📚 Democratic States: A Mathematical Analysis of Voting and Representation | 民主国家：投票与代表的数学分析
A flourishing democratic state relies on fair elections, representative parliaments and informed decision‑making. Behind these political ideals lies a rich layer of mathematics: from counting ballots and distributing seats to testing opinion polls and measuring voting power. This article explores how A‑level Mathematics – especially statistics, probability, combinatorics and decision mathematics – helps us understand and design the machinery of democracy.
繁荣的民主国家依赖公平的选举、代议制议会和明智的决策。这些政治理想的背后隐藏着丰富的数学：从选票计数、席位分配到民意调查的检验和投票权力度量，无不体现数学的威力。本文探讨 A‑level 数学——特别是统计、概率、组合与决策数学——如何帮助我们理解并设计民主的运行机制。
1. Voting Systems and Paradoxes | 投票制度与悖论
In any democratic state, the method used to translate individual preferences into a collective decision can dramatically alter the outcome. The simplest system is plurality voting, where each voter chooses one candidate and the candidate with the most votes wins. However, this can lead to a minority winner when more than two candidates are standing.
在任何民主国家，将个人偏好转化为集体决策的方法会显著影响结果。最简单的制度是简单多数制，即每位选民选一名候选人，得票最多者获胜。然而当候选人多于两人时，这可能导致少数胜出。
Consider three candidates A, B and C with the following preferences among 100 voters: 40 voters rank A > B > C, 35 rank B > C > A, and 25 rank C > B > A. Under plurality, A wins with 40 votes, yet 60% of voters prefer B or C over A. This phenomenon is known as the voting paradox, first described by Condorcet.
假设有三位候选人 A、B、C，100 位选民的偏好如下：40 人排序 A > B > C，35 人 B > C > A，25 人 C > B > A。在简单多数制下，A 以 40 票胜出，但 60% 的选民更偏好 B 或 C 而非 A。这一现象称为投票悖论，由孔多塞首先提出。
The mathematical challenge is to design a voting rule that reflects the true will of the people – a problem that touches on axiomatic set theory and social choice functions.
数学上的挑战在于设计一种能够反映人民真实意愿的投票规则——这是一个涉及公理化集合论和社会选择函数的问题。
2. Condorcet Method and Rankings | 孔多塞方法与排序
proposes that the winner should be the candidate who beats every other candidate in a head‑to‑ comparison. To verify this, we count pairwise comparisons. Using the data above, A beats B? 40 voters prefer A to B, 60 prefer B to A, so B beats A. B beats C? 40+35 = 75 prefer B to C, 25 prefer C to B, so B beats C. C beats A? his creates a cycle: A loses to B, B beats C, but C beats A. There is no Condorcet winner.
孔多塞方法提出，获胜者应当是在两两对决中击败所有其他候选人的那位。为验证这一点，我们进行两两比较计数。用上面的数据，A 对 B：40 人更偏好 A，60 人更偏好 B，故 B 胜 A。B 对 C：40+35=75 人偏好 B，25 人偏好 C，故 B 胜 C。C 对 A：35+25=60 人偏好 C，40 人偏好 A，故 C 胜 A。这形成一个循环：A 输给 B，B 胜 C，但 C 胜 A。不存在孔多塞赢家。
probability of such a cycle can be computed for random preference profiles. If each voter's ranking is uniformly random among 3! nalysis shows the chance of a Condorcet cycle increases with the number of voters but remains remarkably high for three candidates.
这种循环的概率可针对随机偏好组合进行计算。若每位选民对 3! = 6 种可能的排序均匀随机选择，组合分析表明，孔多塞循环的概率随选民人数增加而上升，且对于三位候选人而言始终相当高。
3. Borda Count: Scoring by Rank | 博达计数：按排序评分
An alternative to plurality and Condorcet methods is the Borda count, where voters rank candidates and points are assigned according to position. In an election with n candidates, a first‑place ranking gives n−1 points, second place gives n−2, …, and last place 0. The candidate with the highest total score wins.
与简单多数制和孔多塞方法不同，博达计数让选民对候选人排序，并按位置赋予分值。在有 n 位候选人的选举中，第一名得 n−1 分，第二名得 n−2 分，依此类推，末位得 0 分。总分最高的候选人获胜。
Borda count. This outcome better reflects the overall preference intensity, but the Borda count is susceptible to strategic voting: a voter may rank a strong opponent artificially low.
