📚 Interdisciplinary Applied Statistics for Year 12 OCR | 跨学科综合题型训练
The OCR Year 12 statistics syllabus gives you powerful tools – probability models, binomial and normal distributions, hypothesis testing – that are used across science and social science. In this article we build interdisciplinary fluency by applying these skills to realistic scenarios from biology, economics, psychology, physics and more. Each section presents a contextual problem, step-by-step analysis and full interpretation, exactly the sort of synthesis required by modern cross-topic exam questions.
OCR 十二年级统计大纲为你提供了强大的工具——概率模型、二项分布、正态分布和假设检验——这些在自然科学和社会科学中都有广泛应用。本文通过将技能应用到生物、经济、心理、物理等真实跨学科场景中，培养你的综合解题能力。每一节都给出情境问题、逐步分析和完整解读，这正是现代跨主题综合题所要求的。
1. Biology: Hardy–Weinberg Equilibrium and Binomial Sampling | 生物学：哈代-温伯格平衡与二项抽样
The Hardy– gene with dominant allele A (frequency p) and recessive allele a (frequency q, with q = 1 – number of homozygous recessive individuals can be modelled as a binomial random variable X ~ B(n, q²).
哈代-温伯格原理用于建模大群体随机交配下的等位基因和基因型频率。对于显性等位基因 A（频率 p）和隐性等位基因 a（频率 q，q = 1 – p），期望基因型比例为 AA = p²，Aa = 2pq，aa = q²。从群体中随机抽取 n 个个体，隐性纯合子个体数量可以建模为二项随机变量 X ~ B(n, q²)。
Example: In a population, 4% of individuals express a recessive trait (aa). A researcher takes a random sample of 25 individuals. Find the probability that exactly 2 show the trait and test whether the observed number gives evidence that the frequency has changed.
示例：某群体中有 4% 的个体表现出隐性性状（aa）。研究人员随机抽取 25 人。求恰好有 2 人表现出该性状的概率，并检验观察值是否提供了频率发生变化的证据。
Step 1: If 4% are aa, then q² = 0.04, so q = 0.2 and p = 0.8. The number of aa in a sample of 25 is X ~ B(25, 0.04).
步骤 1：如果 4% 为 aa，则 q² = 0.04，因此 q = 0.2，p = 0.8。样本量为 25 时，aa 个体数 X ~ B(25, 0.04)。
)²³. Using a calculator: ²⁵C₂ = 300, (0.04)² = 0.0016, (0.96)²³ ≈ 0.3879, product ≈ 0.186.
P(X = 2) = ²⁵C₂ × (0.04)² × (0.96)²³。使用计算器：²⁵C₂ = 300，(0.04)² = 0.0016，(0.96)²³ ≈ 0.3879，乘积约为 0.186。
ore extreme. From binomial tables, P(X ≤ 2) ≈ 0.186 + P(X=1)+P(X=0). This is well above any usual significance level, so there is insufficient evidence to reject H₀.
步骤 2：假设检验，H₀: q² = 0.04 对 H₁: q² ≠ 0.04。观察值 X = 2，我们通过求两端更极端概率之和计算双侧 p 值。查表，P(X ≤ 2) 约 0.186 加上 P(X=1)+P(X=0) 后仍远大于常用显著性水平，因此没有足够证据拒绝 H₀。
2. Economics: Market Demand as a Binomial Trial | 经济学：市场需求的二项试验
A company launches a niche product and believes that 35% of target customers will purchase it. In a focus group of 30 potential customers, the number of buyers can be modelled as X ~ B(30, 0.35). Analysts use this model to carry out a hypothesis test on whether a new advertising campaign has increased purchase intent.
一家公司推出一款利基产品，认为 35% 的目标客户会购买。在一个 30 名潜在客户的焦点小组中，购买人数可建模为 X ~ B(30, 0.35)。分析人员使用此模型对新的广告活动是否提升了购买意愿进行假设检验。
Suppose after the campaign 15 customers indicate they will buy. Test, at the 5% significance level, whether the proportion has increased.
