A company sells hot water systems for $\text{\textdollar}4200$ each.
The lifetimes of the systems are normally distributed with a mean of 11 years.
Each system has a warranty. The warranty refunds the full price for systems failing within five years. It refunds half the price for systems failing between five and eight years.
It is known that 1.2% of the systems receive a full refund.
The company's accountant makes the following claim.
On average, warranty refunds will cost the company less than $\text{\textdollar}270$ for each system sold.
Evaluate the reasonableness of this claim.
[5 marks]Have a go at the question cold first. When you are stuck, or finished, open the list and check which of these steps you found for yourself.
The 1.2% pins down the spread of the curve. Its three refund zones become a table, and the expected refund is measured against the claimed $270.
Choose a preset to draw the lifetime curve, shade its three refund zones and set the expected refund against the dashed $270 line. The first three are accepted working. The last four are common slips.
Only the mean is fixed so far. Pick a preset above to find , shade the three refund zones and see the expected refund for that set of working.
The expected refund is more than $270, so the claim is not reasonable. This is the working in the marking steps below.
Rounding $\sigma$ to 2.7 moves the half-refund probability to 0.1201, but $302.66 is still more than $270. The conclusion is the same, so this rounding is accepted.
Recalculating the full refund with $\sigma=2.7$ as well moves it to 0.01313 and the expected refund to $307.43. That is still more than $270, so this is accepted too.
$P(X\lt8)$ also includes every system that fails within five years, so the full-refund systems are counted a second time, at $2100 as well as $4200. The verdict happens to match, but this loses Mark 2. The later marks can still be earned by follow-through.
Without the full refund the expected refund drops to $246.75, below $270, so the claim wrongly looks reasonable. This loses Mark 3. The comment is then marked by follow-through.
With only the full refund the expected refund is $50.40, so the claim wrongly looks reasonable. The half refunds are the larger cost. This loses Marks 2, 3 and 4. The comment is then marked by follow-through.
Reading 1.2% as 0.12 makes the tail ten times too big, so $\sigma$ grows to 5.106 and every later value is wrong. This loses Mark 1.
Try each step yourself before you reveal it.
Let $X$ be the lifetime (years) of a system. $X$ is normally distributed with mean 11 and standard deviation $\sigma$. A full refund means $X\lt5$, so 1.2% is a left-tail area.
$$P(X\lt5)=0.012$$
$$P(Z\lt z)=0.012\Rightarrow z=-2.2571$$
$$\dfrac{5-11}{\sigma}=-2.2571$$
$$\sigma=2.658$$
The mean is given but the standard deviation is not, and the 1.2% is the only other fact about the lifetimes. Inverse normal turns that area into a $z$-score, and the $z$-score formula gives $\sigma$.
Use inverse normal with a left-tail area of 0.012, mean 0 and standard deviation 1 to get $z=-2.2571$. Store $\sigma$ unrounded and use the stored value from here on.
Rounding $\sigma$ to 2.7 is accepted. It gives an expected refund of about $303 to $307, which supports the same conclusion.
Marker: correctly determines the standard deviation of the lifetimes
A half refund is for a system that fails between five and eight years, so its probability is the area from 5 to 8 under $X\sim N\!\left(11,\ 2.658^2\right)$.
$$P(5\lt X\lt8)=0.1175$$
Use normal cdf with lower 5, upper 8, mean 11 and the stored $\sigma$.
Using $P(X\lt8)=0.1295$ for the half refund does not gain this mark, because it counts the full-refund systems a second time. It leads to about $322. The remaining marks can still be earned by follow-through.
Marker: correctly determines the probability that a system receives a half refund
Let $R$ be the refund ($\text{\textdollar}$) for a system. It takes only three values: the full price $4200, half the price $0.5\times4200=\text{\textdollar}2100$, or $0 for a system that lasts beyond eight years. So $R$ is a discrete random variable.
$$P(R=0)=1-0.0120-0.1175=0.8705$$
The probabilities of a discrete random variable must sum to 1, which gives the third probability.
Leaving out the full refund gives about $247 and does not gain this mark. The final comment can still be earned by follow-through.
Marker: recognises that the refund for a system is a discrete random variable and correctly determines its probability distribution
“On average” means the expected value.
$$\begin{aligned}E(R)&=4200(0.0120)+2100(0.1175)\\&\quad+0(0.8705)\\&=297.15\end{aligned}$$
For a discrete random variable, $E(R)=\sum r\,P(R=r)$. Here that is $50.40+246.75+0$.
Including only the full refund gives about $50. That response does not gain this mark, or the second and third marks. The final comment can still be earned by follow-through.
Marker: determines the expected refund for each system
The claim is a threshold of $270, so compare the expected refund with it and state the decision.
$$\text{\textdollar}297.15>\text{\textdollar}270$$
$$\dfrac{297.15}{270}=1.1006$$
Marker: evaluates the reasonableness of the claim using a suitable comment
The question gives you a price, a mean, a refund policy, one percentage and a claim, but no standard deviation, no random variable and no instruction to find an expected value. You have to see that the 1.2% is the left-tail area below 5 years, because a full refund happens exactly when a system fails within five years, and use it to find $\sigma$. Then you have to invent the refund for a system, $R$, as a discrete random variable whose probabilities come from three intervals of a normal distribution. The required procedure is not clear from the way the problem is posed.
Each piece on its own is routine Unit 3 and 4 work: an inverse normal, a normal probability, a probability distribution that sums to 1 and an expected value. What is unfamiliar is the chain. The words “on average” have to become $E(R)$, the half refund has to become the interval from 5 to 8 rather than everything below 8, and the question only asks you to evaluate a claim, so the comparison with $270 and the decision are yours to set up.
Only mark this done when you could do it without help. Reading the solution does not count.
A Tangent Tuition mock exam question. It is not a past QCAA question, and Tangent Tuition is not affiliated with the QCAA.
Spot a mistake on this page? Report it