CAIE FP2 2011 November — Question 9 13 marks

Exam BoardCAIE
ModuleFP2 (Further Pure Mathematics 2)
Year2011
SessionNovember
Marks13
PaperDownload PDF ↗
Mark schemeDownload PDF ↗
TopicZ-tests (known variance)
TypeUnknown variance (t-distribution)
DifficultyStandard +0.8 This question requires calculating a confidence interval from small sample data, then performing a two-sample hypothesis test with unequal sample sizes where variance must be pooled from the first sample. It combines multiple statistical concepts (CI construction, hypothesis testing, variance estimation) and requires careful handling of the 'variance unchanged' assumption across samples, going beyond routine single-procedure questions.
Spec5.05d Confidence intervals: using normal distribution

9 A random sample of five metal rods produced by a machine is taken. Each rod is tested for hardness. The results, in suitable units, are as follows. $$\begin{array} { l l l l l } 524 & 526 & 520 & 523 & 530 \end{array}$$ Assuming a normal distribution, calculate a \(95 \%\) confidence interval for the population mean. Some adjustments are made to the machine. Assume that a normal distribution is still appropriate and that the population variance remains unchanged. A second random sample, this time of ten metal rods, is now taken. The results for hardness are as follows. $$\begin{array} { l l l l l l l l l l } 525 & 520 & 522 & 524 & 518 & 520 & 519 & 525 & 527 & 516 \end{array}$$ Stating suitable hypotheses, test at the \(10 \%\) significance level whether there is any difference between the population means before and after the adjustments.

Question 9:
AnswerMarks Guidance
Answer/WorkingMarks Guidance
Calculate sample mean: \(\bar{d} = 2623/5 = 524.6\)M1
Estimate population variance (1st sample): \(s_1^2 = (1376081 - 2623^2/5)/4 = [13.8\) or \(3.715^2]\)M1 allow biased: \(11.04\) or \(3.323^2\)
Find confidence interval (allow \(z\) in place of \(t\)): \(524.6 \pm t\sqrt{(13.8/5)}\)M1 inconsistent use of 4 or 5 loses M1
Correct tabular value: \(t_{4,\,0.975} = 2.776\) (2 d.p.)A1
Evaluate CI to 3 s.f.: \(524.6 \pm 4.6[1]\) or \([520.0,\;529.2]\)A1 needs correct \(s\), \(t\)
Subtotal: 5 marks
AnswerMarks Guidance
Answer/WorkingMarks Guidance
State hypotheses: \(H_0: \mu_b = \mu_a\), \(H_1: \mu_b \neq \mu_a\)B1
Estimate pop. variance (2nd sample): \(s_2^2 = (2720780 - 5216^2/10)/9 = [572/45\) or \(12.711\) or \(3.565^2]\)M1 allow biased: \(11.44\) or \(3.382^2\)
Pooled variance estimate: \(s^2 = (4 \times 13.8 + 9 \times 12.71)/13 = 848/65\) or \(13.05\)M1 A1
Calculate \(t\) (to 2 dp): \(t = (524.6 - 521.6)/(s\sqrt{(5^{-1} + 10^{-1})}) = 1.52\)M1 *A1
Compare tabular \(t\) value: \(t_{13,\,0.95} = 1.77[1]\)*B1
Conclusion: No difference in meansB1 AEF, dep *A1, *B1
Subtotal: 8 marksTotal: 13 marks
Question 10a:
AnswerMarks Guidance
Answer/WorkingMarks Guidance
Take moments about \(P\) for system [rod]: \(F_A h = Wa\cos\theta\)M1 A1
Take moments about \(B\) for rod: \(F_A h = R_A\cdot 2a\cos\theta\)M1 A1
Eliminate \(F_A\) to give \(R_A\): \(R_A = \frac{1}{2}W + (Wa\sin\theta)/h\)M1
Find inequality for \(\mu\): \(\mu \geq F_A/R_A\), \(\mu \geq 2a\cos\theta/(h + 2a\sin\theta)\)M1 A1 A.G.
Subtotal: 7 marks
Question 10a(i):
AnswerMarks Guidance
Answer/WorkingMarks Guidance
Use \(kW = T\) to express in terms of \(F_A\) or \(R_A\): \(kW = F_A/\sin\theta\) or \((W - R_A)/\cos\theta\)M1
Substitute for \(F_A\) or \(R_A\): \(k = (a/h)\cot\theta\) or \((\frac{1}{2} - (a/h)\sin\theta)/\cos\theta\)M1 A1
Substitute for \(h\) and \(\theta\): \(k = \sqrt{5}/6\) or \(0.373\)A1
Subtotal: 4 marks
Question 10a(ii):
AnswerMarks Guidance
Answer/WorkingMarks Guidance
Find horizontal component \(N_P\): \(N_P = T\sin\theta\) or \(kW\sin\theta\)M1
Substitute for \(k\) and \(\theta\): \((\sqrt{5}/6)(2/3)W = \sqrt{5}W/9\) or \(0.248W\)M1 A1
Subtotal: 3 marksTotal: 14 marks
Question 10b:
Part (i):
AnswerMarks Guidance
Answer/WorkingMark Guidance
Use regression line or 1st normal equation: \(\Sigma y/5 = 2.5\Sigma x/5 - 1.5\)B1
Use data to substitute for \(\Sigma x\) and \(\Sigma y\): \(11 + p + q = 2.5 \times 15 - 5 \times 1.5\)M1
\(p + q = 37.5 - 7.5 - 11 = 19\) A.G.A1
Total: 3 marks
Part (ii):
AnswerMarks Guidance
Answer/WorkingMark Guidance
Use formula for \(b\) or 2nd normal equation: \(2.5 = (32+2p+6q-15\times30/5)/(61-15^2/5)\)M2 A1
*or* \(32+2p+6q = 2.5\times61 - 1.5\times15\)A1
(A.E.F.) \(p + 3q = 49\) (*or* \(3q - p = 41\))
Solve any two simultaneous equations for \(p\), \(q\): \(p = 4,\ q = 15\)M1 A1
Total: 6 marks
Part (iii):
AnswerMarks Guidance
Answer/WorkingMark Guidance
\(r = (32+2p+6q-15\times30/5)/\sqrt{\{(61-15^2/5)(49+p^2+q^2-(11+p+q)^2/5)\}}\)M1
\(= (130 - 15\times30/5)/\sqrt{\{(61-15^2/5)(290-30^2/5)\}}\)A1
*or* \(2.5\sqrt{\{(61-15^2/5)/(49+p^2+q^2-(11+p+q)^2/5)\}}\)(M1)
\(= 2.5\sqrt{\{(61-15^2/5)/(290-30^2/5)\}}\)(A1)
\(= 40/\sqrt{(16\times110)}\) *or* \(2.5\sqrt{(16/110)} = 0.953\)A1
Total: 3 marks
Part (iv):
AnswerMarks Guidance
Answer/WorkingMark Guidance
(a) State equation of actual regression line: \(y = 0.25x - 1.5\)B1
(b) State new value of \(r\) or say unchanged: Same value as found in (iii)B1
Total: 2 marks
Question total: [14]
## Question 9:

| Answer/Working | Marks | Guidance |
|---|---|---|
| Calculate sample mean: $\bar{d} = 2623/5 = 524.6$ | M1 | |
| Estimate population variance (1st sample): $s_1^2 = (1376081 - 2623^2/5)/4 = [13.8$ or $3.715^2]$ | M1 | allow biased: $11.04$ or $3.323^2$ |
| Find confidence interval (allow $z$ in place of $t$): $524.6 \pm t\sqrt{(13.8/5)}$ | M1 | inconsistent use of 4 or 5 loses M1 |
| Correct tabular value: $t_{4,\,0.975} = 2.776$ (2 d.p.) | A1 | |
| Evaluate CI to 3 s.f.: $524.6 \pm 4.6[1]$ or $[520.0,\;529.2]$ | A1 | needs correct $s$, $t$ |

**Subtotal: 5 marks**

| Answer/Working | Marks | Guidance |
|---|---|---|
| State hypotheses: $H_0: \mu_b = \mu_a$, $H_1: \mu_b \neq \mu_a$ | B1 | |
| Estimate pop. variance (2nd sample): $s_2^2 = (2720780 - 5216^2/10)/9 = [572/45$ or $12.711$ or $3.565^2]$ | M1 | allow biased: $11.44$ or $3.382^2$ |
| Pooled variance estimate: $s^2 = (4 \times 13.8 + 9 \times 12.71)/13 = 848/65$ or $13.05$ | M1 A1 | |
| Calculate $t$ (to 2 dp): $t = (524.6 - 521.6)/(s\sqrt{(5^{-1} + 10^{-1})}) = 1.52$ | M1 *A1 | |
| Compare tabular $t$ value: $t_{13,\,0.95} = 1.77[1]$ | *B1 | |
| Conclusion: No difference in means | B1 | AEF, dep *A1, *B1 |