假设广告活动后有 15 人表示会购买。在 5% 显著性水平下检验比例是否上升。
Define X ~ B(30, 0.35). H₀: p = 0.35, H₁: p > 0.35. Test statistic: X = 15. We need P(X ≥ 15 | p=0.35).
定义 X ~ B(30, 0.35)。H₀: p = 0.35，H₁: p > 0.35。检验统计量：X = 15。需要计算 P(X ≥ 15 | p=0.35)。
Using cumulative binomial probabilities, P(X ≥ 15) = 1 – .0271 < 0.05, we reject H₀. There is evidence that the campaign increased purchase intent.
利用二项累积概率，P(X ≥ 15) = 1 – P(X ≤ 14)。查表或使用软件，P(X ≤ 14) ≈ 0.9729，因此 p 值 ≈ 0.0271。因为 0.0271 < 0.05，拒绝 H₀。有证据表明该广告活动提升了购买意愿。
3. Psychology: Reaction Times and the Normal Distribution | 心理学：反应时间与正态分布
s use this model to select individuals for a study requiring exceptionally fast responses.
对视觉刺激的反应时间服从均值为 μ = 250 毫秒、标准差 σ = 30 毫秒的正态分布。心理学家使用该模型筛选出反应特别快的人参与研究。
What is the probability that a randomly chosen participant has a reaction time less than 200 ms? If the fastest 10% are selected, what reaction time cuts off the 10th percentile?
随机选取一名参与者，其反应时间小于 200 毫秒的概率是多少？如果要选取最快的 10%，对应的截断反应时间是多少？
Let R ~ N(250, 30²). For R < 200, z = (200 – 250)/30 = –1.6667. Φ(–1.67) ≈ 0.0475. So about 4.75% of people react faster than 200 ms.
设 R ~ N(250, 30²)。对于 R < 200，z = (200 – 250)/30 = –1.6667。Φ(–1.67) ≈ 0.0475。因此约 4.75% 的人反应快于 200 毫秒。
For the fastest 10%, we find the 10th percentile, i.e. the z-score with left-tail area 0.10, which is z ≈ –1.2816. Then R = μ + zσ = 250 – 1.2816×30 ≈ 211.6 ms. Participants must be faster than 211.6 ms to be in the top 10%.
对于最快的前 10%，需要找到第 10 百分位数，即左尾面积为 0.10 的 z 值，z ≈ –1.2816。然后 R = μ + zσ = 250 – 1.2816×30 ≈ 211.6 毫秒。参与者必须快于 211.6 毫秒才能进入前 10%。
4. Medicine: Diagnostic Testing and Conditional Probability | 医学：诊断检验与条件概率
A disease occurs in 1% of a population. A test for the disease has sensitivity 98% (true positive rate) and specificity 95% (true negative rate). When a person tests positive, what is the probability they actually have the disease? This classic Bayes' theorem problem is a vital statistics application in medical decision-making.
某种疾病在人群中的发病率为 1%。一种检测方法的灵敏度为 98%（真阳性率），特异度为 95%（真阴性率）。当一个人检测呈阳性时，其真正患病的概率是多少？这道经典的贝叶斯定理题目是统计学在医疗决策中的重要应用。
Let D be the event 'has disease', P(D) = 0.01. P(+|D) = 0.98, P(–|not D) = 0.95, so P(+|not D) = 0.05. By Bayes' theorem, P(D|+) = [P(+|D)P(D)] / [P(+|D)P(D) + P(+| 0.01 + 0.05×0.99) = 0.0098 / (0.0098 + 0.0495) = 0.0098/0.0593 ≈ 0.165. So only about 16.5% sitive results are true positives, highlighting the importance of base rates in screening.
