Why it matters in the Letby case
Statistical critique of the Letby shift chart argues the chart suffers from this fallacy: events were defined by association with the suspect, and presence at those events was then offered as evidence of guilt.
Statistical explainer
The Texas sharpshooter fallacy
The Texas sharpshooter fallacy is the error of drawing a target around a cluster of data points after the fact, then claiming the pattern is meaningful. In hospital cluster investigations it appears whenever the suspect is identified first and the events are then defined around their presence.
Statistical analysis2Expert/professional
Last updated
Prosecution position
The Crown's case invited the jury to treat Letby's presence at each of the charted events as the pattern requiring explanation.
Expert challenge / post-conviction reading
Statistical critique argues this is the Texas sharpshooter fallacy applied to a hospital cluster: the suspect was identified first, the event set was then constructed around her presence, and that presence was offered as evidence of guilt. The target was drawn after the shots.
What remains uncertain
Naming a fallacy identifies a risk in how an inference was built; it does not establish what the correctly constructed analysis would have shown.
Common questions
Where does the name come from?
The image of a marksman firing at a barn wall and then painting a target around the tightest cluster of holes — a pattern that looks like skill only because the target was drawn afterwards.
How does it apply to a hospital cluster?
It appears whenever a suspect is identified first and the set of events is then defined by their presence, so the association is guaranteed by construction rather than discovered.
Is this the same as selection bias?
They are closely related. Selection bias is the general problem of events not being chosen independently of the hypothesis; the sharpshooter fallacy is the vivid special case of drawing the boundary after seeing the data.
Does the fallacy prove the pattern is meaningless?
No. It means the pattern cannot be treated as evidence without a properly independent event definition and a comparator.
Related reading
Found an error or missing source? Send a correction with the page URL and supporting source.