| Exam Board | OCR MEI |
|---|---|
| Module | S1 (Statistics 1) |
| Year | 2013 |
| Session | June |
| Marks | 8 |
| Paper | Download PDF ↗ |
| Mark scheme | Download PDF ↗ |
| Topic | Hypothesis test of binomial distributions |
| Type | One-tailed hypothesis test (upper tail, H₁: p > p₀) |
| Difficulty | Moderate -0.3 This is a straightforward one-tailed binomial hypothesis test with standard setup. Parts (i) and (ii) test understanding of hypothesis formulation (guessing = 0.5 probability, researcher suspects people do no better so tests if they do better). Part (iii) requires calculating P(X ≥ 13) under B(20, 0.5) and comparing to 5%, which is routine application of binomial tables or calculation. Slightly easier than average due to clear context and standard procedure. |
| Spec | 2.05a Hypothesis testing language: null, alternative, p-value, significance2.05b Hypothesis test for binomial proportion2.05c Significance levels: one-tail and two-tail |
| Answer | Marks | Guidance |
|---|---|---|
| Marks: E1 | E1 | [2] |
| Answer | Marks |
|---|---|
| Marks: B1 | [1] |
| Answer | Marks | Guidance |
|---|---|---|
| Marks: M1 | B1* | M1* dep |
| Answer | Marks | Guidance |
|---|---|---|
| Marks: B1 | M1* | M1* cao dep on the two correct probabilities |
## Part (i)
**Answer:**
Because if people cannot make a correct identification, then the probability that they guess correctly will be 0.5
For 'equally likely to guess right or wrong' or 'two outcomes with equal probability' or '50:50 chance of success' or 'right one in two occasions on average' or 'two (equally likely) outcomes' etc
**Marks:** E1 | E1 | [2]
**Guidance:**
- E1: For idea of a guess or 'chosen at random'
- E1: For idea of two outcomes. NB The question includes the sentence 'She suspects that people do no better than they would by guessing.', so this on its own does not get the mark for the idea of a guess
## Part (ii)
**Answer:**
'Because people may do better than they would by guessing' or similar
**Marks:** B1 | [1]
**Guidance:**
- B1: For idea of selecting correctly/identifying/knowing. No marks if answer implies that it is because there are over half in the sample who make a correct identification
## Part (iii)
**Answer:**
$$P(X \geq 13) = 1 - P(X \leq 12) = 1 - 0.8684 = 0.1316$$
NB PLEASE ANNOTATE THE TOP AND BOTTOM OF THE EXTRA PAGE IF NOT USED
$$0.1316 > 0.05$$
So not significant
There is insufficient evidence to suggest that people can make a correct identification.
**Marks:** M1 | B1* | M1* dep | A1* | E1* dep | [5]
**Guidance:**
- M1: For notation P(X ≥ 13) or P(X > 12) or 1– P(X ≤ 12)
- B1*: For 0.1316
- M1* dep: For comparison with 5%
- A1*: Allow 'accept H₀' or 'reject H₁'
- E1* dep: NB Point probabilities score zero. Must include 'insufficient evidence' or something similar such as 'to suggest that' ie an element of doubt either in the A or E mark. Must be in context to gain E1 mark. Do not allow 'sufficient evidence to suggest proportion making correct identification is 0.5' or similar
**ALTERNATIVE METHOD – follow method above unless some mention of CR seen**
**Critical region method**
UPPER TAIL
$$P(X \geq 14) = 1 - P(X \leq 13) = 1 - 0.9423 = 0.0577 > 5\%$$
$$P(X \geq 15) = 1 - P(X \leq 14) = 1 - 0.9793 = 0.0207 < 5\%$$
So critical region is {15,16,17,18,19,20}
13 not in CR so not significant
There is insufficient evidence to indicate that people can make a correct identification.
**Marks:** B1 | M1* | M1* cao dep on the two correct probabilities | A1* | E1* dep on A1 | [5]
**Guidance:**
- B1: For either probability
- M1*: For a correct comparison with 5%
- M1* cao dep: No marks if CR not justified. Condone {15, …, 20}, X ≥ 15, oe but not P(X ≥ 15,) etc
- A1*: Must include '13 not in CR'
- E1* dep: Ignore any work on lower critical region. NB If CR found correctly, then P(X=13) subsequently found, but cand says '13 not in CR' then allow up to all five marks. If do not say '13 not in CR' allow no marks
---
5 A researcher is investigating whether people can identify whether a glass of water they are given is bottled water or tap water. She suspects that people do no better than they would by guessing. Twenty people are selected at random; thirteen make a correct identification. She carries out a hypothesis test.\\
(i) Explain why the null hypothesis should be $p = 0.5$, where $p$ represents the probability that a randomly selected person makes a correct identification.\\
(ii) Briefly explain why she uses an alternative hypothesis of $p > 0.5$.\\
(iii) Complete the test at the $5 \%$ significance level.
\hfill \mbox{\textit{OCR MEI S1 2013 Q5 [8]}}