仍用前例，n=3：A 得分 = 40×2 + 35×0 + 25×0 = 80；B = 40×1 + 35×2 + 25×1 = 135；C = 40×0 + 35×1 + 25×2 = 85。博达计数下 B 胜出。这一结果更好地反映了总体偏好强度，但博达计数容易受到策略性投票的影响：选民可能故意将强劲对手排在低位。
4. Probability and Election Polling | 概率与选举民意调查
Democratic states rely heavily on opinion polls to gauge public sentiment. From a statistical standpoint, a poll is a binomial experiment: a random sample of n voters is drawn from a large population, and each response is either ‘support’ (success) or ‘not support’ (failure). If the true population proportion is p, the number of supporters X follows a binomial distribution X ~ B(n, p).
民主国家在很大程度上依赖民意调查来感知公众情绪。从统计学的角度看，民调就是一个二项试验：从大量选民中随机抽取 n 人，每一回答要么“支持”（成功），要么“不支持”（失败）。若真实总体比例为 p，则支持者人数 X 服从二项分布 X ~ B(n, p)。
y of observing exactly k supporters is P(X = k) = C(n, k) pᵏ (1 − p)ⁿ⁻ᵏ, where C(n, k) = n! / [k!(n−k)!]. For large n, the normal approximation to the binomial is used: X ≈ N(np, np(1−p)). A 95% confidence interval for p is then p̂ ± 1.96 × √[p̂(1−p̂)/n], where p̂ = X/n.
恰好观测到 k 名支持者的概率为 P(X = k) = C(n, k) pᵏ (1 − p)ⁿ⁻ᵏ，其中 C(n, k) = n! / [k!(n−k)!]。当 n 很大时，可使用二项分布的正态近似：X ≈ N(np, np(1−p))。则 p 的 95% 置信区间为 p̂ ± 1.96 × √[p̂(1−p̂)/n]，其中 p̂ = X/n。
This interval tells us that if we were to repeat the poll many times, 95% of such intervals would capture the true p. In democratic contexts, this quantification of uncertainty is crucial for interpreting election forecasts.
这个区间告诉我们，若反复进行同样的民调，95% 的此类区间会包含真实的 p。在民主语境中，这种不确定性的量化对解读选举预测至关重要。
5. Hypothesis Testing: Is the Lead Significant? | 假设检验：领先是否显著？
Election polls often report a candidate leading by a few percentage points. A‑ is statistically significant. Suppose a poll of n = 1000 voters shows candidate A with 52% support and candidate B with 48%. We can test the null hypothesis H₀: p = 0.5 against the two‑tailed alternative H₁: p ≠ 0.5, where p is the true proportion supporting A.
选举民调常报道某候选人领先几个百分点。A‑level 统计学提供了检验这种领先是否显著的工具。假设一项对 n = 1000 名选民的调查显示候选人 A 支持率 52%，B 为 48%。我们可以检验原假设 H₀: p = 0.5，对立假设为双尾 H₁: p ≠ 0.5，其中 p 为支持 A 的真实比例。
Under H₀, the test statistic is z = (p̂ − 0.5) / √[0.5×0.5/n] = 0.02 / √0.25/1000 = 0.02 / 0.01581 ≈ 1.265. At the 5% value is 1.96, so we fail to reject H₀. The observed lead is not statistically significant – it could easily arise from sampling error.
在 H₀ 下，检验统计量为 z = (p̂ − 0.5) / √[0.5×0.5/n] = 0.02 / √0.25/1000 = 0.02 / 0.01581 ≈ 1.265。在 5% 显著性水平下，临界值为 1.96，因此我们不能拒绝 H₀。观测到的领先并无统计显著性——它很可能只是由抽样误差引起。
This use of hypothesis testing helps prevent overinterpretation of small fluctuations in poll results, thereby protecting the democratic discourse from false claims of momentum.
如此运用假设检验有助于避免对民调结果的小幅波动过度解读，从而保护民主话语，防止“势头”的虚假宣称。
6. Apportionment: Allocating Seats Fairly | 席位分配：公平分配议席
In many democracies, legislative seats are distributed to regions (states or constituencies) in proportion to their populations. This apportionment problem is mathematically nontrivial because seats are indivisible and fractions must be rounded. Several historical methods exist, each with its own biases.
