Consider the following question taken from this link, question number $25$:

We have four boxes. Box $1$ contains $2000$ components of which $5$ percent are defective. Box $2$ contains $500$ components of which $40$ percent are defective. Boxes $3$ and $4$ contain $1000$ each with $10$ percent defective. We select at random one of the boxes and we remove at random a single component. What is the probability that the selected component is defective?

The solution is given in the form of conditional probabilities as

$$ P(B) = P(A1) P(B/A1) + P(A2) P(B/A2) + P(A3) P(B/A3) + P(A4) P(B/A4)$$

$$P(B) = 1/4 (0.05 + .4 + .1 + .1) = 0.1625$$

I wonder, since the probabilities of selecting each of the box is equal, why the probability of a defective part is not: $$ \frac{\text{Total defective components}}{\text{Total Number of Components}} = 500/4500 =0.1111 $$


The boxes have different numbers of components. So if we choose a box, and then a component, not all components are equally likely to be chosen. The ones in the smaller boxes have an edge.

An extreme example will I think make things clear. Suppose Boxes 1, 2, and 3 each hold one component, which is defective. Box 4 holds $97$ components, all of them good. It is clear that if we pick a box at random, and then a component, the probability the component chosen is bad is $3/4$. If we used Total Number of Bad divided by Total Number of Components, we would get $3/100$, very far from the truth.

  • $\begingroup$ It makes me think what would have been the box selection probabilities so that the answer were consistent with the total sum probabilities as written in my "incorrect" calculation. $\endgroup$ – Abu Bakar Feb 8 '16 at 8:28
  • 1
    $\begingroup$ Box selection probability proportional to its size. So Box 1 with probability $2000/4500$, box 2 with probability $500/4500$, and so on. $\endgroup$ – André Nicolas Feb 8 '16 at 8:31

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Not the answer you're looking for? Browse other questions tagged or ask your own question.