📚 The Factors That Explain the Outcome of These Elections | 解释这些选举结果的因素
An election result is not just a political headline; it can be treated as a data-generating process. In Edexcel A-Level Mathematics, applied statistics gives us a set of tools to identify possible factors, measure their strength, and decide whether an apparent pattern is statistically meaningful or simply due to chance. This article explains how concepts such as sampling, confidence intervals, correlation, regression, hypothesis testing, and probability models can be used to understand why elections produce particular outcomes.
选举结果不仅是政治头条，也可以被视为一个数据生成过程。在 Edexcel A-Level 数学中，应用统计学为我们提供了一套工具，用于识别可能的影响因素、衡量其强度，并判断表面模式在统计上是否显著或仅由偶然造成。本文将解释抽样、置信区间、相关、回归、假设检验和概率模型等概念如何用于理解选举为何产生特定结果。
1. Elections as a Statistical Problem | 选举作为一个统计问题
An election result is a set of observed values: vote shares, turnout rates, and seat counts. In statistics we define a response variable, such as the percentage vote for a party in a constituency, and explanatory variables, such as unemployment rate, age profile, or campaign spending. These variables allow us to test which factors have a measurable influence on the result.
选举结果是一组观测值：得票率、投票率和席位数量。在统计学中，我们定义响应变量，例如某选区中一个政党的得票率，以及解释变量，例如失业率、年龄构成或竞选支出。这些变量使我们能够检验哪些因素对结果有可衡量的影响。
In Edexcel A-Level Mathematics, the applied statistics section covers data collection, presentation, probability, correlation and regression, discrete and continuous distributions, and hypothesis testing. Each of these topics can be applied directly to electoral data. For example, a scatter diagram of unemployment against incumbent vote share can reveal whether economic conditions are associated with election outcomes.
在 Edexcel A-Level 数学中，应用统计部分涵盖数据收集、数据呈现、概率、相关与回归、离散与连续分布以及假设检验。这些主题中的每一个都可以直接应用于选举数据。例如，失业率与执政党得票率的散点图可以揭示经济状况是否与选举结果相关。
2. Variables and Data Collection | 变量与数据收集
Before any analysis, we must decide what to measure. The response variable might be the share of votes won by the incumbent party in a general election. Explanatory variables could include turnout, median income, percentage of young voters, immigration rate, and campaign expenditure. Good data collection requires a clear definition of each variable and a consistent unit of measurement across constituencies.
在进行任何分析之前，我们必须决定要测量什么。响应变量可以是执政党在大选中的得票率。解释变量可以包括投票率、中位收入、年轻选民比例、移民率和竞选支出。良好的数据收集要求清晰定义每个变量，并在各选区之间使用一致的测量单位。
In the exam, you may be given a table of data from several constituencies and asked to plot a scatter diagram or calculate summary statistics. Always identify which column is the explanatory variable and which is the response variable. Mixing them up changes the direction of any regression line and weakens the conclusion.
在考试中，你可能会得到一张来自多个选区的数据表，并被要求绘制散点图或计算汇总统计量。务必要确定哪一列是解释变量，哪一列是响应变量。混淆两者会改变回归直线的方向，并削弱结论。
3. Sampling and Polling Error | 抽样与民调误差
Opinion polls attempt to estimate the true level of support for a party by surveying a random sample of voters. If x out of n sampled voters say they will vote for party A, the sample proportion is p̂ = x / n. This is an unbiased estimator of the true proportion p, but different samples give slightly different values of p̂.
民意调查通过访问选民的随机样本来估计一个政党的真实支持率。如果在 n 位受访选民中有 x 位表示将投票给 A 党，则样本比例为 p̂ = x / n。这是真实比例 p 的无偏估计量，但不同样本会给出略有不同的 p̂ 值。
The standard error of the sample proportion is given by:
样本比例的标准误差由下式给出：
SE = √( p(1 − p) / n )
A larger sample size n reduces the standard error, making the poll more precise. This is why national polls with 1,000 or more respondents are usually more reliable than small local surveys.
较大的样本容量 n 会降低标准误差，使民调更加精确。这就是为什么拥有 1,000 名或更多受访者的全国性民调通常比小型地方调查更可靠。
4. Confidence Intervals for Vote Share | 投票率的置信区间
A 95% confidence interval for a party’s true vote share p is calculated as:
一个政党真实得票率 p 的 95% 置信区间计算如下：
p̂ ± 1.96 × √( p̂(1 − p̂) / n )
If the confidence intervals for two competing parties overlap, the poll cannot be confident about which party is ahead. This often happens in marginal constituencies where the gap between parties is smaller than the margin of error.
