I was best able to understand it by running the numbers. Take a condition with a 0.1% occurrence rate in the population: then 1,000 out of every million people

I was best able to understand it by running the numbers. Take a condition with a 0.1% occurrence rate in the population: then 1,000 out of every million people have this condition. Then, say you develop a test for the condition that's "99% accurate". Sounds great! Out of the thousand people with the condition, 99%, or 990, will get a positive test result! Unfortunately, of the people without the condition, 1%, or 9,990, will get a positive test result. So if you're one of the 10,980 people who get a positive test result from this "99% accurate" test, it turns out you still only have a 9% chance of actually having the condition.