📚 Interdisciplinary Problem-Solving in CIE A-Level Statistics | 跨学科综合题型训练
Interdisciplinary problem-solving is a cornerstone of the CIE A-Level Statistics syllabus, requiring students to apply statistical concepts to real-world contexts spanning physics, biology, economics, and more. This article offers a comprehensive training set of mixed-format questions that blend statistical theory with practical scenarios, helping Year 13 learners sharpen their analytical skills and examination technique.
跨学科综合题型是CIE A-Level统计课程的核心要求之一，学生需要将统计概念应用于物理、生物、经济等现实情境。本文提供一套完整的综合题型训练，将统计理论与实际场景相融合，帮助Year 13学生提升分析能力和应试技巧。
1. Error Analysis in Physics | 物理学中的误差分析
When measuring the acceleration due to gravity g using a simple pendulum, repeated trials yield a sample mean of 9.78 m s⁻² and an unbiased estimate of the population variance of 0.015 m² s⁻⁴ from 30 trials. Assume the measurements are normally distributed. Construct a 98% confidence interval for the true value of g, and interpret whether the accepted value of 9.81 m s⁻² lies within this interval.
使用单摆测量重力加速度g时，30次重复实验得到样本均值9.78 m s⁻²和无偏总体方差估计值0.015 m² s⁻⁴。假设测量数据服从正态分布，构建g真实值的98%置信区间，并判断公认值9.81 m s⁻²是否落在该区间内。
The standard error is s/√n = √0.015/√30 ≈ 0.02236. With 29 degrees of freedom, the t-value for a 98% side this interval, the experimental result is consistent with the accepted value at the 2% significance level.
标准误为s/√n = √0.015/√30 ≈ 0.02236。自由度为29时，98%置信水平的t值约为2.462。区间为9.78 ± 2.462 × 0.02236，即(9.725, 9.835)。9.81落入该区间，因此在2%显著性水平下实验结果与公认值一致。
2. Poisson Distribution in Cell Counting | 生物中的泊松分布与细胞计数
A biologist examines yeast cells under a microscope on a haemocytometer grid. The average number of cells per 0.1 mm² square is 4.2, and the cells are randomly distributed. Find the probability that a randomly chosen square contains exactly 3 cells. If the biologist examines 25 squares, approximate the probability that the total number of cells observed is at least 110.
一位生物学家在血球计数板的方格网下观察酵母细胞。每个0.1 mm²方格中的平均细胞数为4.2，且细胞随机分布。求随机选取的一个方格中恰好有3个细胞的概率。若生物学家检查25个方格，近似计算观察到的细胞总数至少为110的概率。
For a single square, X ~ Po(4.2), P(X=3) = e⁻⁴'² × 4.2³ / 3! ≈ 0.1854. For 25 squares, the total Y ~ Po(25 × 4.2) = Po(105). Using a normal approximation, Y ~ ontinuity correction: z = (109.5 − 105)/√105 ≈ 0.439, giving a probability of about 0.330.
单个方格X ~ Po(4.2)，P(X=3) = e⁻⁴'² × 4.2³ / 3! ≈ 0.1854。25个方格的总数Y ~ Po(25 × 4.2) = Po(105)。用正态近似Y ~ N(105, 105)，连续性修正后P(Y ≥ 110) = P(Y > 109.5)，z = (109.5 − 105)/√105 ≈ 0.439，概率约为0.330。
3. Regression Analysis in Economics | 经济模型中的回归与相关
10 households. The summary statistics are: Σx = 620, Σy = 410, Σx² = 40 200, Σy² = 18 900, Σxy = 27 560. Calculate the equation of the least squares regression line of y on x and the product moment correlation coefficient. Interpret the slope in context.
