Ines is a sophomore at Kettle State. During orientation, the campus health center offered a free screening for a condition that 1 in 100 students have, usually without knowing it.
The pamphlet says the test is right 90% of the time, for students who have the condition and for students who don’t. This morning, Ines’s result came back positive. How worried should Ines be?
1,000 students
Each mark is one student at the screening. The 10 circles are the students who have the condition, 1 in 100. The 990 squares are the students who don’t.
Has the conditionDoesn’t have it
Nobody has been tested yet.
With a positive result, how likely is it that Ines has the condition?
You chose about 9 in 10.
No. It’s much lower. The pamphlet’s 90% is about how often the test is right for each group. Ines needs something else: of the students who test positive, how many have the condition? Here are all 1,000, with their results.
You chose about half.
No. It’s much lower. The pamphlet’s 90% is about how often the test is right for each group. Ines needs something else: of the students who test positive, how many have the condition? Here are all 1,000, with their results.
You chose less than 1 in 10.
That’s right. The pamphlet’s 90% is about how often the test is right for each group. Ines needs something else: of the students who test positive, how many have the condition? Here are all 1,000, with their results.
You said you’re not sure.
Fair enough. The pamphlet’s 90% is about how often the test is right for each group. Ines needs something else: of the students who test positive, how many have the condition? Here are all 1,000, with their results.
Everyone’s result
The circles are still the students with the condition, and the squares are the students without it. Now each mark also shows a test result: filled means tested positive, and open means tested negative.
Of the 10 circles, 9 are filled. The test is wrong for 1 in 10 students without the condition, so 99 squares are filled too.
Has the conditionDoesn’t have itFilled: tested positiveOpen: tested negative
Ines tested positive, so the question is about the students who did too: press Set aside those who tested negative.
The 108 students who tested positive are now at the top, and only 9 of them have the condition: 9 out of 108, or 1 in 12.
The 99 without it outnumber the 9 with it. Here, 1 in 10 of a big group is more than 9 in 10 of a small one.
The pamphlet’s 90% and Ines’s 1 in 12 are both conditional probabilities. The 90% is about students with the condition. The 1 in 12 is about students who test positive.
P(positive | condition) = 9/10 = 0.9
P(condition | positive) = 9/108 = 0.0833
What it adds up to
So how worried should Ines be? Enough to get a follow-up test, but not to panic: only 1 in 12 students who test positive have the condition.
The probability of a positive result, given the condition, is not the probability of the condition, given a positive result. When the condition is rare, the two can be far apart.
