The Labour Party | 工党

📚 The Labour Party | 工党

In A‑Level Mathematics (Edexcel), statistical methods are often applied to real‑world contexts such as opinion polls. Using data about the Labour Party, we can reinforce core topics including the binomial distribution, normal approximations, confidence intervals and hypothesis testing. This article works through a typical polling scenario to illustrate these techniques.

在A‑Level数学（Edexcel）中，统计学方法经常被应用于民意调查等实际背景。利用关于工党的数据，我们可以巩固二项分布、正态近似、置信区间和假设检验等核心主题。本文将通过一个典型的民调情景来展示这些技巧。

1. Modelling Poll Responses: The Binomial Distribution | 民调响应建模：二项分布

Suppose a polling organisation randomly selects n voters and records whether each supports the Labour Party. The number of Labour supporters, X, can be modelled by a binomial distribution: X ~ B(n, p), where p is the true proportion of Labour voters in the population.

假设某民意调查机构随机选取 n 名选民，记录每人是否支持工党。支持工党的人数 X 可以用二项分布建模：X ~ B(n, p)，其中 p 是人口中工党支持者的真实比例。

Each selection is a Bernoulli trial with two outcomes: ‘Labour’ (success) or ‘not Labour’ (failure). If the sample is sufficiently small relative to the population, the trials are effectively independent and the probability p stays constant for each observation.

每次选择都是一个伯努利试验，只有两种结果：“支持工党”（成功）或“不支持工党”（失败）。如果样本相对于总体足够小，那么各次试验实际上是独立的，且概率 p 在每次观测中保持不变。

---

2. Conditions for a Binomial Model | 二项模型的条件

For the binomial model to be valid, we require a fixed number of trials n, each trial must have exactly two possible outcomes, the trials must be independent, and the probability of success p must be the same for every trial. In a well‑conducted opinion poll using random sampling without replacement, the independence condition is approximately met when the sample size is less than 10% of the population.

要使二项模型成立，我们需要有固定的试验次数 n，每次试验必须恰好有两种可能结果，各次试验必须相互独立，且每次试验的成功概率 p 必须相同。在采用无放回随机抽样的良好民调中，当样本量小于总体量的 10% 时，独立性条件可近似满足。

When these conditions hold, X is a discrete random variable whose probability mass function is given by P(X = k) = ⁿCₖ pᵏ (1 – p)ⁿ⁻ᵏ for k = 0,1,…,n.

当这些条件成立时，X 是一个离散随机变量，其概率质量函数为 P(X = k) = ⁿCₖ pᵏ (1 – p)ⁿ⁻ᵏ，其中 k = 0,1,…,n。

---

3. Estimating Labour Support: Sample Proportion | 估计工党支持率：样本比例

The observed sample proportion is p̂ = X/n. This statistic is an unbiased estimator of the population proportion p, and its sampling distribution has mean p and variance p(1 – p)/n. For a given sample of size n = 1000, suppose 420 people express support for Labour; then p̂ = 0.42.

观测到的样本比例为 p̂ = X/n。该统计量是总体比例 p 的无偏估计量，其抽样分布的均值为 p，方差为 p(1 – p)/n。设一个样本量为 n = 1000 的样本中，有 420 人表示支持工党；则 p̂ = 0.42。

Because p̂ is based on a random sample, it will vary from poll to poll. The standard error of p̂ is √(p(1 – p)/n), which quantifies this variability. In practice, we estimate the standard error using p̂ when p is unknown.

由于 p̂ 基于随机样本，它会因调查而异。p̂ 的标准误差为 √(p(1 – p)/n)，它量化了这种变异性。在实践中，当 p 未知时，我们使用 p̂ 来估计标准误差。

---

4. Normal Approximation to the Binomial | 二项分布的正态近似

When n is large and p is not too close to 0 or 1, the distribution of X can be approximated by a normal distribution: X ~ N(np, np(1 – p)) approximately. Consequently, the sample proportion satisfies p̂ ~ N(p, p(1 – p)/n) approximately. A common rule of thumb is that both np and n(1 – p) should be at least 10.

当 n 很大且 p 不太接近 0 或 1 时，X 的分布可用正态分布近似：X 近似服从 N(np, np(1 – p))。因此，样本比例满足 p̂ 近似服从 N(p, p(1 – p)/n)。常用的经验法则是 np 和 n(1 – p) 均应至少为 10。

For the Labour poll with n = 1000 and p̂ = 0.42, we check np̂ = 420 and n(1 – p̂) = 580, both well above 10, so the normal approximation is appropriate.

