OCR S2 2013 June — Question 7 11 marks

Exam BoardOCR
ModuleS2 (Statistics 2)
Year2013
SessionJune
Marks11
PaperDownload PDF ↗
Mark schemeDownload PDF ↗
TopicApproximating Binomial to Normal Distribution
TypeOne-tailed hypothesis test
DifficultyStandard +0.3 This is a straightforward one-tailed hypothesis test using normal approximation to binomial with continuity correction. Students must check np>5 and nq>5, set up H₀: p=0.35 vs H₁: p>0.35, apply continuity correction (49.5), calculate z-score, and compare to critical value (1.282 at 10% level). While it requires multiple steps, each is standard procedure taught explicitly in S2 with no novel problem-solving required, making it slightly easier than average.
Spec2.04d Normal approximation to binomial5.04b Linear combinations: of normal distributions5.05c Hypothesis test: normal distribution for population mean

7 Past experience shows that \(35 \%\) of the senior pupils in a large school know the regulations about bringing cars to school. The head teacher addresses this subject in an assembly, and afterwards a random sample of 120 senior pupils is selected. In this sample it is found that 50 of these pupils know the regulations. Use a suitable approximation to test, at the \(10 \%\) significance level, whether there is evidence that the proportion of senior pupils who know the regulations has increased. Justify your approximation.

Question 7:
AnswerMarks Guidance
AnswerMarks Guidance
\(H_0: p = 0.35\); \(H_1: p > 0.35\)B2 One error (e.g. \(\mu\), no symbol, 2-tailed): B1, but \(\bar{x}\), \(t\) etc: B0. Allow \(\pi\)
\(B(120, 0.35)\)M1 \(B(120, 0.35)\) stated or implied
\(\approx N(42,\ 27.3)\)M1 \(N(np,\ npq)\), their attempt at \(120 \times 0.35\). \(120 \times 0.35 \times 0.65\) *Not* \(N(np,\ nq)\)
\(\alpha\): \(z = \dfrac{49.5 - 42}{\sqrt{27.3}} = 1.435 > 1.282\) [or \(0.0757 < 0.1\)]A1 ft A1 ft Standardise with their \(np\) and \(\sqrt{npq}\), right cc. Allow both 49.5 and 50.5 and both in CR. \(z\) in range \([1.43, 1.44]\) before rounding. Comparison with 1.282, ft on \(z/p\) or \(\sqrt{120}\)
\(\beta\): \(CV = 42.5 + 1.282 \times \sqrt{27.3} = 49.198\); \(z = 1.282\) and compare 50; \(CR \geq 50\) or \(\geq 49.2\)A1 ft A1 A1 ft CV \(42.5 + z\sqrt{27.3}\), ignore LH, ft on \(np\), \(npq\). \(z = 1.282\) used in RH CV and compare 50. CV correct ft on \(z\), but don't worry about \(\geq\). Must round up: 49 from 49.2: A1A1A0
Reject \(H_0\)M1 Consistent first conclusion, needs correct method and comparison. Can give M1A1 even if comparison not explicit. Allow from exact binomial
Significant evidence that proportion who know regulations has increasedA1 ft Contextualised, needs "who know regulations" or "pupils", and "evidence"
\(np > 5\) \([= 42]\) from normal attemptedM1 From \(p = 0.35\) or 5/12, don't need 42. *or* \(n\) large or \(p\) close to 0.5 asserted
\(nq = 78 > 5\) and no others apart from \(n\) largeA1 Need 78, or 70 from 5/12, *not* \(npq\). *and* the other qualitative reason asserted
Total: 11
# Question 7:
| Answer | Marks | Guidance |
|--------|-------|----------|
| $H_0: p = 0.35$; $H_1: p > 0.35$ | B2 | One error (e.g. $\mu$, no symbol, 2-tailed): B1, but $\bar{x}$, $t$ etc: B0. Allow $\pi$ |
| $B(120, 0.35)$ | M1 | $B(120, 0.35)$ stated or implied |
| $\approx N(42,\ 27.3)$ | M1 | $N(np,\ npq)$, their attempt at $120 \times 0.35$. $120 \times 0.35 \times 0.65$ *Not* $N(np,\ nq)$ |
| $\alpha$: $z = \dfrac{49.5 - 42}{\sqrt{27.3}} = 1.435 > 1.282$ [or $0.0757 < 0.1$] | A1 ft A1 ft | Standardise with their $np$ and $\sqrt{npq}$, right cc. Allow both 49.5 and 50.5 and both in CR. $z$ in range $[1.43, 1.44]$ before rounding. Comparison with 1.282, ft on $z/p$ or $\sqrt{120}$ |
| $\beta$: $CV = 42.5 + 1.282 \times \sqrt{27.3} = 49.198$; $z = 1.282$ and compare 50; $CR \geq 50$ or $\geq 49.2$ | A1 ft A1 A1 ft | CV $42.5 + z\sqrt{27.3}$, ignore LH, ft on $np$, $npq$. $z = 1.282$ used in RH CV and compare 50. CV correct ft on $z$, but don't worry about $\geq$. Must round up: 49 from 49.2: A1A1A0 |
| Reject $H_0$ | M1 | Consistent first conclusion, needs correct method and comparison. Can give M1A1 even if comparison not explicit. Allow from exact binomial |
| Significant evidence that proportion who know regulations has increased | A1 ft | Contextualised, needs "who know regulations" or "pupils", and "evidence" |
| $np > 5$ $[= 42]$ from normal attempted | M1 | From $p = 0.35$ or 5/12, don't need 42. *or* $n$ large or $p$ close to 0.5 asserted |
| $nq = 78 > 5$ and no others apart from $n$ large | A1 | Need 78, or 70 from 5/12, *not* $npq$. *and* the other qualitative reason asserted |
| | **Total: 11** | |

---
7 Past experience shows that $35 \%$ of the senior pupils in a large school know the regulations about bringing cars to school. The head teacher addresses this subject in an assembly, and afterwards a random sample of 120 senior pupils is selected. In this sample it is found that 50 of these pupils know the regulations. Use a suitable approximation to test, at the $10 \%$ significance level, whether there is evidence that the proportion of senior pupils who know the regulations has increased. Justify your approximation.

\hfill \mbox{\textit{OCR S2 2013 Q7 [11]}}