📚 The case for and against referendums in a representative democracy | 代议制民主中公投的利弊：数学视角分析
In a representative democracy, decisions are typically delegated to elected officials, yet the referendum—a direct vote by the electorate on a specific issue—remains a powerful and controversial tool. Mathematical modelling offers a rigorous way to evaluate its strengths and weaknesses, moving beyond rhetoric to analyse voting paradoxes, statistical representativeness, and strategic behaviour. This article explores the quantitative case for and against referendums through the lens of A‑Level Mathematics, including logic, probability, and decision theory.
在代议制民主中，决策通常由民选代表做出，但公投——选民就某一特定议题进行的直接投票——仍然是一种强大且充满争议的工具。数学建模提供了一种严谨的评估方式，超越修辞，分析投票悖论、统计代表性以及策略行为。本文将通过 A‑Level 数学的视角，包括逻辑、概率和决策理论，探讨支持与反对公投的量化论证。
1. The Condorcet Paradox: Why Majority Rule Can Be Cyclical | 孔多塞悖论：为何多数规则可能产生循环
When three or more options are presented, pairwise majority voting can lead to intransitive collective preferences, even if every individual has rational, transitive preferences. Consider three voters with ranked choices over options A, B, C: Voter 1: A≻B≻C, Voter 2: B≻C≻A, Voter 3: C≻A≻B. A beats B by 2–1, B beats C by 2–1, yet C beats A by 2–1, forming a cycle A≻B≻C≻A. No option is a clear ‘will of the people’, undermining the legitimacy of any referendum result that relies on a single pairwise choice.
当存在三个或更多选项时，两两多数票决可能导致集体偏好出现非传递性，即便每个个体都有理性、可传递的偏好。假设三位选民对选项 A、B、C 的排序为：选民1：A≻B≻C，选民2：B≻C≻A，选民3：C≻A≻B。A 以 2–1 击败 B，B 以 2–1 击败 C，但 C 以 2–1 击败 A，形成循环 A≻B≻C≻A。没有任何选项能代表清晰的“人民意志”，这削弱了任何依赖单一二元选择的公投结果的合法性。
2. Arrow’s Impossibility Theorem: No Perfect Voting System | 阿罗不可能定理：无完美的投票系统
Kenneth Arrow proved that no ranked voting system can simultaneously satisfy a set of seemingly minimal fairness conditions: unrestricted domain, non‑dictatorship, Pareto efficiency, and independence of irrelevant alternatives (IIA). Any referendum design—whether simple majority, two‑round, or alternative vote—must violate at least one of these. For instance, the IIA condition is often breached when a third option dropped from the ballot alters the relative ranking of the remaining two, meaning the framing of a binary referendum can manipulate the outcome.
肯尼斯·阿罗证明，任何基于排序的投票制度都无法同时满足一组看似最低限度的公平条件：无限制定义域、非独裁、帕累托效率以及无关选项独立性（IIA）。任何公投设计——无论是简单多数制、两轮投票还是排序复选制——都必然违反其中至少一个条件。例如，当从选票中移除第三个选项会改变剩余两个选项的相对排序时，IIA 条件即被破坏，这意味着二元公投的议题框定可以操纵结果。
3. Turnout and Statistical Representativeness | 投票率与统计代表性
Referendum outcomes are often treated as the voice of the entire electorate, yet low turnout can produce large sampling errors. If turnout is only 40%, the result can be modelled as a sample from the population of eligible voters. The margin of error for a proportion p̂ is approximately z × √[p̂(1−p̂)/n], where n is the number of actual voters. With a close result like 52% to 48% and n = 10 million, the 95% confidence interval can span several percentage points, potentially reversing the apparent majority. This statistical uncertainty challenges claims of a decisive mandate.
公投结果常被当作全体选民的声音，但低投票率可能产生很大的抽样误差。若投票率仅为 40%，结果可视为从合格选民总体中抽取的样本。比例 p̂ 的误差范围约为 z × √[p̂(1−p̂)/n]，其中 n 为实际投票人数。若出现 52% 对 48% 的接近结果且 n = 1000 万，其 95% 置信区间可跨越数个百分比，足以逆转表面的多数。这种统计不确定性对声称获得明确授权构成挑战。
4. The Condorcet Jury Theorem: Wisdom of Crowds Under Conditions | 孔多塞陪审团定理：条件化下群体的智慧
If each voter has an independent probability p > 0.5 of choosing the ‘correct’ option, then as the number of voters N increases, the probability that a majority vote selects the correct outcome approaches 1. This supports referendums when the electorate is well‑informed and unbiased. Mathematically, the majority accuracy is P(N) = Σ (from k=⌈N/2⌉ to N) C(N,k) p^k (1−p)^(N−k). However, if p < 0.5 due to misinformation, the majority becomes increasingly likely to err as N grows, making a large‑scale referendum dangerously amplification of error.
