Witness evidence: can it be trusted?

Basic statistical principles are often widely misunderstood (and often mistrusted) by many members of the public at large. It comes as no shock that this also applies to legal practitioners. If there is an ignorance of the intricacies of statistics by the judiciary and counsel, it is not surprising that erroneous conclusions based on the evidence have arisen. These have even been afforded their own special designation as 'fallacies', including those of both the prosecutor and defence. Whether or not the underlying mathematics is fully understood, it is essential that practitioners can correctly interpret information provided by witnesses, expert or otherwise. This article provides one contribution in that direction toward cases that specifically involve medical situations. It considers some of the more outrageous conclusions made, and accepted, in several high profile cases and considers the evidence more carefully. The challenge is to be able to determine the likelihood that any assertion made is actually true from a statistical point of view. Examples of how to make this determination are provided, including the all-important notion of conditional probability.