[T-hawk]

This is known as the multiple-endpoints fallacy, where you deliberately pick the outlying data point after the fact and ignore all the mundane ones.

Another example is when you see reports of some county experiencing a once-in-500-year flood, trying to establish the shock value of how unusual and terrible a catastrophe that is.

But if you're observing 500 such counties, then every year that will happen to some one of them. You just never notice the 499 data points where it didn't.