Popular Ignorance of Statistics

I thought this thread might do well with a nice concrete example. From another AH thread going on right now.

http://www.boards.ie/vbulletin/showthread.php?p=91269146#post91269146

For some short context, girl went home with someone she didn't know and got raped.

The poster below then pulls out the following post (I shortened the post slightly by deleting the lines at the end)

floggg wrote: »

Exactly. I have done it plenty of times, be it for parties or sex. As have most people here.

And don't the statistics say that you are more likely to be raped by somebody you know? Which means you can never really tell when you might be at risk.

So if you argue that going back to somebody's house on the first night is risky, so too is going back to their house after 5 dates, or two years of friendship.

You have no way of knowing what will happen of when.

And the only safe way to avoid being assaulted (sexual or otherwise) is to avoid all contact with people.

I think it's a good example of statistics being terribly misused by someone (I think probably out of ignorance rather than a deliberate attempt to mislead although that's not always the case).

Now I've often heard the point being made that you are most likely to be raped by someone you know. I've no idea whether it's true although I'd imagine so, and for the purpose of this lets assume it is true.

The generalisation that the poster is drawing from this "statistic" is that because rapes happen more often by someone you know, the girl in this case wasn't increasing her risk factor by going home with a stranger.

This misses quite a few key points (I'm not going to list them all out, just a few of them)

1) By definition you spend most of your life with people that you know. So almost anything you care to measure will happen more likely to you by someone you know (I'm more likely to be insulted by someone I know, I'm more likely to be punched by someone I know etc etc)
2) Assuming for a minute that most rapes happen when you are along with someone, the amount of time you spend alone with people you know dwarfs the amount of time you spend alone with strangers
3) As a general rule you can't take a very general statistic and apply it to a very specific situation. For example, I might know that the chances of a person in Ireland catching an illness today might be 0.001%. However I can't then decide to sit in a room full of people that have that illness and say "my chances of of catching that disease today are 0.001%" simply because I happen to a member of the population of Ireland and that's the rate for that population.....

Mr. Boo wrote: »

I thought this was obviously implied by my point, and thus did not require explicit statement.

You're right i feel silly. reading comprehension 101

rockbeast wrote: »

Something will either happen or it won't, no?

See cmssjone's post.

rockbeast wrote: »

If you believe you can "predict" the future, fair enough, Mr Boo. I fully believe there is a need for statistics but certainly reject the "soothsayer" aspect some over-zealous proponents attach to them.

I can see a clear difference between soothsaying and calculating the probability of some event (with a measure of uncertainty) based on observations of that event.