设 D 为“患病”事件，P(D) = 0.01。P(+|D) = 0.98，P(–|未患病) = 0.95，所以 P(+|未患病) = 0.05。根据贝叶斯定理，P(D|+) = [P(+|D)P(D)] / [P(+|D)P(D) + P(+|未患病)P(未患病) (0.98×0.01) / (0.98×0.01 + 0.05×0.99) = 0.0098 / (0.0098 + 0.0495) = 0.0098/0.0593 ≈ 0.165。因此只有约 16.5% 的阳性结果为真阳性，这凸显了筛查中基础比率的重要性。
We can also represent the data in a contingency table per 10,000 people to verify the result.
我们也可以用每 10 000 人的列联表来验证结果。
| Status | Test + | Test – | Total |
|---|---|---|---|
| Disease | 98 | 2 | 100 |
| No disease | 495 | 9405 | 9900 |
| Total | 593 | 9407 | 10000 |
From the table, P(D|+) = 98/593 ≈ 0.165.
由上表，P(D|+) = 98/593 ≈ 0.165。
5. Physics: Measurement Errors and Normal Modelling | 物理学：测量误差与正态建模
In a physics lab, repeated measurements of the acceleration due to gravity g are taken with an instrument known to have random error distributed normally with mean 0 and standard deviation 0.05 m/s². If the accepted value is 9.81 m/s², what is the probability that a single measurement deviates from the true value by more than 0.1 m/s²? Physicists use such probability to set tolerance limits.
在物理实验室中，用一台已知随机误差服从均值为 0、标准差为 0.05 m/s² 正态分布的仪器重复测量重力加速度 g。如果公认值为 9.81 m/s²，单次测量偏离真值超过 0.1 m/s² 的概率是多少？物理学家用此概率设定容差界限。
Measurement M ~ N(9.81, 0.05²). The absolute deviation |M – 9.81| > 0.1 corresponds to M > 9.91 or M < 9.71. Standardising: z = 0.1/0.05 = 2. So we need P(|Z| > 2) = 2 × Φ(–2) = 2 × 0.0228 = 0.0456. The chance is about 4.6%.
测量值 M ~ N(9.81, 0.05²)。绝对偏差 |M – 9.81| > 0.1 对应于 M > 9.91 或 M < 9.71。标准化：z = 0.1/0.05 = 2。于是需求 P(|Z| > 2) = 2 × Φ(–2) = 2 × 0.0228 = 0.0456。这一概率约为 4.6%。
If a physicist takes 4 independent measurements and averages them, the standard deviation of the sample mean becomes σ/√n 2 = 0.025. The probability that the average differs from 9.81 by more than 0.05 becomes P(|Z| > 0.05/0.025=2) = 0.0456 again — but for a smaller margin. This shows how averaging improves precision.
如果物理学家进行 4 次独立测量并取平均值，样本均值的标准差变为 σ/√n = 0.05/2 = 0.025。平均值与 9.81 相差超过 0.05 的概率为 P(|Z| > 0.05/0.025=2) = 0.0456 —— 虽然概率相同，但允许的误差边界更小，说明平均化提高了精度。
6. Geography: Flood Frequency and Normal Approximation | 地理学：洪水频率与正态近似
A river gauge records days each year when water level exceeds a critical mark. Over a 100-year period, such exceedance days are recorded. Assume each year's number of exceedances is independent and has a constant probability, but we model the total over a century with a normal approximation to the binomial. Suppose the annual probability of at least one exceedance day is 0.2, and we are interested in the total number of years with exceedance in 100 years. Use a normal approximation to find the approximate probability that at most 15 years see such flooding.
一个河流水位计记录每年水位超过临界线的天数。在 100 年间记录超标天数。假设每年是否出现超标独立且概率恒定，但我们用正态近似来建模百年中超标年份总数。假设每年至少出现一次超标天的概率为 0.2，我们想知道 100 年中至多有 15 年发生此种洪水的近似概率。
Let Y ~ = np(1-p) = 16, σ = 4. Using continuity correction, P(Y ≤ 15) ≈ P(Y ≤ 15.5) = P(Z ≤ (15.5 – 20)/4) = Φ(–1.125) ≈ 0.130. So about 13% chance of observing 15 or fewer flood years.
