Detailed Explanation
This is a classic Bayes' theorem problem. We are told the result of the test (positive) and asked to find the probability of the actual cause (having HIV). Bayes' theorem lets us reverse conditional probabilities: instead of asking "given HIV, what is the chance of a positive test", we ask "given a positive test, what is the chance of HIV". A key idea is that even a good test can give a low probability of actually being sick when the disease is very rare in the population, because there are so many more healthy people that even a small false-positive rate produces many positive results.
Formula Used
Bayes' theorem: \[P(E_1 \mid A) = \dfrac{P(E_1)\,P(A \mid E_1)}{P(E_1)\,P(A \mid E_1) + P(E_2)\,P(A \mid E_2)}\]
Solution
- Step 1: Define the events. Let \(E_1\) be the event that the person has HIV, and \(E_2\) be the event that the person is free of HIV. Let \(A\) be the event that the test reports the person as HIV positive.
- Step 2: Write down the prior probabilities from the population data. Since only \(0.1\%\) of the population have HIV: \[P(E_1) = \frac{0.1}{100} = 0.001\] \[P(E_2) = 1 - 0.001 = 0.999\]
- Step 3: Write down the conditional probabilities of a positive test. Of people having HIV, \(90\%\) test positive: \[P(A \mid E_1) = \frac{90}{100} = 0.9\] Of people free of HIV, \(1\%\) are wrongly diagnosed as positive: \[P(A \mid E_2) = \frac{1}{100} = 0.01\]
- Step 4: We want \(P(E_1 \mid A)\), the probability that the person actually has HIV given a positive report. Apply Bayes' theorem: \[P(E_1 \mid A) = \frac{P(E_1)\,P(A \mid E_1)}{P(E_1)\,P(A \mid E_1) + P(E_2)\,P(A \mid E_2)}\]
- Step 5: Substitute all the values: \[P(E_1 \mid A) = \frac{0.001 \times 0.9}{0.001 \times 0.9 + 0.999 \times 0.01}\]
- Step 6: Compute the numerator and denominator. Numerator: \[0.001 \times 0.9 = 0.0009\] Denominator: \[0.0009 + 0.999 \times 0.01 = 0.0009 + 0.00999 = 0.01089\]
- Step 7: Divide to get the final probability: \[P(E_1 \mid A) = \frac{0.0009}{0.01089} = \frac{90}{1089} = \frac{30}{363} \approx 0.0826\]
- Step 8: Interpret the result. Even though the test is quite accurate, the probability that a person reported positive actually has HIV is only about \(8.26\%\). This is because HIV is very rare, so most positive results come from false positives among the huge healthy population.
Final Answer
The probability that the person actually has HIV is \(\dfrac{90}{1089} \approx 0.083\), that is about \(8.3\%\).
Common Mistake
A common mistake is to answer \(0.9\), thinking that because the test detects \(90\%\) of HIV cases the person must have a \(90\%\) chance of being infected. This ignores the prior probability \(P(E_1)\) and the false positives from the large healthy population, which Bayes' theorem correctly accounts for.