42 How to Win Every Argument
coincidence, or for the operation of an outside factor which
separately influences those events.
A tourist met a Spanish peasant and his wife on a train. They had never
seen bananas before, so he offered one to each of them. As the farmer bit
into his, the train entered a tunnel. 'Don't eat it, Carmen,' he shouted,
'They make you blind. 'Like the post hoc fallacy which links events because they occur
consecutively, the cum hoc fallacy links them because they occur
simultaneously. It is a fallacy because of its unwarranted
assumption that either of the events would not occur without the
other one.
Things are happening all the time. Scarcely a day goes by
without rain, electricity bills, televised show-jumping, or the
Guardian newspaper. It is inviting to link these endemic dis-
comforts with simultaneous events, and conclude that they are
all somehow connected. In primitive societies such assumptions
are made routinely, and one of the jobs of the witchdoctor is to
sort out which actions are linked with various consequences. In
our society, alas, life is more complicated.
The field of statistics provides a natural habitat for the cum hoc
fallacy to lurk undetected. Indeed, a whole branch of statistics
called regression analysis is devoted to measuring the frequency
and extent of simultaneous events, and to calculating the
probability of their being linked. Correlation coefficients are
produced, with percentages attached showing the likelihood
that mere chance was involved. Statisticians routinely offer us
relationships with a 95 per cent or a 99 per cent probability that
'more than chance is involved'.
A statistician looking over figures for pupil performance was astounded
to discover in the 7-12 age-group that neatness of handwriting matched
with size of shoe. He checked the figures for hundreds of children, but it