设 Y ~ B(100, 0.2) 为发生洪水的年数。μ = np = 20，方差 σ² = np(1-p) = 16，σ = 4。使用连续性校正，P(Y ≤ 15) ≈ P(Y ≤ 15.5) = P(Z ≤ (15.5 – 20)/4) = Φ(–1.125) ≈ 0.130。因此观察到不超过 15 年洪水的概率约为 13%。
Geographers compare observed flood decades with this distribution to detect climate-related changes.
地理学家将实际观测的洪水年代与此分布对比，以检测气候变化带来的异常。
7. Business: Quality Control via Binomial Hypothesis Testing | 商业：二项假设检验与质量控制
A machine produces components and the manufacturer claims that at most 5% are defective. A quality control inspector samples 50 components and finds 6 defectives. Test, at the 5% significance level, whether the defect rate exceeds 5%.
一台机器生产元件，制造商声称次品率不超过 5%。质检员抽取 50 个元件，发现 6 个次品。在 5% 显著性水平下检验次品率是否超过 5%。
X ~ B(50, 0.05). H₀: p = 0.05, H₁: p > 0.05. Observed X = 6. P(X ≥ 6) = 1 – .05, there is sufficient evidence to reject H₀ and conclude the defect rate has increased.
X ~ B(50, 0.05)。H₀: p = 0.05，H₁: p > 0.05。观察值 X = 6。P(X ≥ 6) = 1 – P(X ≤ 5)。查二项分布表，P(X ≤ 5) ≈ 0.9622，故 p 值 ≈ 0.0378。由于 0.0378 < 0.05，有足够证据拒绝 H₀，推断次品率已上升。
A critical region approach: find the smallest c such that P(X ≥ c) ≤ 0.05. P(X ≥ 7) = 1 – P(X ≤ 6) ≈ 1 – 05. So the critical region is X ≥ 6, and our observed 6 falls in it, confirming rejection.
临界区方法：找到最小 c 使得 P(X ≥ c) ≤ 0.05。P(X ≥ 7) ≈ 0.0118 < 0.05，P(X ≥ 6) ≈ 0.0378 < 0.05。因此临界区为 X ≥ 6，而我们的观察值 6 落入该区域，再次确认拒绝 H₀。
8. Sports Science: Performance Consistency and Cumulative Probabilities | 体育科学：表现稳定性与累积概率
A basketball player's free-throw success probability is 0.8. In a game, she takes 10 free throws; her team's success often hinges on getting at least 8 baskets. Model the number of successes as X ~ B(10, 0.8) and find the probability she scores exactly 8, and the probability she scores at least 8.
一名篮球运动员的罚球命中概率为 0.8。一场比赛中她罚球 10 次；球队的胜利往往需要她至少命中 8 球。将命中次数建模为 X ~ B(10, 0.8)，求她恰好命中 8 球的概率，以及至少命中 8 球的概率。
= 10) = 0.1074. Summing gives P(X ≥ 8) ≈ 0.302 + 0.268 + 0.107 = 0.677. So there is a 67.7% chance she achieves the required target.
P(X = 8) = ¹⁰C₈ × (0.8)⁸ × (0.2)² = 45 × 0.1678 × 0.04 ≈ 0.302。P(X = 9) = 10 × 0.1342 × 0.2 ≈ 0.268。P(X = 10) = 0.1074。求和得 P(X ≥ 8) ≈ 0.302 + 0.268 + 0.107 = 0.677。因此她有 67.7% 的概率达到所需目标。
Coaches use such probabilities to decide whether to rest the player or design plays that lead to higher-percentage shots.
教练利用这类概率决定是否让球员休息或设计更高命中率的投篮战术。
9. Environmental Science: Detecting Rare Events with Poisson Approximation? (Binomial Approach) | 环境科学：稀有事件的检测（用二项分布处理）
Water samples are taken from a lake and each sample is tested for a harmful algal toxin. Historically, 2% of samples test positive. In a new monitoring round, 120 samples are taken. The binomial distribution models the number of positives. Use binomial calculations (and a normal approximation for practice) to find the probability that 5 or more samples test positive, and discuss whether this suggests an outbreak.
