The shift chart — selection bias presented as proof
Prosecution claim
A chart shown to the jury plotted 25 suspicious events against the nurses on duty for each. Letby was the only nurse present at all 25. The prosecution argued the improbability of this pattern, if she were innocent, was proof of her guilt.
Counter-evidence
The 25 events were selected in part because Letby was there. Collapses where she was not on shift were excluded from the chart. Statisticians including Prof. Richard Gill (Leiden, instrumental in the Lucia de Berk exoneration) and the Royal Statistical Society have characterised this as a textbook 'Texas sharpshooter' fallacy: painting the target around the bullet hole. triedbystats.com models the chart in detail — base rates of shift attendance alone mean that whoever works the most unsociable shifts will end up plotted against an unusual cluster of deteriorations, without any wrongdoing. When other nurses' attendance is plotted against the full set of collapses (including those excluded from the trial chart), the pattern dissolves.
"The chart shown to the jury is statistically meaningless. You cannot select the events because Letby was there and then use her presence as evidence of guilt." — Prof. Richard Gill
What the jury heard
A rota-style grid of 25 events with every nurse's shifts shaded; Letby's row was the only one fully shaded. The Crown's opening invited the jury to infer from this single image that coincidence was impossible.
What the Panel says
The Panel did not opine on statistics directly but referenced the Royal Statistical Society's published warnings that selection effects vitiate any inference from the chart. Its medical review concluded that every collapse was explicable without deliberate harm — which would itself dissolve the chart's premise.
What independent experts add
- Prof. Richard Gill (Leiden) — public letters and detailed statistical critique; instrumental in the Lucia de Berk exoneration in the Netherlands.
- Royal Statistical Society — formal public commentary on selection bias.
- triedbystats.com — visual, interactive model showing how the chart would look for every nurse on the unit if the selection were not pre-restricted to Letby.
- Prof. Peter Green (Bristol) has separately written about base rates and the Texas sharpshooter problem in this case.