Sorry, I put that badly. I should have used probability not risk.
I mean, if people think a probability can be controlled, controlling it takes on a disproportionate importance and any downside of controlling it is ignored.
Out of 100000 men and 100000 women, there will be more child abusers amongst the men. But out of 100000 men and 100000 women there will be many more men than women who started a pub fight, say.
So it's true to say that if a UM sits next to a man, she has a greater probability of sitting next to someone who has started a pub fight than if the UM sits next to a woman.
But the risk of the person sitting next to the UM starting a fight on a plane cannot be quantified by their "group" probability of having previously started a fight.
Not to mention that if you grouped the probabilities I. Other ways (under 60s and over 60s, for example) you could come up with a different seating plan that would also not be "risk related"
I may or may not be making myself clearer here...