从湖中取水样，每个样本都检测一种有害藻毒素。历史上，2% 的样本呈阳性。在新一轮监测中，采集了 120 个样本。用二项分布建模阳性样本数。使用二项计算（并练习正态近似）求 5 个或更多样本呈阳性的概率，并讨论这是否暗示爆发。
X ~ B(120, 0.02). Exact P(X ≥ 5) = 1 – . P(X ≤ 4) from binomial formula or tables ≈ 0.056 + 0.138 + 0.169 + 0.138 + 0.084 = 0.585? Wait, let' X=3)≈0.215, P(X=4)≈0.130. Sum≈0.918. So P(X ≥ 5) ≈ 0.082. This is a fairly low probability – observing 5 or more would be unusual and may trigger an investigation.
X ~ B(120, 0.02)。精确 P(X ≥ 5) = 1 – P(X ≤ 4)。由二项公式或查表：P(X=0)=(0.98)¹²⁰≈0.089，P(X=1)=120×0.02×(0.98)¹¹⁹≈0.218，P(X=2)≈0.266，P(X=3)≈0.215，P(X=4)≈0.130。合计≈0.918，因此 P(X ≥ 5) ≈ 0.082。该概率相当低——观察到 5 个或更多阳性是不寻常的，可能触发调查。
Normal approximation check: μ = np = 2.4, σ = √2.352 ity correction, P(X ≥ 5) ≈ P(X ≥ 4.5) = P(Z ≥ (4.5-2.4)/1.53 ≈ 1.37) ≈ 0.0853, very close.
正态近似检验：μ = 2.4，σ = √2.352 ≈ 1.53。使用连续性校正，P(X ≥ 5) ≈ P(X ≥ 4.5) = P(Z ≥ 1.37) ≈ 0.0853，非常接近。
10. Social Policy: Survey Analysis and Confidence Concepts (Normal Distribution) | 社会政策：调查分析与置信概念（正态分布）
5. Assuming the standard deviation of the sampling distribution can be approximated by √(p̂(1-p̂)/n), calculate an approximate 95% l for the true proportion. Interpret the interval and discuss the assumption of normality.
一家政策智库随机调查了 400 位市民，以估计支持新提案的比例。样本比例为 0.55。假定抽样分布的标准差可用 √(p̂(1-p̂)/n) 近似，计算真实比例的一个近似 95% 置信区间。解读此区间并讨论正态性假设。
Let p̂ = 0.55, n=400. Standard error SE ≈ √0.55×0.45/400 = √0.00061875 ≈ 0.02487. For 95% confidence, z* = 1.96. The interval is p̂ SE = 0.55 ± 1.96×0.02487 = 0.55 ± 0.0487, giving (0.5013, 0.5987). We are approximately 95% confident that the true proportion lies between 50.1% and 59.9%.
令 p̂ = 0.55，n=400。标准误 SE ≈ √0.55×0.45/400 = √0.00061875 ≈ 0.02487。对于 95% 置信度，z* = 1.96。区间为 p̂ ± 1.96 × SE = 0.55 ± 0.0487，得 (0.5013, 0.5987)。我们有约 95% 的把握认为真实比例在 50.1% 到 59.9% 之间。
The normal approximation is valid because np̂ = 220 > 10 and n(1-p̂ s an approximate method; in Year 12 this provides a foundation for later formal inference.
正态近似有效，因为 np̂ = 220 > 10 且 n(1-p̂)=180 > 10。但该区间使用了估计的标准误，因此为近似方法；在十二年级阶段这为后续正式的推断打下基础。
Published by TutorHao | Statistics Revision Series | aleveler.com
更多咨询请联系16621398022（同微信）
或截图后用「扫一扫 → 相册」选择本图