Question: Do you think your research is ever 100% accurate?

Wei Xun answered on 13 Jun 2011:

Hi Jack:

To answer your question: no! As I said before, there are no certainties in science. Most scientists will also tell you how likely what they find is actually reflecting the “truth”, we call this the p-value. To save us scientisits arguing about this all day, there’s a mutually agreed limit for p-value at 5%: this is how likely we have found find something interesting, even though there’s nothing there at all, just by chance (ie false positive).

For example, if you feel ill and go to take a flu test, but really you don’t have flu but you didn’t know at the time. If you take the same test 20 times, 19 times it comes back negative, but once it is positive just because of bad luck (ie the technician did the test wrong, or they used someone else’s blood etc). So we’ll say that there’s a 1 in 20 chance (5%) that you might have flu but we’re willing to take that chance and say that you really don’t.

You can also see it this way, that we’re 95% sure that you don’t have flu.

I hope that wasn’t too confusing 😛