(Close Window)
Topic: Check my math
Message: Posted by: MeetMagicMike (Jul 26, 2019 05:22PM)
How can these two statements both be true?

A) Tigers are more dangerous than wolves because if you see a tiger there is a 49% chance you will be attacked. For wolves it's only 5%

B) Wolves are more dangerous than tigers because last year more people were attacked by wolves than by tigers.

Obviously, wolves can't be both more dangerous and less dangerous than tigers but I do believe the supporting statisticcs given can both be accurate. So which is true?
Message: Posted by: landmark (Jul 26, 2019 10:45PM)
A) Voldemort is more dangerous than a swimming pool. If you encounter Voldemort there is a 99% chance you will be obliterated. For encounters with a swimming pool, it's much less.

B) Swimming pools are more dangerous than Voldemort. More people died from swimming pools last year than from Voldemort.

And so on. We might instead want to define the annual danger index as the product of the number of encounters expected in a year by the probability of any single such encounter being fatal.
Message: Posted by: katarinag (Jul 28, 2019 09:47PM)
One animal is more dangerous in a one-on-one interaction.

The other lives closer to populated areas (and there are way, way more of them all over the world) so you’re a whole lot more likely to encounter it.

Different criteria / measures. Which is basically the same point Landmark made.
Message: Posted by: MeetMagicMike (Jul 28, 2019 10:37PM)
Thanks guys. I think this is a good example of how statistical data can be abused.
Message: Posted by: ddyment (Jul 29, 2019 10:26AM)
This isn't so much an example of the misuse of statistics as it is the misuse of language. The discrepancy arises not because of any mathematical concerns, but because the word "dangerous" is used with two different meanings.

In the first example, it is taken to mean "likely risk of death in an encounter"; in the second it is taken to mean "likely risk of death to an overall population". So it's more an "apples and oranges" problem than one of statistics.