Expert Witness Journal Issue 67 June/July 2026 - Flipbook - Page 98
In fact, Professor Meadow’s 昀椀gures were entirely
bogus, and misleading in any event. Recognising
that the infants were related, the chances of two
cot deaths was orders of magnitude higher than
1 in 73 million, possibly as high as 1 in 200. Even if
a much lower 昀椀gure were correct, it was wrong to
imply that this equated to the chance of innocence
(the ‘prosecutor’s fallacy’). As the Royal Statistical
Society explained to the Lord Chancellor, the fact
that two deaths by SIDS is quite unlikely is, taken
alone, of little value. Two deaths by murder may
be even more unlikely. What matters is the relative
likelihood of the deaths under each explanation, not
just how unlikely they are under one explanation.
the babies. Defence experts subsequently found the
evidence inconclusive and the causes of death not
established – however, by that stage, Sally Clark was
a defendant in a criminal prosecution.
The police, the prosecutors, the state (it would
appear) thought her guilty of murder. The jury
had to reach its conclusion on the basis of a mass of
contradictory expert medical evidence. There were
nine experts – four for the prosecution and 昀椀ve for
the defence. Some prosecution experts had changed
their minds on fundamental issues.
How are juries to decide between competing experts?
Fortunately, people are machines for jumping to
conclusions, and no evidential uncertainty is too
great to overcome.
Sally Clark was convicted. When the statistical errors
were explained to the Court of Appeal in Sally’s
Clark’s 昀椀rst appeal, they were not fully understood
and not acted upon. The Court of Appeal only
recognised the signi昀椀cance of Professor Meadow’s
misleading evidence in Sally Clark’s second appeal
– when supported by new medical evidence pointing
to natural death based on previously undisclosed
microbiological tests performed at post-mortem.
Sally Clark’s convictions were 昀椀nally overturned.
Evaluating expert evidence
Daniel Kahneman and his colleague, Amos Tversky,
found that people often solve di昀케cult questions
by (unconsciously) 昀椀nding an easier question and
answering that instead. Rather than conducting
a forensic analysis of the technical and scienti昀椀c
evidence, a juror is liable to substitute a question
like: which expert seems more authoritative?
A wider problem
This is just one example, of course, but there are
many others. We do not have to wade too far into the
Letby debate to note that Dr Dewi Evans, the lead
prosecution expert, identi昀椀ed deaths as suspicious
on the basis of medical records that had not been
‘blinded’ and when Letby was the only suspect. That
the allegedly suspicious nature of those deaths
has now been challenged (robustly) by a panel of
international experts leaves the prosecution case
perilously circular – ‘Letby was the only one present
for all suspicious deaths’ becomes ‘deaths were
suspicious because Letby was present’.
The juror will be aided by all manner of cognitive
shortcuts. As well as authority bias causing jurors to
prefer experts who appear to have a higher status,
or simply a more impressive courtroom demeanour,
anchoring bias may lead to jurors using the 昀椀rst
expert opinion they hear (usually prosecution
experts) as their reference point – if the initial
expert provided compelling evidence any expert
providing contradictory evidence might be viewed
as less credible; the availability heuristic means
that jurors may be more in昀氀uenced by evidence
that is easier to understand, vividly presented, or
emotionally compelling; con昀椀rmation bias might
come into play, with jurors favouring experts who
support their intuition as to the defendant’s guilt or
innocence.
This issue is far broader than the Sally Clark and
Lucy Letby cases. Peoples’ preconceptions and
motives shape their interpretation of the world
around them. Not merely by noting and giving
weight to evidence that supports pre-existing
theories and disregarding evidence that does not
(con昀椀rmation bias) but in di昀昀ering interpretations
of the same evidence. Even for something that might
be thought as open and shut as 昀椀ngerprint analysis,
if examiners are told that a suspect has an alibi, they
are less likely to conclude that a print found at the
crime scene came from the suspect.
There is another major problem. Expert witnesses
are as susceptible to bias as everyone else, sometimes
more so. Alternatively, they might just be wrong.
Sally Clark (Part 2)
Cutting through the noise of Sally Clark’s trial
the eminent paediatrician, Sir Roy Meadow, gave
evidence that there was a 1 in 73 million chance
of two cot deaths. It was, he said, something that
would happen only ‘once in every 100 years’. He later
compared the chances of both children dying natural
deaths to picking four 80 to 1 winners of the Grand
National in successive years. That a defence expert
witness pointed out that these 昀椀gures involved an
illegitimate oversimpli昀椀cation is unlikely to have
been so impactful.
EXPERT WITNESS JOURNAL
The more subjective the assessment, the greater
the potential for bias. For example, determining
whether deaths are suspicious when the causes of
death are unclear. Or determining whether there
has been a failure in care in contentious and di昀케cult
areas of medicine. Now add to that uncertainty
a suspicion, whether warranted or not, against a
96
JUNE 2026
