📚 Referendums and How They Are Used: A Statistical Perspective | 公投及其运用：统计学的视角
A referendum is a direct vote in which an entire electorate is invited to vote on a particular proposal. While this may seem purely a political process, mathematics—particularly statistics—plays a crucial role in understanding, designing, and interpreting referendums. From opinion polls that measure public sentiment to hypothesis tests that determine whether a result is truly decisive, statistical tools help us make sense of these democratic exercises. In this article, we will explore how mathematical concepts from the Edexcel A-Level syllabus, such as binomial distributions, confidence intervals, and hypothesis testing, can be applied to analyse referendums and their usage.
公投是一种直接投票形式，全体选民受邀就某一特定提案作出表决。虽然这看似纯属政治进程，但数学——尤其是统计学——在理解、设计和解读公投中起着至关重要的作用。从测量民意的民调，到判断结果是否真正决定性的假设检验，统计工具帮助我们理解这些民主实践。本文将探讨来自Edexcel A-Level教学大纲的数学概念，如二项分布、置信区间和假设检验，如何被用于分析公投及其运用。
1. What is a Referendum? | 什么是公投？
A referendum is a mechanism of direct democracy where citizens vote yes or no on a specific question. Unlike general elections where representatives are chosen, a referendum asks the public to make a policy decision directly. The outcome is typically determined by a simple majority, though some require a supermajority. The use of referendums varies across countries, with some employing them for constitutional changes, others for major national issues like EU membership or independence.
公投是直接民主的一种机制，公民对特定问题投票表示赞成或反对。与选举代议士的普选不同，公投直接请公众作出政策决定。结果通常由简单多数决定，尽管有些要求绝对多数。公投的运用因国而异，有的用于宪法修改，有的用于像欧盟成员资格或独立这样的重大国家议题。
2. Types and Uses of Referendums | 公投的类型与用途
Referendums can be classified into several types based on their legal force and purpose: mandatory (required by constitution), advisory (non-binding), and popular initiatives (triggered by citizen petitions). They are used to legitimise important decisions, resolve contentious issues, or approve constitutional amendments. The design of a referendum—wording of the question, timing, and threshold for victory—can significantly influence the outcome, and statistical analysis can help assess the fairness and reliability of these processes.
公投可根据其法律效力和目的分为几类：强制性（宪法要求）、咨询性（非约束性）和公民创制（由公民请愿触发）。公投用于使重要决策合法化、解决争议性问题或批准宪法修正案。公投的设计——问题的措辞、时机和获胜门槛——能显著影响结果，统计分析有助于评估这些过程的公正性和可靠性。
- Mandatory referendum: Automatically triggered for specific issues (e.g. constitutional reform).
强制性公投：针对特定议题自动触发（如宪法改革）。 - Advisory referendum: Government seeks public opinion but is not legally bound.
咨询性公投：政府征询民意但不受法律约束。 - Popular initiative: Citizens collect signatures to force a vote.
公民创制：公民收集签名强制进行投票。
3. Opinion Polls and Sampling Error | 民意调查与抽样误差
Before a referendum, polling organisations survey a sample of the electorate to estimate the proportion who will vote ‘Yes’. Since we cannot ask everyone, sample statistics are used to infer population parameters. The sample proportion p̂ = x/n is an unbiased estimator of the true proportion p. However, each sample varies; the sampling distribution of p̂ is approximately normal for large n, with standard error SE = √(p̂(1-p̂)/n). This allows us to construct confidence intervals for p.
在公投之前，民调机构会抽选部分选民进行调查，以估计将投“赞成”票的比例。由于无法询问所有人，我们使用样本统计量推断总体参数。样本比例 p̂ = x/n 是真实比例 p 的无偏估计量。然而，每个样本都有差异；对于大样本量 n，p̂ 的抽样分布近似正态，标准误为 SE = √(p̂(1-p̂)/n)。这使我们能够构建 p 的置信区间。
95% Confidence Interval: p̂ ± 1.96 × √(p̂(1-p̂)/n)
95% 置信区间：p̂ ± 1.96 × √(p̂(1-p̂)/n)
For example, if a poll of 1000 voters shows 520 supporting a measure (p̂ = 0.52), the 95% confidence interval is 0.52 ± 1.96 × √(0.52×0.48/1000) ≈ (0.489, 0.551). This means we are 95% confident that the true support lies between 48.9% and 55.1%. The margin of error here is about 3.1%. Such intervals help gauge the uncertainty in pre-referendum polls.
例如，如果一项对1000名选民的民调显示520人支持某议案（p̂ = 0.52），则95%置信区间为 0.52 ± 1.96 × √(0.52×0.48/1000) ≈ (0.489, 0.551)。这意味着我们有95%的信心认为真实支持率介于48.9%至55.1%之间。此处的误差幅度约为3.1%。这样的区间有助于评估公投前民调的不确定性。
4. Determining Sample Size | 样本量的确定
Pollsters must decide how many individuals to sample to achieve a desired margin of error E. The required sample size for estimating a proportion with confidence level (1-α) is n = (z² × p̂(1-p̂)) / E². Since p̂ is unknown before the survey, the maximum variability p̂ = 0.5 is often used to give the most conservative (largest) sample size. For a 95% confidence level and E = 3%, n = (1.96² × 0.5×0.5) / 0.03² ≈ 1067. This explains why many national polls sample around 1000 people.
