probability that ignored both family shared-environment correlation and the independence assumption. The RSS’s post-Clark framework identifies exactly the kind of error Meadow made and requires courts to avoid it. Green chaired the RSS during the period that framework was being applied institutionally; his involvement in the 2022 Letby-relevant report is a continuation of that professional-body work.

Professional background

Emeritus Professor of Statistics, University of Bristol; previously held the chair of Probability and Statistics there (1989–2018).
President of the Royal Statistical Society, 2002–2003 — the post-Sally-Clark period during which the RSS developed its canonical framework on statistical evidence in criminal trials.
Bayesian statistician; author on complex stochastic systems, Markov chain Monte Carlo methods, and applied statistical modelling.
Co-author, RSS report ‘Healthcare serial killer or coincidence?’ (October 2022).
Fellow of the Royal Society (FRS); elected 2003.

Why this biography is on the site

The statistical architecture of the Letby case is the most specialist part of the evidential picture. Readers who are not statisticians themselves need access to the institutional authority of the people making statistical claims, because the jury received these claims through layers of expert presentation. Green’s institutional standing — past RSS President, Emeritus at Bristol, FRS, co-author of the RSS framework publication on this specific case type — is why his public commentary on the shift-rota chart is load-bearing. This page supplies that institutional context.

Read alongside

Source

Royal Statistical Society report ‘Healthcare serial killer or coincidence? Statistical issues in investigation of suspected medical misconduct’ (Green, Hutton, Steventon; October 2022); RSS archived president list; University of Bristol School of Mathematics faculty profile; Royal Society Fellows Directory; RSS public framework documents on statistical evidence in criminal trials (post-Sally-Clark guidance).