对于工党民调中 n = 1000、p̂ = 0.42 的情况，我们检验得 np̂ = 420，n(1 – p̂) = 580，均远大于 10，因此正态近似是合适的。

---

5. Confidence Intervals for Labour’s True Support | 工党真实支持率的置信区间

An approximate 95% confidence interval for p is given by p̂ ± z* × √(p̂(1 – p̂)/n), where z* = 1.96 is the critical value from the standard normal distribution. For our data, the standard error is √(0.42 × 0.58 / 1000) ≈ 0.0156.

p 的近似 95% 置信区间为 p̂ ± z* × √(p̂(1 – p̂)/n)，其中 z* = 1.96 是来自标准正态分布的临界值。对于我们的数据，标准误差为 √(0.42 × 0.58 / 1000) ≈ 0.0156。

95% CI: 0.42 ± 1.96 × 0.0156 → (0.389, 0.451)

95% 置信区间：0.42 ± 1.96 × 0.0156 → (0.389, 0.451)

We are 95% confident that the true proportion of Labour supporters in the population lies between 38.9% and 45.1%. This interval captures values both below and above 40%, meaning the poll data alone does not give a definitive picture of whether Labour is above the 40% mark.

我们有 95% 的把握认为，人口中工党支持者的真实比例介于 38.9% 到 45.1% 之间。此区间同时包含了低于和高于 40% 的值，这意味着仅凭民调数据尚不能明确工党支持率是否超过 40% 这一关口。

---

6. Hypothesis Testing: Is Labour Above 40%? | 假设检验：工党支持率是否高于40%？

A political analyst might ask: does this poll provide significant evidence that Labour’s true support exceeds 40%? We set up the null hypothesis H₀: p = 0.40 against the alternative H₁: p > 0.40, using a significance level of α = 0.05.

政治分析人士可能会问：本次民调是否提供了显著证据，表明工党的真实支持率超过 40%？我们设定原假设 H₀: p = 0.40，备择假设 H₁: p > 0.40，并采用显著性水平 α = 0.05。

Under H₀, the test statistic is z = (p̂ – p₀) / √(p₀(1 – p₀)/n). Plugging in the numbers: z = (0.42 – 0.40) / √(0.40 × 0.60 / 1000) = 0.02 / √(0.00024) ≈ 0.02 / 0.01549 ≈ 1.29.

在 H₀ 下，检验统计量为 z = (p̂ – p₀) / √(p₀(1 – p₀)/n)。代入数字：z = (0.42 – 0.40) / √(0.40 × 0.60 / 1000) = 0.02 / √(0.00024) ≈ 0.02 / 0.01549 ≈ 1.29。

The critical value for a one‑tailed test at α = 0.05 is z₀.₀₅ = 1.645. Since 1.29 < 1.645, we do not reject H₀. There is insufficient evidence to claim that Labour's true support is greater than 40%.

在 α = 0.05 下的单尾检验临界值为 z₀.₀₅ = 1.645。因 1.29 < 1.645，我们不拒绝 H₀。没有足够证据表明工党的真实支持率大于 40%。

---

7. One‑Tailed vs Two‑Tailed Tests | 单尾与双尾检验

The choice between a one‑tailed and a two‑tailed test depends on the research question. If we were simply testing whether Labour’s support differs from 40%, the hypotheses would be H₀: p = 0.40 and H₁: p ≠ 0.40, requiring a two‑tailed test with critical values ±1.96 at α = 0.05.

选择单尾检验还是双尾检验取决于研究问题。如果我们只是检验工党的支持率是否不同于 40%，则假设应为 H₀: p = 0.40 和 H₁: p ≠ 0.40，这需要在 α = 0.05 下使用临界值 ±1.96 进行双尾检验。

A one‑tailed test has greater power to detect an effect in the specified direction, but it must be justified before seeing the data. In our scenario, the claim that Labour is above 40% justifies a one‑sided test; the observed z = 1.29 is still not significant even in this setting.

单尾检验在特定方向上检测效应的功效更强，但必须在观察数据之前进行合理说明。在我们的场景中，“工党支持率超过 40%”的主张使单侧检验合理；但即使如此，观测到的 z = 1.29 也不显著。

---

8. p‑Values and Significance | p值与显著性

The p‑value is the probability of obtaining a test statistic at least as extreme as the one observed, assuming H₀ is true. For the one‑tailed test, p‑value = P(Z > 1.29) = 1 – Φ(1.29) ≈ 0.0985.

p 值是在 H₀ 为真的前提下，获得一个至少与观测值同样极端的检验统计量的概率。对于单尾检验，p‑value = P(Z > 1.29) = 1 – Φ(1.29) ≈ 0.0985。

Since 0.0985 > 0.05, the result is not statistically significant at the 5% level. This means that if the true support were indeed 40%, we would see a sample proportion of 0.42 or higher about 9.85% of the time purely by random sampling variability.

