On Newsnight today, Cameron made the case for Marriage. He said that if you look at a random child of unmarried parents, by the age of five there is a 1 in 5 chance that the parents will have separated. The chance for a married couple is 1 in 12. Therefore, he argues, marriage provides a more stable basis in which to raise children.
He's fallen into a logical fallacy here. Correlation doesn't imply causation. It could simply be that couples which are more stable are more likely to get married.
To make it a fair test, you would have to take a random sampling of unmarried people with a newborn child and then randomly assign them to be married or not married (with no option to become married). I very much doubt if this has been done, not least because of the ethical issues.
If you're going to make an argument, you should use the data fairly - or at least tacitly admit its flaws (i.e. 'it looks like marriage is a more stable arrangement in which to raise children'). I'd like to think that Cameron realises the flaw in his argument, but is paraphrasing for the benefit of the camera - I'm slightly worried that a potential Prime Minister hasn't spotted the logical flaw and what this implies about his reasoning ability.
It is a shame that nobody who interviewed Cameron understood the distinction between causation and correlation.