若每位选民选择“正确”选项的独立概率 p > 0.5，那么随着选民数量 N 的增加，多数票选中正确结果的概率趋近于 1。这在选民信息充分且无偏见的条件下支持公投。数学上，多数正确的概率为 P(N) = Σ (from k=⌈N/2⌉ to N) C(N,k) p^k (1−p)^(N−k)。然而，若因误导信息导致 p < 0.5，随着 N 增大，多数犯错的可能性反而增加，使得大规模公投成为危险的错误放大机制。
5. Bayesian Framing Effects: How Question Wording Shifts Priors | 贝叶斯框架效应：问题措辞如何改变先验概率
Referendum questions are rarely neutral; their wording can activate different prior beliefs in voters’ minds. By Bayes’ theorem, the posterior probability a voter assigns to a proposition after reading the question is P(H|E) = [P(E|H) × P(H)] / P(E). Slight changes in prompted framing (the ‘evidence’ E) can shift posterior support significantly, even if the factual content is identical. A question like ‘Should the UK remain in the EU?’ versus ‘Should the UK leave the EU?’ may trigger different mental models, making the outcome sensitive to agenda‑setting rather than true preferences.
公投问题极少中立；其措辞能激活选民心中不同的先验信念。根据贝叶斯定理，选民在阅读问题后赋予一个命题的后验概率为 P(H|E) = [P(E|H) × P(H)] / P(E)。略微改变所引发的框定（即“证据”E）就能显著改变后验支持度，即便事实内容相同。诸如“英国是否应留在欧盟？”与“英国是否应脱离欧盟？”的问题可能激活不同的心智模型，使结果更易受议题设定影响而非反映真实偏好。
6. Decision Theory: Rational Ignorance and the Marginal Voter | 决策论：理性无知与边际选民
From an expected utility standpoint, the incentive for a citizen to become informed before a referendum is minuscule. The probability of a single vote being pivotal in a large electorate is close to zero, so the expected benefit of deep research is dwarfed by its cost. This ‘rational ignorance’ means many voters rely on heuristics or partisan cues rather than substantive analysis, undermining the epistemic quality of the collective decision. Expected value of voting = p × B − C, where p is pivot probability, B is the benefit of one’s preferred outcome winning, and C is the cost of becoming informed. For large N, p → 0, making informed voting individually irrational.
从期望效用视角看，公民在公投前获取信息的激励微乎其微。在大规模选民中，单一选票成为关键票的概率趋近于零，因而深入研究的期望收益远低于其成本。这种“理性无知”意味着许多选民依赖启发法或党派线索而非实质性分析，损害了集体决策的认知品质。投票期望值 = p × B − C，其中 p 为关键票概率，B 为心仪结果获胜的收益，C 为获取充足信息的成本。对于大 N，p → 0，使得充分知情投票在个体层面非理性。
7. Game Theory: Referendums as Coordination Devices | 博弈论：公投作为协调机制
In representative democracies, legislatures can become deadlocked on contentious issues. A referendum can serve as a coordination device, providing a focal point that resolves a game of chicken between parties. Consider two parties each preferring their own policy but fearing the electoral cost of compromise. The promise of a binding referendum transforms the game into a Stag Hunt where both gain by accepting the popular verdict, avoiding mutually destructive stalemate. This is a positive mathematical argument for referendums when representative institutions fail to aggregate preferences efficiently.
在代议制民主中，立法机关可能在有争议的议题上陷入僵局。公投可作为一种协调机制，提供一个聚点以化解政党间的斗鸡博弈。假设两个政党各自偏好自身政策，但又担心妥协带来的选举代价。绑定性公投的承诺将博弈转变为猎鹿博弈，双方通过接受民意裁决而获益，避免两败俱伤的僵局。当代表机构无法有效聚合偏好时，这为公投提供了一个正面的数学论据。
8. Gerrymandering of the Plebiscite: Strategic Date and Threshold Setting | 公投的变相操纵：策略性日期与门槛设定
The timing of a referendum can be modelled as an optimisation problem for the incumbent. By choosing a date when their support is seasonally high or when opposition turnout is expected to be low, the initiator shifts the sample of actual voters away from a random draw. Additionally, imposing supermajority thresholds or quorum rules changes the probability of success. A mathematical analysis using binomial distributions shows that a 60% supermajority requirement drastically lowers the probability of passing legislation even if true support is 55%, creating a status‑quo bias that can entrench minority rule.
