This tutorial covers false positives and false negatives. Now with tests of any kind, whether it's a medical test or some sort of evaluation, they're never perfect. You're not going to be 100% accurate. So it's a starting point for this lesson, tutorial.
Now if you get a positive test result, that means the thing you were looking for, whatever condition it is, you found it. If you get a negative test result, whatever the thing you're looking for is, you didn't find it. So with a false positive, that's when it's falsely positive, when the test says the condition is there and it's not. With a false negative, it's when the test says the condition isn't there when it really is.
In a medical setting, both of these situations are really concerning. With a false positive, there's someone who thinks that they have a disease or something like that and they really don't. So that's why with particularly stressful tests or ones that are really important, like an HIV test, you have to get two positive results in a row. The chances of getting two false positives in a row are really unlikely, so this helps to remove some of the uncertainty around something like that. Same with a false negative. If you think you don't have a disease and you actually do, then you have a false sense of security. So that's something else that doctors are really vigilant about.
This example, we're going to look at a lie detector test. So for a positive result, they're finding the thing they're looking for. They're finding that you're a liar. For a negative result, the test is saying that nope, you're telling the truth. So a false positive would be when a true statement gets marked as a lie. A false negative would be when an untrue statement gets marked as true.
Now in this particular setting, a company is trying to find out who is stealing from them. And they have this test and they know that 90% of true statements gets marked as lies. And 90% of untrue statements get marked as true. Let's tease out the numbers a little bit.
So let's say that the company has 200 employees, and 20 of them have stolen and would want to lie about what they had done. So if you look at the non-thieves side, if 20 have stolen, then 180 haven't. And if 90% of true statements get marked as lies, that means 90% of these people are going to have their statements marked as lies. And 10% of these people will get marked as true. So it turns out to be 162 people are going to get marked as telling lies and 18 people are going to get marked as telling the truth.
On the other hand, we have 20 thieves, and they are going to lie. They're not going to want the company to find out that they were stealing. So 90% of their statements, their lies, are going to be marked as true. So those 18 statements by the thieves that get marked as true, those are false negatives. So they're marked as true. They're not actually, though.
That's about as many statements that are true do get marked as true. So in this situation here, the false stuff is almost as accurate, is almost as prevalent, as the actual. So that's a problem. So this test is actually really, really bad.
This has been your tutorial on false positives and false negatives. They're when a test result isn't accurate. And these are very important to look at, particularly in medical situations, but can be applied any time there's a test involved.