如果两个竞争政党的置信区间重叠，民调就无法有把握地确定哪个政党领先。这种情况经常发生在边缘选区，因为政党之间的差距小于误差范围。
For example, if party A polls at 52% and party B at 48% with a margin of error of ±3%, the intervals are [49%, 55%] and [45%, 51%]. Because they overlap, the election outcome is genuinely uncertain. This statistical uncertainty is a key factor in explaining unexpected results.
例如，如果 A 党民调为 52%，B 党为 48%，误差幅度为 ±3%，则置信区间为 [49%, 55%] 和 [45%, 51%]。由于两者重叠，选举结果确实存在不确定性。这种统计不确定性是解释意外结果的关键因素。
5. Correlation and Regression with Economic Factors | 与经济因素的相关与回归
Pearson’s product moment correlation coefficient r measures the strength and direction of a linear relationship between two variables, such as unemployment rate x and incumbent vote share y. The value of r lies between −1 and +1. A negative r suggests that constituencies with higher unemployment tend to give lower support to the incumbent party.
皮尔逊积矩相关系数 r 衡量两个变量之间线性关系的强度和方向，例如失业率 x 与执政党得票率 y。r 的取值在 −1 到 +1 之间。负的 r 表明失业率较高的选区往往对执政党的支持率较低。
The least squares regression line is written as:
最小二乘回归直线写为：
y = a + bx
Here b is the gradient, representing the estimated change in vote share for a one-unit increase in unemployment. The intercept a is the predicted vote share when unemployment is zero, though this may not be meaningful in real contexts.
其中 b 是斜率，表示失业率每增加一个单位时得票率的估计变化量。截距 a 是失业率为零时的预测得票率，但在实际情境中可能没有意义。
When interpreting a regression line, always comment on the direction and size of b, and remember that correlation does not imply causation. A third factor, such as regional identity, could affect both unemployment and voting behaviour.
在解释回归直线时，务必评论 b 的方向和大小，并记住相关并不意味着因果。第三个因素，例如地区认同，可能同时影响失业率和投票行为。
6. Hypothesis Testing for a Factor | 对某一因素进行假设检验
To test whether an explanatory variable has a statistically significant effect on election outcomes, we use a hypothesis test. The null hypothesis states that the factor has no effect, while the alternative hypothesis states that it does have an effect.
为了检验某个解释变量是否对选举结果具有统计上显著的影响，我们使用假设检验。零假设表明该因素没有影响，而备择假设表明该因素确有影响。
H₀: β = 0 and H₁: β ≠ 0
The test statistic is the estimated coefficient divided by its standard error:
检验统计量为估计系数除以其标准误差：
t = b / SE(b)
Under the null hypothesis, this statistic follows a t-distribution with n − 2 degrees of freedom for simple linear regression. If the p-value is less than the significance level, usually 0.05, we reject H₀ and conclude that the factor is statistically significant.
在零假设下，该统计量服从自由度为 n − 2 的 t 分布（对于简单线性回归）。如果 p 值小于显著性水平，通常为 0.05，我们拒绝 H₀，并得出该因素在统计上显著的结论。
For example, a test might show that campaign spending has a significant positive effect on vote share, while the effect of average rainfall on polling day is not significant. This helps researchers focus on the factors that truly explain outcomes.
例如，检验可能显示竞选支出对得票率有显著的正向影响，而投票日平均降雨量的影响不显著。这有助于研究者聚焦于真正解释结果的因素。
7. Probability Models of Voter Turnout | 投票率的概率模型
Voter turnout can be modelled using a binomial distribution. If there are n registered voters in a constituency and each has probability p of voting, the number of voters X follows a binomial distribution:
投票率可以用二项分布建模。如果某选区有 n 位登记选民，且每位选民投票的概率为 p，则投票人数 X 服从二项分布：
X ~ B(n, p)
The expected number of voters is E(X) = np and the variance is Var(X) = np(1 − p). Any factor that changes p, such as weather, distance to polling station, or voter ID requirements, directly alters the distribution of turnout.
投票人数的期望值为 E(X) = np，方差为 Var(X) = np(1 − p)。任何改变 p 的因素，例如天气、到投票站的距离或选民身份要求，都会直接改变投票率的分布。
In close seats, even a small change in p can flip the result. This is why election campaigns invest heavily in get-out-the-vote operations targeting groups with historically low turnout probabilities.
在竞争激烈的席位上，即使 p 的微小变化也可能逆转结果。这就是为什么竞选活动投入大量资源进行催票行动，针对历史上投票概率较低的群体。
8. Normal Approximation and Polling Distributions | 正态近似与民调分布
When n is large and p is not too close to 0 or 1, the binomial distribution can be approximated by a normal distribution with mean np and variance np(1 − p):
当 n 很大且 p 不接近 0 或 1 时，二项分布可以用均值为 np、方差为 np(1 − p) 的正态分布来近似：
X ≈ N( np, np(1 − p) )
This approximation allows us to calculate probabilities of different turnout levels. The standardised value is:
该近似使我们能够计算不同投票率水平的概率。标准化值为：
z = ( X − np ) / √( np(1 − p) )
Using standard normal tables, we can estimate the probability that a party wins a seat given a particular lead in the polls. If the calculated probability is only 0.55, the result is far from certain, so small shifts in voter behaviour can change the outcome.