一位经济学家收集了10户家庭的收入(x，千美元)与奢侈品年支出(y，百美元)数据。汇总统计量为：Σx = 620, Σy = 410, Σx² = 40 200, Σy² = 18 900, Σxy = 27 560。计算y对x的最小二乘回归线方程和积矩相关系数，并结合背景解释斜率。
Sxx = 40200 − 620²/10 = 40200 − 38440 = 1760
Syy = 18900 − 410²/10 = 18900 − 16810 = 2090
Sxy = 27560 − (620×410)/10 = 27560 − 25420 = 2140
386 ≈ −34.386. Regression line: y = −34.39 + 1.216x. The correlation coefficient r = Sxy / √Sxx Syy = 2140 / √1760×2090 ≈ 0.927. The slope means that for every additional $1000 in household income, luxury spending increases by about $121.60 on average.
斜率b = Sxy / Sxx = 2140 / 1760 ≈ 1.2159，截距a = 41 − 1.2159×62 ≈ −34.39。回归线：y = −34.39 + 1.216x。相关系数r = 2140 / √1760×2090 ≈ 0.927。斜率表示家庭收入每增加1000美元，奢侈品支出平均增加约121.60美元。
4. Binomial Distribution in Quality Control | 质量控制中的二项分布
A factory produces electronic chips, and historically 8% are defective. To test if a new production process reduces the defect rate, a random sample of 50 chips is taken. The number of defective chips, X, is found to be 1. Using a binomial model, test at the 5% significance level whether there is evidence that the defect rate has decreased. State the null and alternative hypotheses, the p-value, and your conclusion.
某工厂生产电子芯片，历史上次品率为8%。为检验新工艺是否降低了次品率，随机抽取50片芯片，其中次品数X=1。使用二项模型在5%显著性水平下检验是否有证据表明次品率下降。写出原假设和备择假设、p值及结论。
H₀: p = 0.08, H₁: p < 0.08. Under H₀, X ~ 0.0837. Since 0.0837 > 0.05, we do not reject H₀. There is insufficient evidence at the 5% level to conclude that the defect rate has decreased.
H₀: p = 0.08, H₁: p < 0.08。在原假设下X ~ B(50, 0.08)。p值 = P(X ≤ 1) = (0.92)⁵⁰ + 50×0.08×(0.92)⁴⁹ ≈ 0.0156 + 0.0681 = 0.0837。因为0.0837 > 0.05，不拒绝H₀。在5%显著性水平下尚未有足够证据表明次品率下降。
5. Sampling Methods in Environmental Science | 环境科学中的抽样方法
An environmental agency wants to estimate the mean concentration of a pollutant (in µg m⁻³) in a large city. The city is divided into 4 districts with differing industrial activity. The agency decides to use stratified random sampling, with the strata being districts and sample sizes proportional to district area. Explain why this design is preferable to simple random sampling in this context and describe how to select the sample. If one district has a very small area but high pollution, what might be a better allocation strategy?
某环境机构欲估计大城市中某污染物的平均浓度(µg m⁻³)。该城市分为4个工业活动不同的区。机构拟采用分层随机抽样，各层为区，样本量按区面积比例分配。解释此设计为何在此情境下优于简单随机抽样，并阐述样本选取方法。若某区面积很小但污染严重，何种分配策略更佳？
vide the city into the 4 district strata, determine the total sample size n, allocate to each stratum proportionally to its area (nᵢ = n × Aᵢ / Atotal), then take a simple random sample within each stratum. If a small-area district has high pollution, proportional allocation may underrepresent it, leading to a biased estimate. A better strategy would be Neyman allocation, where sample sizes are proportional to stratum size times estimated standard deviation, or simply oversample the high-pollution district.
