Lies, damned lies and statistics.

Consider the following two statements -
(1) In this country, at midnight, it is dark.
(2) We are in this country and it is dark, therefore it is midnight.
You will clearly see (at least I hope you will!) that this is utter tosh. Now consider these two statements -
(1)In this country, in April, the probability that the temperature at midday will be greater than 15ºC is 80%
(2)We are in this country, it is midday and the temperature is greater than 15ºC, therefore the probability that it is April is 80%
Perhaps not quite so obviously, but this is also tosh (what about May, June, July, August, September...?)
Now look at these two statements -
(1) A crime has been committed. Blood of a particularly rare type has been found at the scene and it is accepted that it belongs to the criminal. The chances of a person chosen at random having this blood type are 1 in 1,000,000.
(2) You have this blood type. It is therefore a million times more likely than not that it is your blood, and that therefore you committed the crime.
Still tosh - you bet! And yet, substitute DNA for blood type and one in a billion for one in a million, and this sort of argument is being used all the time in court cases, and juries are falling for it. Perhaps we need a resident court statistics expert to keep the record straight.