Jeremy Stangroom

If you want more examples of this sort of thing, you could take a look at a book I wrote with Ophelia Benson, "Does God Hate Women?". I think you'd be horrified at some of the things that happen to women in the name of religion.

Okay, well this is quite a difficult topic. Basically, you're absolutely right, you can't say something like men tend to behave badly, therefore Tom is bad. You can't even say, men tend to behave badly, therefore Tom is probably bad.

Why can't you make the latter claim?

Because you can't move from what is statistically true of a distribution (e.g., 60% of men are bad) to a claim that there's a 60% chance any particular man is bad. In fact, it's hard to make sense of a probability statement about a particular individual (he's either bad or not).

Why not? If your only data point on this individual is that he's a man, why can't you do this? If 99% of snakes are venomous, is it not reasonable to assume any given snake you encounter is venomous, or at least likely enough to be venomous that caution is advised?

The snake you come across either is venomous or it isn't - probability doesn't come into it. What does it mean to say of a snake that isn't venomous that there's a 90% chance it's venomous?

The snake either is or isn't venomous. It's either 100% venomous or 0% venomous. But until you know that, you're reasonable to treat the snake with caution, since your only data point is that it's a snake, and 99% of them are venomous. Right?