使用标准正态表，我们可以估计在民调中处于特定领先位置时某政党赢得席位的概率。如果计算出的概率仅为 0.55，那么结果远非确定，因此选民行为的微小变化就可能改变结局。
9. Conditional Probability and Swing Voters | 条件概率与摇摆选民
Conditional probability is a powerful tool for explaining how different groups respond to election issues. If A is the event ‘voter is aged 18 to 24’ and B is ‘voter supports party X’, then P(B|A) is the probability that a young voter supports party X. This may be very different from the overall probability P(B).
条件概率是解释不同群体如何对选举议题作出反应的有力工具。如果 A 是“选民年龄在 18 至 24 岁”这一事件，B 是“选民支持 X 党”，则 P(B|A) 是年轻选民支持 X 党的概率。这可能与总体概率 P(B) 有很大差异。
The formula is:
公式为：
P(B|A) = P(A ∩ B) / P(A)
A policy such as tuition fee changes can alter P(B|A), shifting the overall vote share even if the size of the group remains the same. Thus, understanding conditional probabilities helps explain why campaigns target specific demographic groups.
像学费变化这样的政策可以改变 P(B|A)，即使该群体的规模保持不变，也会改变整体得票率。因此，理解条件概率有助于解释为什么竞选活动针对特定的人口群体。
10. Data Presentation and Misleading Graphs | 数据呈现与误导性图表
Edexcel A-Level Mathematics places strong emphasis on accurate data presentation. A misleading graph can overstate the influence of a factor. For example, truncating the y-axis of a vote share chart can make a small change of 1% look like a dramatic swing. Omitting relevant variables can also create false impressions of causality.
Edexcel A-Level 数学非常强调准确的数据呈现。误导性图表会夸大某个因素的影响。例如，截断得票率图表的 y 轴会使 1% 的小变化看起来像剧烈的摇摆。遗漏相关变量也会造成虚假的因果印象。
Good practice involves using clear histograms, box plots, and scatter diagrams with labelled axes and appropriate scales. Box plots are particularly useful for comparing turnout distributions across regions, while scatter diagrams reveal relationships between variables.
良好做法包括使用清晰的直方图、箱线图和散点图，并标注坐标轴和适当的刻度。箱线图特别适合比较各地区的投票率分布，而散点图则能揭示变量之间的关系。
11. Combining Factors: Multiple Regression | 综合因素：多元回归
Real election outcomes are rarely explained by a single variable. Multiple regression extends the linear model to include several explanatory variables simultaneously:
真实的选举结果很少能由单一变量解释。多元回归将线性模型扩展为同时包含多个解释变量：
y = a + b₁x₁ + b₂x₂ + … + bₖxₖ
Here y could be the incumbent vote share, x₁ the unemployment rate, x₂ the percentage of voters with a university degree, and x₃ the campaign spending per voter. Each b coefficient measures the unique effect of that variable while holding the others constant.
其中 y 可以是执政党得票率，x₁ 是失业率，x₂ 是拥有大学学位的选民比例，x₃ 是每位选民的竞选支出。每个 b 系数衡量该变量在保持其他变量不变时的独特效应。
Multiple regression is a more realistic model because it controls for confounding variables. However, it also requires larger data sets and careful checking of assumptions such as linearity and absence of multicollinearity between explanatory variables.
多元回归是一个更现实的模型，因为它控制了混杂变量。然而，它也需要更大的数据集，并仔细检查线性假设以及解释变量之间是否存在多重共线性。
12. Conclusion and Exam Tips | 结论与考试提示
Election outcomes can be explained by a combination of statistical factors: sampling variability, economic variables, turnout probabilities, demographic effects, and measurement errors. In Edexcel A-Level Mathematics, exam questions often ask you to interpret r, construct a confidence interval, carry out a hypothesis test, or comment on a regression model in an election-based context.
选举结果可以由一系列统计因素共同解释：抽样变异性、经济变量、投票概率、人口效应和测量误差。在 Edexcel A-Level 数学考试中，题目经常要求你在选举情境下解释 r、构建置信区间、进行假设检验或评论回归模型。
Always define your variables clearly, state hypotheses using correct notation, show all steps in calculations, and finish with a conclusion in the context of the question. Remember that statistical significance does not automatically mean a factor is practically important, and correlation does not prove causation.
务必清晰定义变量，使用正确符号陈述假设，展示计算的所有步骤，并在题目情境下给出结论。请记住，统计显著并不自动意味着该
Published by TutorHao | A-Level Mathematics 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