There’s a fairly simple statistical trick that I’ve gotten a ton of leverage out of. This is probably only interesting to people who aren’t statistics experts.
The trick is how to calculate the chance that an event won’t occur in N trials. For example, in N dice rolls, what’s the chance of never rolling a 6?
The chance of a 6 is 1⁄6, and there’s a 5⁄6 chance of not getting a 6. Your chance of never rolling a 6 is therefore (5/6)N.
More generally, the chance of an event X never occurring is (1−X)N. The chance of the event occurring at least once is 1−(1−X)N.
This has proved importantly helpful twice now:
Explaining why the death of 1 of 6 mice during a surgery for an exploratory experiment is not convincing evidence that the treatment was harmful.
Explaining why quadrupling the amount of material we use in our experiment is not likely to appreciably increase our likelihood of an experimental success.
There’s a fairly simple statistical trick that I’ve gotten a ton of leverage out of. This is probably only interesting to people who aren’t statistics experts.
The trick is how to calculate the chance that an event won’t occur in N trials. For example, in N dice rolls, what’s the chance of never rolling a 6?
The chance of a 6 is 1⁄6, and there’s a 5⁄6 chance of not getting a 6. Your chance of never rolling a 6 is therefore (5/6)N.
More generally, the chance of an event X never occurring is (1−X)N. The chance of the event occurring at least once is 1−(1−X)N.
This has proved importantly helpful twice now:
Explaining why the death of 1 of 6 mice during a surgery for an exploratory experiment is not convincing evidence that the treatment was harmful.
Explaining why quadrupling the amount of material we use in our experiment is not likely to appreciably increase our likelihood of an experimental success.