**Subtotal: 8 marks | Total: 13 marks**

---

## Question 10a:

| Answer/Working | Marks | Guidance |
|---|---|---|
| Take moments about $P$ for system [rod]: $F_A h = Wa\cos\theta$ | M1 A1 | |
| Take moments about $B$ for rod: $F_A h = R_A\cdot 2a\cos\theta$ | M1 A1 | |
| Eliminate $F_A$ to give $R_A$: $R_A = \frac{1}{2}W + (Wa\sin\theta)/h$ | M1 | |
| Find inequality for $\mu$: $\mu \geq F_A/R_A$, $\mu \geq 2a\cos\theta/(h + 2a\sin\theta)$ | M1 A1 | A.G. |

**Subtotal: 7 marks**

## Question 10a(i):

| Answer/Working | Marks | Guidance |
|---|---|---|
| Use $kW = T$ to express in terms of $F_A$ or $R_A$: $kW = F_A/\sin\theta$ or $(W - R_A)/\cos\theta$ | M1 | |
| Substitute for $F_A$ or $R_A$: $k = (a/h)\cot\theta$ or $(\frac{1}{2} - (a/h)\sin\theta)/\cos\theta$ | M1 A1 | |
| Substitute for $h$ and $\theta$: $k = \sqrt{5}/6$ or $0.373$ | A1 | |

**Subtotal: 4 marks**

## Question 10a(ii):

| Answer/Working | Marks | Guidance |
|---|---|---|
| Find horizontal component $N_P$: $N_P = T\sin\theta$ or $kW\sin\theta$ | M1 | |
| Substitute for $k$ and $\theta$: $(\sqrt{5}/6)(2/3)W = \sqrt{5}W/9$ or $0.248W$ | M1 A1 | |

**Subtotal: 3 marks | Total: 14 marks**

# Question 10b:

## Part (i):
| Answer/Working | Mark | Guidance |
|---|---|---|
| Use regression line or 1st normal equation: $\Sigma y/5 = 2.5\Sigma x/5 - 1.5$ | B1 | |
| Use data to substitute for $\Sigma x$ and $\Sigma y$: $11 + p + q = 2.5 \times 15 - 5 \times 1.5$ | M1 | |
| $p + q = 37.5 - 7.5 - 11 = 19$ **A.G.** | A1 | |
| **Total: 3 marks** | | |

## Part (ii):
| Answer/Working | Mark | Guidance |
|---|---|---|
| Use formula for $b$ or 2nd normal equation: $2.5 = (32+2p+6q-15\times30/5)/(61-15^2/5)$ | M2 A1 | |
| *or* $32+2p+6q = 2.5\times61 - 1.5\times15$ | A1 | |
| (A.E.F.) $p + 3q = 49$ (*or* $3q - p = 41$) | | |
| Solve any two simultaneous equations for $p$, $q$: $p = 4,\ q = 15$ | M1 A1 | |
| **Total: 6 marks** | | |

## Part (iii):
| Answer/Working | Mark | Guidance |
|---|---|---|
| $r = (32+2p+6q-15\times30/5)/\sqrt{\{(61-15^2/5)(49+p^2+q^2-(11+p+q)^2/5)\}}$ | M1 | |
| $= (130 - 15\times30/5)/\sqrt{\{(61-15^2/5)(290-30^2/5)\}}$ | A1 | |
| *or* $2.5\sqrt{\{(61-15^2/5)/(49+p^2+q^2-(11+p+q)^2/5)\}}$ | (M1) | |
| $= 2.5\sqrt{\{(61-15^2/5)/(290-30^2/5)\}}$ | (A1) | |
| $= 40/\sqrt{(16\times110)}$ *or* $2.5\sqrt{(16/110)} = 0.953$ | A1 | |
| **Total: 3 marks** | | |

## Part (iv):
| Answer/Working | Mark | Guidance |
|---|---|---|
| **(a)** State equation of actual regression line: $y = 0.25x - 1.5$ | B1 | |
| **(b)** State new value of $r$ or say unchanged: Same value as found in (iii) | B1 | |
| **Total: 2 marks** | | |

**Question total: [14]**
9 A random sample of five metal rods produced by a machine is taken. Each rod is tested for hardness. The results, in suitable units, are as follows.

$$\begin{array} { l l l l l } 
524 & 526 & 520 & 523 & 530
\end{array}$$

Assuming a normal distribution, calculate a $95 \%$ confidence interval for the population mean.

Some adjustments are made to the machine. Assume that a normal distribution is still appropriate and that the population variance remains unchanged. A second random sample, this time of ten metal rods, is now taken. The results for hardness are as follows.

$$\begin{array} { l l l l l l l l l l } 
525 & 520 & 522 & 524 & 518 & 520 & 519 & 525 & 527 & 516
\end{array}$$

Stating suitable hypotheses, test at the $10 \%$ significance level whether there is any difference between the population means before and after the adjustments.

\hfill \mbox{\textit{CAIE FP2 2011 Q9 [13]}}