Dian Koeswandi

IE 417 HW#4

Parisay’s comments are in red.

Problem 9 page 772

The government is attempting to determine whether immigrants should be tested for a

contagious disease. Let’s assume the decision will be made on a financial basis. Assume
that each immigrant who is allowed into the country and has disease costs the United
States $100,000, and each immigrant who enters and does not have the disease will
contribute $10,000 to the national economy. Assume that 10% of all potential immigrants
have the disease. The government may admit all immigrants, admit no immigrants, or test
immigrants for the disease before determining whether they should be admitted. It costs
$100 to test a person for the disease; the test result is either positive or negative. If the
test result is positive, the person definitely has the disease. However, 20% of all people
who do have the disease test negative. A person who does not have the disease always
test negative. The government’s goal is to maximize (per potential immigrant) expected
benefits minus expected costs.

Bayes’ calculation?
Prob(has disease | test result is negative) = 1/46. This value is rounded to 0.02 on the

following tree and it will effect the final expected value and in some cases the final

decision. When it come to such probabilities it is better to use 3 to 4 decimal points.

1
Input

Output

2
Decision Tree

EVSI: expected value from test – expected value from no test: $7,176 – $0 = $7,176

EVPI: If a company can provide perfect information the information can be:

a) A person has disease, then the best decision is not to admit, it will cost $0 and has

probability of 0.1 to happen.

b) A person does not have disease, then the best decision is to admit, it will benefit

$10,000 and has probability of 0.9 to happen.

Therefore, expected value of perfect info = 10,000(0.9) + 0(0.1) = 9000

EVPI = 9000 – 0 = 9000

Notice always EVPI > = EVSI

3
Report to manager:

Considering all the given information, the best decision is to test the immigrants and will

give us an expected profit of $7,076. Expected profit is calculated based on probability of

having profit on some people and having a loss on other people, so expected profit of

$7,076 is based on many immigrants and not just one individual and it’s not a fixed profit

that you will get consistently from each one immigrant. The test result can be either

positive or negative, and the chance of the test to be positive is 8% and negative is 92%.

If the test is negative then the best decision is to admit. If you decide to admit them,

there is a 98% chance that they don’t have the disease. If the test result is positive, then

the best decision is not to admit.

The Expected Value of Sample Information is $7,176, which shows the cost limit that

we’re willing to spend to do the test. Since the test cost is $100, so we are safe to do the

test. The Expected Value of Perfect information is $9,000 which means the maximum that

we are willing to spend to get the perfect information.