在许多民主国家，立法席位按人口比例分配给各地区（州或选区）。这个席位分配问题在数学上并非易事，因为席位不可分割，必须对小数进行取整。历史上有多种方法，各有偏倚。
/10 = 1000. Exact quotas are q₁ = 5.3, q₂ = 2.4, q₃ = 2.3. The simplest method is Hamilton's: give each state the integer part (5,2,2) – total 9 – at to the state with the largest fractional remainder (state 1 with 0.3). Result: 6, 2, 2.
设想一个由三个州组成的小型联邦，议会共 10 席。各州人口为：P₁ = 5300，P₂ = 2400，P₃ = 2300。标准除数为总人口除以总席位 = 10000/10 = 1000。精确配额为 q₁ = 5.3，q₂ = 2.4，q₃ = 2.3。最简单的是汉密尔顿法：给各州整数部分（5,2,2）——共 9 席，剩余一席分配给小数余数最大的州（州 1，0.3）。结果为 6, 2, 2。
Other methods, such as Jefferson's (using a modified divisor to round down all quotas to integers that sum exactly to the total seats) or Webster's (rounding to the nearest integer using a modified divisor), can produce slightly different allocations and may violate quota rules. The table below illustrates a possible divergence:
其他方法，如杰斐逊法（用一个修正除数使所有配额向下取整后总数恰好等于总席位）或韦伯斯特法（用修正除数四舍五入到整数），可能产生略有不同的分配，并可能违反配额规则。下表显示了可能的分歧：
| Method / 方法 | State 1 | State 2 | State 3 |
|---|---|---|---|
| Exact quota / 精确配额 | 5.3 | 2.4 | 2.3 |
| Hamilton / 汉密尔顿 | 6 | 2 | 2 |
| Jefferson (divisor ≈ 960) / 杰斐逊 | 5 | 3 | 2 |
| Webster (divisor ≈ 1005) / 韦伯斯特 | 5 | 2 | 3 |
The choice of apportionment method can alter the balance of power in a parliament, illustrating the deep connection between mathematics and representative democracy.
分配方法的选择可能改变议会中的权力平衡，显示出数学与代议制民主之间的深刻联系。
7. Voting Power Indices: Banzhaf and Shapley‑Shubik | 投票权指数：班扎夫与沙普利‑舒比克
In a democratic state with weighted voting (e.g. a council where members represent different numbers of constituents), it is not enough to know the number of votes each member casts; we need to measure their voting power – the ability to influence outcomes. The Banzhaf power index does this by counting the number of winning coalitions in which a voter is critical.
在采用加权投票的民主国家（如理事会内成员代表不同数量的选民），仅知道每位成员所投票数是不够的；我们需要度量其投票权——亦即影响结果的能力。班扎夫权力指数通过统计某个投票者在获胜联盟中成为关键者的次数来实现这一度量。
A motion needs a majority (≥51) to pass. Winning coalitions and critical members: {A,B} – ecause any two suffice. The Banzhaf power count for A: appears critical in {A,B} and {A,C} – 2 times. B: in {A,B} and {B,C} – 2. C: in {A,C} and {B,C} – 2. Total critical defections = 6, so each has Banzhaf index = 2/6 = 1/3.
设想一个简单的投票体，三位成员 A（30 席）、B（40）、C（30），共 100。议案需要多数（≥51）通过。获胜联盟及关键成员：{A,B} — A 关键（30+40=70≥51，若无 A 则 B 单独只有 40，不获胜），B 关键；{A,C}：两者均关键；{B,C}：两者均关键；{A,B,C}：无人关键，因为任意两位即可。A 的班扎夫权力计数：在 {A,B} 和 {A,C} 中关键 — 2 次。B：在 {A,B} 和 {B,C} 中 — 2 次。C：在 {A,C} 和 {B,C} 中 — 2 次。总关键背叛数为 6，故每人班扎夫指数 = 2/6 = 1/3。
Surprisingly, although B has 40 seats versus A’s 30, their power is equal under this index. This highlights why mathematically naive weighting can distort democratic representation.
令人惊讶的是，虽然 B 有 40 席而 A 只有 30 席，但因该指数两者权力相等。这凸显了为何数学上简单的加权可能会扭曲民主代表。
8. Coalition Formation and Game Theory | 联盟形成与博弈论
Democratic states are often run by coalitions of multiple parties. Game theory models such situations as cooperative games where players (parties) form binding agreements to achieve a majority. The Shapley value, a concept from cooperative game theory, distributes the total ‘payoff’ (cabinet posts, policy influence) according to each party’s marginal contribution averaged over all possible joining orders.