民调人员必须决定抽取多少人样本才能达到期望的误差幅度 E。估计总体比例所需样本量的计算公式为 n = (z² × p̂(1-p̂)) / E²，置信水平为(1-α)。由于在调查前 p̂ 未知，常采用 p̂ = 0.5 以获得最大变异性下的最保守（最大）样本量。对于95%置信水平和 E = 3%，n = (1.96² × 0.5×0.5) / 0.03² ≈ 1067。这解释了为什么许多全国性民调的样本量约为1000人。
5. Binomial Distribution in Referendum Predictions | 二项分布与公投预测
In a referendum with a large electorate, each voter’s choice can be modelled as a Bernoulli trial with probability p of voting ‘Yes’. The total number of Yes votes X out of n voters follows a binomial distribution, X ~ B(n, p). When n is large, the binomial can be approximated by a normal distribution: X ≈ N(np, np(1-p)). This approximation lets us calculate probabilities about the outcome, such as the likelihood of a Yes victory given a certain true support level.
在选民众多的公投中，每位选民的选择可被建模为一次伯努利试验，投“赞成”的概率为 p。总赞成票数 X（共 n 位选民）服从二项分布 X ~ B(n, p)。当 n 很大时，可以用正态分布近似二项分布：X ≈ N(np, np(1-p))。这种近似使我们能够计算关于结果的概率，例如在某一真实支持水平下，“赞成”获胜的可能性。
If the true support is p = 0.52 and 1,000,000 people vote, the expected Yes votes are 520,000 with standard deviation √(1,000,000×0.52×0.48) ≈ 499.6. The probability of Yes receiving more than half the votes is P(X > 500,000) ≈ P(Z > (500,000 – 520,000)/499.6) = P(Z > -40.03), virtually 1. Thus, even a 2% lead in true support gives near certainty with large electorates.
如果真实支持率为 p = 0.52，并且有100万人投票，则期望的赞成票数为520,000，标准差为 √(1,000,000×0.52×0.48) ≈ 499.6。赞成票数过半的概率为 P(X > 500,000) ≈ P(Z > (500,000 – 520,000)/499.6) = P(Z > -40.03)，几乎为1。因此，在大量选民的情况下，即便真实支持率仅领先2%，胜出的确定性也接近百分之百。
6. Hypothesis Testing: Is the Result Conclusive? | 假设检验：结果是否确凿？
After a referendum, we can use hypothesis testing to decide whether the winning side’s lead is statistically significant. We can set up a null hypothesis H₀: p = 0.5 (the population is evenly split) against an alternative H₁: p ≠ 0.5. Given the observed sample proportion p̂ from the full population (which can be treated as a sample from a super-population or simply test if the result deviates from 0.5 due to chance), we compute the test statistic z = (p̂ – 0.5) / √(0.5×0.5/n). If p-value < α, we reject H₀, concluding that the result is not a random fluctuation.
公投结束后，我们可以使用假设检验来判断获胜方的领先是否具有统计显著性。我们可以设立零假设 H₀: p = 0.5（总体意见均匀分布）和备择假设 H₁: p ≠ 0.5。根据从全部选民中观测到的样本比例 p̂（可视为来自超级总体的样本，或直接检验结果是否因偶然偏离0.5），我们计算检验统计量 z = (p̂ – 0.5) / √(0.5×0.5/n)。如果 p值 < α，我们拒绝H₀，得出结论：该结果并非随机波动。
For example, in a referendum with 30 million votes, 15.6 million Yes (p̂ = 0.52). Then z = (0.52 – 0.5)/√(0.25/30,000,000) ≈ 69.3; p-value is astronomically small. Even for a very close 50.1% result with the same n, z ≈ 3.46, p-value ≈ 0.0005, still highly significant. Thus, with large populations, tiny leads can be statistically significant, but political significance is another matter.
例如，在一场有3000万张选票的公投中，赞成票为1560万（p̂ = 0.52）。则 z = (0.52 – 0.5)/√(0.25/30,000,000) ≈ 69.3；p值小得惊人。即使对于同样人口规模下50.1%的极微弱结果，z ≈ 3.46，p值 ≈ 0.0005，仍然高度显著。因此，对大规模人口而言，极小的领先在统计上也是显著的，但政治显著性则是另一回事。
7. The Margin of Victory and Confidence | 胜幅与置信度
We can also estimate the margin of victory in terms of a confidence interval for the difference between the Yes and No proportions. If d = p_yes – p_no, then d̂ = 2p̂ – 1. The standard error of d̂ is √(4p̂(1-p̂)/n). A 95% CI can be constructed. If this interval does not include 0, we can be confident that the Yes side truly won. For a 52% Yes vote among 30 million, d̂ = 0.04, SE ≈ 0.00018, 95% CI = (0.0396, 0.0404), which firmly excludes zero.