分层抽样保证每个区按其大小被代表，减少因工业影响不均带来的偏差。方法：将城市分为4个区层，确定总样本量n，按面积比例分配每层样本量(nᵢ = n × Aᵢ / Atotal)，然后在每层内进行简单随机抽样。若小面积区污染严重，按比例分配会使其代表不足，导致估计偏差。更好的策略是Neyman分配，即样本量与层大小乘标准差的估计值成比例，或直接对该区超额抽样。
6. Chi-Squared Test in Medical Trials | 医学试验中的卡方检验
In a clinical trial, 200 patients were randomly assigned to receive either a new drug or a placebo. The outcomes were recorded as 'recovered' or 'not recovered'. The observed frequencies are:
| Recovered | Not recovered | Total | |
|---|---|---|---|
| Drug | 72 | 28 | 100 |
| Placebo | 58 | 42 | 100 |
| Total | 130 | 70 | 200 |
在一项临床试验中，200名患者被随机分配服用新药或安慰剂。结果记录为“康复”或“未康复”。观察频数如上表。在5%显著性水平下进行独立性卡方检验。给出假设、检验统计量、自由度、临界值及结论。
35, Placebo-Recovered = 65, Placebo-Not = 35. χ² = Σ(O−E)²/E = (72−65)²/65 + (28−35)²/35 + (58−65)²/65 + (42−35)²/35 = 49/65 + 49/35 + 49/65 + 49/35 ≈ 0.7538 + 1.4 + 0.7538 + 1.4 = 4.3076. Degrees of freedom = (2−1)(2−1) = 1. Critical value at 5% 08 > 3.841, reject H₀. There is evidence of an association between treatment and recovery.
H₀：康复与处理方式独立；H₁：不独立。期望频数：药物康复 = 65，药物未康复 = 35，安慰剂康复 = 65，安慰剂未康复 = 35。χ² = 4.3076，自由度1，5%临界值3.841。因4.308 > 3.841，拒绝H₀，有证据表明处理方式与康复有关联。
7. Exponential Distribution in Engineering Reliability | 工程可靠性中的指数分布
rs. Find the probability that a component lasts more than 600 hours. If 5 components are used independently, find the probability that at least 2 of them fail before 400 hours. State the memoryless property of the exponential distribution and explain its implication for maintenance scheduling.
某机器部件的寿命T(小时)服从指数分布，均值μ = 500小时。求一个部件寿命超过600小时的概率。若独立使用5个部件，计算至少有2个在400小时前失效的概率。简述指数分布的无记忆性及其对维护排程的意义。
For exponential, rate λ = 1/500 = 0.002. P(T > 600) = e⁻⁰'⁰⁰²×⁶⁰⁰ = e⁻¹'² ≈ 0.3012. P(T < 400) = 1 − e⁻⁰'⁸ = 1 − 0.4493 = 0.5507. Let Y ~ 4493)⁵ − 5×0.5507×(0.4493)⁴ ≈ 1 − 0.0184 − 0.1127 = 0.8689. Memoryless property: P(T > s+t | good as a new one; replacement schedules based on age are not beneficial under this model.
速率λ = 1/500 = 0.002。P(T > 600) = e⁻¹'² ≈ 0.3012。P(T < 400) = 0.5507。Y ~ B(5, 0.5507)，P(Y ≥ 2) ≈ 0.8689。无记忆性：P(T > s+t | T > s) = P(T > t)。在维护中意味着已使用部件与新部件没有区别；基于年龄的更换计划在此模型下无益。
8. Confidence Intervals in Psychology | 心理学中的置信区间
A psychologist measures reaction times (ms) to a stimulus in a random sample of 25 participants after sleep deprivation. The sample mean is 285 ms with a sample standard deviation of 42 ms. Construct a 95% confidence interval for the population mean reaction time. The researcher claims that the mean reaction time under normal sleep is 250 ms. Does the interval support that sleep deprivation increases reaction time? Justify your answer.
一名心理学家测量了25名参与者在睡眠剥夺后对刺激的反应时间(毫秒)。样本均值为285 ms，样本标准差为42 ms。构建总体平均反应时间的95%置信区间。研究者声称正常睡眠下的平均反应时间为250 ms。该区间是否支持睡眠剥夺增加反应时间？说明理由。
g t-distribution with 24 df, t(0.025) = 2.064. The confidence interval: 285 ± 2.064 × (42/√25 267.66, 302.34). The value 250 is far below the lower limit, indicating the population mean is significantly higher than 250 ms. The interval supports the claim that sleep deprivation increases mean reaction time.
