"Statistics show that children are most often harmed and killed by family"

[debunker]

This mantra is frequently used against the McCanns by people who harass them.

IT ignores the basic rule of statistics that statistical prediction does not apply to individual cases- there are always exceptions and outliers

[debunker]

Statistics is the collection of many facts (each 'statistic' is a single fact) and then drawing general conclusions from that collection of facts. Statistical prediction is broad brush and ignores specifics. Counter examples are not unexpected

Most murderers are male but there are female murder ers

Most basketball players in major league are very tall but some  top basketball players are very short.

Most people with severe mental illness smoke, but some do not

Most midwives are female, but some are male.

Most sexual abuse of  children occurs in family groups or acquaintances, but strangers also abuse.



And so on. Statistical inference does not control individual incidents.

[Eleanor]

That is such a pathetic argument.  Children do get abducted by strangers.  That is an indisputable fact.

[ferryman]

Shades of Roy Meadow and cot death in infants.

"Meadow's rule"

To lose one baby to a cot death is a tragedy.

To lose two is suspicious

To lose three is murder

Meadow worked on simple statistics: general instances of cot death are something like 1 in 7000

Meadow argued that for that to happen twice would be that number squared, and three times, the number cubed.

Meadow overlooked that genetic/hereditary factors might actually shorten, not lengthen, the odds of repeat occurrences.

But that wasn't his most serious error.

Even if you discount genetic/heredity or environmental influences, for a parent to kill his or her child is one rare and independent event; for parents to lose a child to a cot death is a second rare and independent event.

There is no correlation between the two, and no attempt should be made to correlate them ....

[debunker]

Excellent argument.