民主国家常由多党联盟治理。博弈论将此类情形建模为合作博弈，其中局中人（政党）结成有约束力的协议以取得多数。合作博弈中的沙普利值按照各党在所有可能加入顺序中的边际贡献平均值来分配总“收益”（内阁职位、政策影响力）。
For a 3‑ probability that it is pivotal when parties join a coalition in random order. Orders: ABC – A not needed initially? Let' al in 2 of 6 orders → Shapley value 2/6 = 1/3; B also 1/3; C 1/3. This equal division contrasts with the seat shares of 0.45, 0.35, 0.20, revealing how small parties can hold disproportionate bargaining power in coalition negotiations.
考虑三党情景：A 有 45 席，B 35 席，C 20 席，共 100，多数门槛 51。联盟若总席位 ≥51 则获胜。某党的沙普利值是其按随机顺序加入联盟时成为关键者的概率。顺序 ABC：开始空集，A 45（未胜），加入 B → 80（胜）；B 关键。顺序 ACB：A 45，加 C → 65，C 关键。顺序 BAC：B 35，加 A → 80，A 关键。顺序 BCA：B 35，加 C → 55，C 关键。顺序 CAB：C 20，加 A → 65，A 关键。顺序 CBA：C 20，加 B → 55，B 关键。因此 A 在 6 种顺序中关键 2 次 → 沙普利值 2/6 = 1/3；B 也是 1/3；C 1/3。这种均分与席位份额 0.45、0.35、0.20 形成对比，揭示了小党在联盟谈判中可能拥有不成比例的议价能力。
9. Statistical Modelling: Democracy Index | 统计建模：民主指数
Researchers often quantify the ‘health’ of a democracy using composite indices such as the Democracy Index. These indices aggregate several indicators (electoral process, civil liberties, political participation) into a single score. From a statistical perspective, this involves data normalisation, weighting and sometimes regression analysis to relate scores to underlying factors.
研究者常使用诸如“民主指数”之类的复合指数来量化民主的“健康”程度。这些指数将多个指标（选举流程、公民自由、政治参与等）整合为一个分值。从统计学角度看，这涉及数据归一化、加权，有时还使用回归分析将得分与潜在因素联系起来。
ed using the least squares method. The coefficients βᵢ are estimated to minimise Σ(yᵢ − β₀ − β₁x₁ᵢ − β₂x₂ᵢ)². Standard hypothesis tests on each βᵢ (t‑ the corresponding variable has a significant linear relationship with the democracy score.
假设我们有 n 个国家的数据集，预测变量 x₁（人均 GDP）、x₂（新闻自由评分），响应变量 y（民主指数）。可以用最小二乘法拟合多元线性回归模型 y = β₀ + β₁x₁ + β₂x₂ + ε。系数 βᵢ 的估计通过最小化 Σ(yᵢ − β₀ − β₁x₁ᵢ − β₂x₂ᵢ)² 得到。对每个 βᵢ 的标准假设检验（t 检验）可以表明相应变量是否与民主得分有显著的线性关系。
Such models help political scientists test theories about what drives democratisation, and they rely directly on the inferential techniques taught in A‑level Statistics modules.
这类模型有助于政治学家检验关于民主化驱动力的理论，而这些模型直接依赖于 A‑level 统计模块所教授的推断技术。
10. From Mathematics to Democratic Design | 从数学到民主设计
Mathematics does not dictate how a democratic state must be run, but it reveals the logical consequences of each choice: which voting rule avoids paradoxes, how seat allocation formulas affect proportionality, and how power is truly distributed. Formal models of voting, apportionment and coalition power expose hidden biases and empower citizens to demand fairer systems.
数学并不规定民主国家应如何运作，但它揭示了每种选择的逻辑后果：哪种投票规则能避免悖论、议席分配公式如何影响比例性，以及权力究竟如何分布。关于投票、席位分配和联盟权力的形式化模型暴露了隐藏的偏倚，并赋予公民要求更公平制度的能力。
The mathematical concepts explored in this article – combinatorics, binomial probabilities, hypothesis tests, game‑theoretic indices and linear regression – form part of the Edexcel A‑level Mathematics and Further Mathematics specifications. Understanding them not only prepares students for exams but also for informed participation in the democratic process.
本文探讨的数学概念——组合学、二项概率、假设检验、博弈论指数和线性回归——是 Edexcel A‑level 数学和进阶数学课程大纲的一部分。理解这些内容不仅有助于备考，更能使他们在民主进程中获得知情参与的能力。
Published by TutorHao | Mathematics 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