我们还可以从赞成与反对比例之差的置信区间来估算胜幅。设 d = p_yes – p_no，则 d̂ = 2p̂ – 1。d̂ 的标准误为 √(4p̂(1-p̂)/n)。可以构建95%置信区间。如果该区间不包含0，我们就能确信赞成方确实获胜。对3000万人口中52%的赞成票，d̂ = 0.04，SE ≈ 0.00018，95% CI = (0.0396, 0.0404)，完全排除0。
8. Exit Polls and Quick Counts | 出口民调与快速计票
Exit polls survey voters immediately after they have cast their ballots, allowing early estimates of the outcome. Because they are based on samples, the same statistical principles apply. Quick counts, used in some countries, involve sampling polling stations rather than individuals. The proportion of votes for each option in the sample is used to estimate the final proportion, with confidence intervals updated as results arrive. This is essentially a continuous sampling process.
出口民调在选民投完票后立即进行调查，从而能提前估计结果。由于它们基于样本，同样的统计原理适用。某些国家使用的快速计票，是抽取投票站而非个人作为样本。样本中各选项得票比例被用于估计最终比例，并随着结果陆续出炉更新置信区间。这本质上是一个持续抽样过程。
9. Voting Paradoxes and Mathematical Challenges | 投票悖论与数学挑战
Beyond simple statistical analysis, referendums raise deeper mathematical issues about aggregating preferences. The Condorcet paradox shows that majority rule can lead to cyclical social preferences when there are more than two options. Arrow’s impossibility theorem states that no voting system can simultaneously satisfy all reasonable fairness criteria. While most referendums are binary, the design of multi-option referendums (e.g., using ranked-choice voting) can lead to strategic voting and counterintuitive outcomes.
除了基础的统计分析，公投还引发了关于偏好汇总的更深层次数学问题。孔多塞悖论表明，当选项超过两个时，多数规则可能导致循环的社会偏好。阿罗不可能定理指出，没有任何投票制度能同时满足所有合理的公平性标准。虽然多数公投是二选一，但多选项公投的设计（如采用排序投票制）可能导致策略性投票和违反直觉的结果。
10. Case Study: Brexit Referendum Analysis | 案例分析：英国脱欧公投分析
The 2016 UK referendum on EU membership saw 51.9% vote Leave versus 48.1% Remain, with a turnout of 72.2% (about 33.6 million votes). Applying a hypothesis test with H₀: p = 0.5, we get z ≈ (0.519 – 0.5)/√(0.5×0.5/33,577,342) ≈ 69.8, an overwhelmingly significant result statistically. A 95% confidence interval for the Leave proportion is 0.519 ± 1.96×√(0.519×0.481/33,577,342) ≈ (0.5185, 0.5195). While the vote was close in percentage terms, the massive sample size leaves no doubt that Leave had more supporters in the voting population.
2016年英国有关欧盟成员资格的公投中，51.9%支持脱欧，48.1%支持留欧，投票率为72.2%（约3360万张选票）。应用假设检验，零假设 H₀: p = 0.5，得到 z ≈ (0.519 – 0.5)/√(0.5×0.5/33,577,342) ≈ 69.8，统计上极其显著。脱欧比例的95%置信区间为 0.519 ± 1.96×√(0.519×0.481/33,577,342) ≈ (0.5185, 0.5195)。尽管票数百分比接近，庞大的样本量使脱欧方在投票人口中拥有更多支持者这一结论确信无疑。
11. Conclusion | 结语
Referendums are more than just political events; they are rich with mathematical applications. Statistical reasoning helps us design accurate polls, interpret voting outcomes, and evaluate the decisiveness of a result. By using binomial distributions, confidence intervals, and hypothesis tests, we can separate genuine trends from random noise. Understanding these mathematical underpinnings empowers citizens and analysts to engage more critically with direct democracy. The skills from A-Level Mathematics thus extend far beyond the classroom, into the very heart of our civic life.
公投不仅仅是政治事件，它们充满了数学应用。统计推理帮助我们设计精确的民意调查，解读投票结果，并评估结果的决断力。通过运用二项分布、置信区间和假设检验，我们能将真实的趋势与随机噪音区分开来。理解这些数学基础，能使公民和分析者更具批判性地参与直接民主。因此，A-Level数学中的技能远超课堂，直抵我们公民生活的核心。
Published by TutorHao | 数学 Revision Series | aleveler.com
更多咨询请联系16621398022（同微信）
屏轩国际教育cambridge primary/secondary checkpoint, cat4, ukiset,ukcat,igcse,alevel,PAT,STEP,MAT, ibdp,ap,ssat,sat,sat2课程辅导，国外大学本科硕士研究生博士课程论文辅导Cancel reply