由于 0.0985 > 0.05，该结果在 5% 水平上不具有统计显著性。这意味着如果真实支持率确实为 40%，那么单纯由于随机抽样波动，我们有大约 9.85% 的机会观测到 0.42 或更高的样本比例。

---

9. Type I and Type II Errors in Polling | 民调中的第一类和第二类错误

A Type I error occurs if we reject H₀ when it is actually true—falsely concluding that Labour’s support exceeds 40%. The probability of a Type I error is α, which we set at 0.05. A Type II error happens if we fail to reject H₀ when H₁ is true—missing a real increase in support.

如果我们拒绝了实际上为真的 H₀，即错误地得出工党支持率超过 40% 的结论，就会发生第一类错误。第一类错误的概率为 α，我们设定为 0.05。第二类错误发生在 H₁ 为真时我们未能拒绝 H₀，即错过了真正的支持率上升。

The power of the test (1 – β) is the probability of correctly rejecting H₀ for a given true p. Power increases with larger sample sizes, a larger true effect size, and a higher α. With n = 1000 and a true p of, say, 0.45, the power to detect a significant increase above 0.40 would be considerably higher than for a true p of 0.42.

检验的功效（1 – β）是对于某一给定的真实 p 值，正确拒绝 H₀ 的概率。功效随着样本量的增加、真实效应量的增大以及 α 的升高而提高。当 n = 1000 且真实 p 为 0.45 时，检测出超过 0.40 的显著增加的把握要远高于真实 p 为 0.42 时的情况。

---

10. Determining Sample Size for Desired Accuracy | 确定所需精度的样本量

To design a poll with a specified margin of error E at a given confidence level, we solve E = z* × √(p̂(1 – p̂)/n) for n. Using a conservative estimate of p̂ = 0.5 (which maximises variance) gives n = (z*/E)² × 0.25. For a 3% margin of error at 95% confidence, n = (1.96/0.03)² × 0.25 ≈ 1067.

要设计一个在给定置信水平下具有指定误差幅度 E 的民调，我们求解 E = z* × √(p̂(1 – p̂)/n) 得到 n。使用保守估计 p̂ = 0.5（使方差最大）可得出 n = (z*/E)² × 0.25。对于 95% 置信度下 3% 的误差幅度，n = (1.96/0.03)² × 0.25 ≈ 1067。

If a polling agency wants to detect a 2‑percentage‑point difference from 40% with high power, the required sample size can be computed using power analysis. Larger samples reduce both Type I and Type II error risks, but come with greater cost and logistical effort.

如果民调机构希望以高功效检测出与 40% 的 2 个百分点的差异，则所需样本量可通过功效分析计算。更大的样本可以同时降低第一类和第二类错误的风险，但也伴随着更高的成本和更复杂的操作。

---

11. Interpreting Results in Context | 在背景中解释结果

Statistical significance does not always imply practical importance. Even if a poll showed a statistically significant 1% increase above 40%, it might not change political strategy. Conversely, a non‑significant result with a wide confidence interval could hide a meaningful shift, especially if the sample size is small.

统计显著性并不总意味着实际重要性。即使某项民调显示出统计上显著的、高于 40% 的 1% 增长，也未必会改变政治策略。反之，一个不显著但置信区间很宽的结果可能会掩盖有意义的变化，尤其在样本量较小时。

In our example, the poll suggests Labour support could be as low as 39% or as high as 45%. Decision‑makers should combine the confidence interval with subject‑matter knowledge and the consequences of potential errors before reaching firm conclusions.

在我们的例子中，民调表明工党支持率可能低至 39% 或高达 45%。决策者在下定论之前，应结合置信区间与专业知识以及潜在错误的后果。

---

12. Conclusion and Exam Tips | 结论与考试技巧

The Labour Party polling example brings together many key ideas in Edexcel A‑Level statistics: building a binomial model, checking conditions, using the normal approximation, calculating confidence intervals, and performing hypothesis tests. Examiners look for clear statements of hypotheses, correct use of standard errors, proper interpretation of p‑values, and awareness of the assumptions behind each procedure.

工党民调的实例汇聚了 Edexcel A‑Level 统计学中的许多关键思想：建立二项模型、检查条件、使用正态近似、计算置信区间以及进行假设检验。考官期望看到清晰的假设陈述、标准误差的正确使用、对 p 值的恰当解释，以及对每个过程背后假设的意识。

Always state the distribution of the test statistic under H₀, verify that approximation conditions are satisfied, and write a conclusion in context, linking the p‑value or test statistic to the original claim. Practising with realistic polling numbers helps build confidence for the exam.

务必写出在 H₀ 下检验统计量的分布，验证近似条件是否满足，并结合背景写出结论，将 p 值或检验统计量与原始主张联系起来。用真实的民调数字进行练习，有助于为考试积累信心。

Published by TutorHao | Mathematics Revision Series | aleveler.com

更多咨询请联系16621398022（同微信）