公投的时机选择可被建模为执政者的一个最优化问题。通过选择一个己方支持率季节性高涨或预期反对派投票率较低的日期，发起者使得实际投票者样本偏离随机抽样。此外，设定绝对多数门槛或法定人数规则会改变成功的概率。利用二项分布的数学分析显示，即使真实支持度为 55%， 60% 的绝对多数要求也会显著降低法案通过的概率，制造出固化少数派统治的现状偏见。
9. Regression Analysis of Money and Media Influence | 资金与媒体影响的回归分析
Empirical studies using multiple regression have shown that spending by campaigns and biased media coverage can significantly predict referendum outcomes, controlling for underlying public sentiment. A model Y = β₀ + β₁X₁ + β₂X₂ + ε, where Y is the ‘Yes’ vote share, X₁ is campaign spending imbalance, and X₂ is a media slant index, often yields significant positive β₁. This suggests that outcomes can be purchased or engineered, distorting the ideal of a pure expression of popular will. Mathematical modelling confirms that unequal resources violate the assumption of equal and independent voter decision‑making.
利用多元回归的实证研究表明，在控制基本公众情感的情况下，竞选活动的花费和有偏见的媒体报道能显著预测公投结果。模型 Y = β₀ + β₁X₁ + β₂X₂ + ε，其中 Y 为“赞成”票份额，X₁ 为竞选花费的不平衡度，X₂ 为媒体倾斜指数，常得出显著的正向 β₁。这表明公投结果可能被收买或操纵，扭曲了纯粹表达公意的理想。数学建模证实，资源不平等违反了选民独立决策且影响均等的假定。
10. Counterargument: Referendums as a Randomised Check on Elites | 反论点：公投作为对精英的随机化制衡
From a statistical perspective, occasional referendums can be seen as a form of stratified sampling or randomised audit of representative decisions. If representatives’ decisions are modelled as a stochastic process that sometimes deviates from the median voter, a referendum resets the mean with high certainty. The law of large numbers ensures that a well‑designed, multi‑issue deliberative poll can approximate the true preference distribution. Thus, mathematically, referendums need not be abolition of representation but a complementary zero‑force mechanism that restores alignment when divergence exceeds a critical threshold Δ.
从统计学视角看，偶尔举行的公投可被视为对代议决策的一种分层抽样或随机化审计。若将代表的决策建模为一个有时偏离中间选民的随机过程，那么公投即是以高确定性重置均值。大数定律保证，一个精心设计、多议题的协商式民调能够逼近真实的偏好分布。因此，从数学上说，公投不必是代议制的废除，而是一种互补的归零机制，当偏差超过临界阈值 Δ 时恢复一致性。
11. The Risk of Polarisation Feedbacks: Nonlinear Dynamics | 两极化反馈的风险：非线性动力学
Referendums can trigger positive feedback loops in opinion dynamics. Using differential equation models of social influence, dX/dt = k X (1 − X)(p − q) where X is the proportion supporting a view, p is persuasion rate among supporters, and q among opponents, a binary divisive referendum can increase k by amplifying group identity. This can push the system toward a stable extreme equilibrium, reducing the moderate centre. Such bifurcation analysis warns that referendums on identity‑heavy issues may irreversibly fracture the electorate, a mathematical case against their frequent use.
公投可能触发舆论动力学中的正反馈循环。利用社会影响的微分方程模型，dX/dt = k X (1 − X)(p − q)，其中 X 为支持某一观点的比例，p 为支持者之间的劝服率，q 为反对者之间的劝服率，一个二元分裂性的公投可通过放大群体认同而增加 k 值。这会将系统推向一个稳定的极端均衡，缩小温和中间派。这种分岔分析警示，在身份色彩浓厚的议题上举行公投可能不可逆地撕裂选民，这是反对频繁使用公投的数学依据。
12. Conclusion: A Probabilistic Defence of Circumscribed Use | 结论：有限使用的概率性辩护
Mathematical reasoning reveals that referendums are neither inherently democratic nor inherently dangerous; their value is highly contingent on design parameters such as turnout, threshold, question neutrality, and voter information. When conditions approximate the assumptions of the Condorcet Jury Theorem—large N, p > 0.5, independence—referendums can enhance legitimacy. When they violate those conditions, they amplify noise and bias. A mathematically informed representative democracy would use referendums sparingly, with robust statistical safeguards, treating them as calibrated instruments rather than populist panaceas.
数学推理表明，公投既非天然民主，也非天然危险；其价值高度取决于设计参数，如投票率、门槛、议题中立性以及选民信息水平。当条件接近孔多塞陪审团定理的假定——大 N、p > 0.5、独立性——公投可增强合法性。当这些条件被违反时，它们则会放大噪声与偏差。一个有数学素养的代议制民主应当审慎地使用公投，并配备健全的统计保障，将其视作校准过的工具而非民粹主义的万灵药。
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