使用自由度为24的t分布，t(0.025) = 2.064。置信区间：285 ± 2.064 × (42/√25) = 285 ± 17.34，即(267.66, 302.34)。250远低于下限，表明总体均值显著高于250 ms。该区间支持睡眠剥夺增加反应时间的说法。
9. Geometric Distribution in Sports | 体育统计中的几何分布
miss. State the distribution of X, including its parameter. Find P(X = 5) and P(X > 5). Calculate the expected number of attempts until the first miss. The coach is concerned if a miss occurs in the first 2 attempts; find the probability of this event.
斯蒂芬在篮球中的罚球命中概率为0.85，各次独立。设X为直到第一次不中的尝试次数。写出X的分布及其参数。求P(X = 5)和P(X > 5)。计算直到第一次不中的期望尝试次数。若教练担心在前两次尝试中出现不中，求此事件的概率。
X is the trial number of the first failure, so X ~ Geo(p) where p = P(miss) = 0.15. For geometric with support {1,2,3,...}, P(X = x) = (0.85)x−1 0.15 = 0.0783. P(X > 5) = (0.85)⁵ ≈ 0.4437. E(X) = 1/p = 1/0.15 ≈ 6.67 attempts. The event 'miss in first 2 attempts' is X ≤ 2: P(X=1) + P(X=2) = 0.15 + 0.85×0.15 = 0.2775.
X为首次失败的尝试次数，X ~ Geo(0.15)。P(X=5) = (0.85)⁴ × 0.15 ≈ 0.0783。P(X > 5) = (0.85)⁵ ≈ 0.4437。E(X) = 1/0.15 ≈ 6.67次。前两次有不中的概率为P(X ≤ 2) = 0.15 + 0.85×0.15 = 0.2775。
10. Integrated Data Analysis Case Study | 综合案例演练：跨学科数据分析
A research study investigates the effect of a new teaching method on student performance across three subjects: Mathematics, Physics, and Economics. Data include pre-test and post-test scores, time spent on e-learning platforms, and categorical variables such as gender and prior qualification level. Suggest appropriate statistical techniques to: (i) compare mean improvement scores among the three subjects; (ii) test whether the proportion of students reaching a benchmark differs between genders; (iii) model the relationship between time spent and post-test score after accounting for pre-test score; (iv) check for association between qualification level and choice of subject. Briefly justify each choice.
某项研究探讨新教学法对学生数学、物理、经济三科表现的影响。数据包括前测与后测成绩、在线学习平台使用时间，以及性别、先前资格等级等分类变量。请就以下目标提出合适的统计技术：(i) 比较三科之间的平均进步分数；(ii) 检验达到基准分的学生比例是否存在性别差异；(iii) 在控制前测成绩后，模拟学习时间与后测成绩之间的关系；(iv) 检验资格等级与所选科目之间的关联。简要说明理由。
(i) One-way ANOVA (or repeated measures if same students across subjects) to compare mean improvement, checking normality and homogeneity assumptions. (ii) A two-sample z-test for proportions or a chi-squared test of homogeneity to compare proportions of students reaching the benchmark by gender. (iii) Multiple linear regression with post-test score as response, and predictors: time spent and pre-test score; an ANCOVA model can determine the effect of time after adjusting for pre-test. (iv) A chi-squared test of independence between two categorical variables 'qualification level' and 'subject', provided expected frequencies are large enough. If not, Fisher's exact test may be used.
(i) 单因素方差分析(若为同一批学生则用重复测量ANOVA)比较平均进步分数，需检验正态性和方差齐性。(ii) 两样本比例z检验或同质性卡方检验比较不同性别达到基准的比例。(iii) 多元线性回归，以时间与前测成绩为预测变量，后测成绩为响应变量；或采用ANCOVA调整前测后分析时间效应。(iv) 用独立性卡方检验检验“资格等级”与“所选科目”两分类变量的关联，若期望频数不足可使用Fisher精确检验。
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