Is the variance the sum of the sum of conditional variances and variance of conditional expected value?

43 Views Asked by At

Here is the following question:

"I buy a drill and wonder about its lifespan (which is a real positive number). With a probability of 0.25 the drill has a hidden manufacturing defect and its service life is evenly distributed between 1 and 7 years. If it is not defective (probability 0.75), its lifespan is a random variable uniformly distributed between 9 and 15 years."

My problem is that I would like to calculate the variance of the lifespan. My problem is I am unsure of the formula to use (I didn't really understand it on wikipedia)

I use a variable T for the lifespan and D=0 being a defective tool and D =1 being a nondefective tool. For the conditional variances I obtain the following: $Var(T|D = 0)$ and $Var(T|D = 1) = 3$ and for the variance of the conditional expected value $\mathbb{E}[T|D]$ I obtain 12.

In the solution of the exercise, it is given 15 as the total variance. From what I've understood the total variance should be 18 and not 15.

1

There are 1 best solutions below

0
On BEST ANSWER

Considering the two uniform random variables representing the two random lifespans:

1)

$$A\sim U[1;7]$$

2)

$$B\sim U[9;15]$$

The total lifespan's density is a mixture (a linear combination) of the two uniform densities:

$$f_T(t)=0.25\cdot f_A(t)+0.75\cdot f_B(t)$$

Thus

$$\mathbb{E}[T]=0.25\cdot 4+0.75\cdot12=10$$

$$\mathbb{E}[T^2]=0.25\cdot (3+4^2)+0.75\cdot(3+12^2)=115$$

Concluding:

$$\mathbb{V}[T]=115-10^2=15$$

...your textbook is right!