📚 Globalisation Critique: A Statistical Perspective | 全球化批判：统计学视角
Globalisation is often presented as an unstoppable force for economic prosperity, connecting markets, ideas, and cultures. However, a closer look at the data reveals a more complex and uneven picture. In this article we apply your A-level statistical toolkit—averages, measures of spread, correlation, regression, and hypothesis testing—to critique the glossy narrative of globalisation. We will question whether ‘free trade lifts all boats’ or if the numbers mask rising inequality, environmental costs, and statistical manipulation.
全球化常被描绘成一股不可阻挡的力量，能为各国带来经济繁荣，将市场、思想和文化紧密相连。然而，深入审视数据便会发现一个更为复杂且不均衡的图景。本文将运用你在 A-level 阶段掌握的统计学工具——平均值、离散度量、相关、回归与假设检验——来批判光鲜的全球化叙事。我们将质疑“自由贸易能抬高所有船只”的说法是否成立，还是那些数字掩盖了日益加剧的不平等、环境代价以及统计操纵。
1. The Story of Averages: Why Mean Is Not Enough | 均值的故事：为何只谈平均数不够
When advocates claim that globalisation has raised average incomes in developing countries, they are often quoting the arithmetic mean. But the mean is highly sensitive to extreme values. A handful of billionaires can pull the average income upward, creating the illusion of widespread prosperity while the majority of households see little or no gain. Your A-level studies teach you that the median is a more resistant measure of central tendency when distributions are skewed. In globalisation debates, the choice between mean and median is a political act as much as a statistical one.
当全球化的支持者声称发展中国家的人均收入提高了，他们往往引用的是算术平均值。但均值对极端值极其敏感。少数亿万富翁就能拔高平均收入，制造出全社会普遍繁荣的幻象，而大多数家庭的收入却几乎毫无增长。你的 A-level 课程告诉你，当分布呈偏态时，中位数是更具抗干扰性的集中趋势度量。在全球化的争论中，选择均值还是中位数，既是统计选择，也是一种政治行为。
Consider a fictitious country, Globellia, where five families earn £5,000 per year and one multinational executive enters the economy with an income of £1 million. The mean income becomes (£1,025,000 / 6) ≈ £170,833, but the median remains £ median clearly reveals deep income divisions. Globalisation cheerleaders often commit this ‘sin of averaging’ to make their case look stronger than it really is.
设想一个虚构的国家“格罗贝利亚”，有五户家庭的年收入为 5,000 英镑，随后一位跨国公司高管进入该国，年收入为 100 万英镑。此时均值收入约为 (1,025,000 ÷ 6) ≈ 170,833 英镑，而中位数仍然是 5,000 英镑。如果只报告均值，会让人觉得中产阶级欣欣向荣；而中位数则清晰地揭示了严重的收入鸿沟。全球化的鼓吹者常常犯下这种“平均数的罪过”，让他们的论调远好于事实。
2. Income Inequality and the Gini Coefficient | 收入不平等与基尼系数
The Gini coefficient is a key measure of inequality derived from the Lorenz curve, which plots the cumulative proportion of income against the cumulative proportion of population. A perfectly equal society has a Gini coefficient of 0, while complete inequality gives a value of 1. In the context of globalisation, many nations that opened their markets rapidly in the 1990s and 2000s experienced an initial spike in Gini values. This challenges the assumption that trade liberalisation automatically narrows the gap between rich and poor.
基尼系数是度量不平等程度的关键指标，它来源于洛伦兹曲线——将累计收入比例与累计人口比例进行对比的一条曲线。一个完全平等的社会基尼系数为 0，而完全不平等则为 1。在全球化的背景下，许多在 1990 年代和 2000 年代迅速开放市场的国家都经历过基尼系数起初急剧攀升。这挑战了自由贸易会自动缩小贫富差距的假设。
rea under the Lorenz curve. For a quick calculation, an A-level student might use the formula G = (2 / (n² × x̄ ation of manufacturing jobs to lower-cost countries can increase the incomes of the top 10% ng wage growth for the middle class, thereby pushing the Gini upward in developed nations.
基尼系数可以简洁地表示为 G = A / (A + B)，其中 A 是完全平等线与洛伦兹曲线之间的面积，B 是洛伦兹曲线下方的面积。A-level 学生可以套用公式 G = (2 / (n² × x̄)) × Σ(i × xᵢ) - (n + 1) / n 进行快速计算，其中 xᵢ 是排序后的收入。我们能够看到，全球化将制造业岗位转移到低成本国家，可能提高了最富有的 10% 人口的收入，同时抑制了中产阶级的工资增长，从而推高了发达国家的基尼系数。
3. Correlation vs. Causation: The Trade–Growth Fallacy | 相关与因果：贸易与增长的谬误
A classic move in globalisation reports is to show a scatter plot of ‘trade openness’ against GDP growth and present a positive correlation as proof that free trade causes prosperity. But any A-level statistics student knows that correlation does not imply causation. There may be lurking variables—such as institutional quality, education levels, or colonial history—that drive both trade and growth. Without a controlled experiment, the observed r-value (Pearson correlation coefficient) is merely a descriptive number, not a causal decree.
全球化报告中一个经典的手法是出示一张“贸易开放度”与 GDP 增长的散点图，并把正相关当作自由贸易导致繁荣的证据。然而任何学过 A-level 统计的学生都知道，相关并不意味着因果。可能存在教育质量、制度水平或殖民历史等潜在变量，同时驱动贸易与增长。在没有对照实验的情况下，观测到的 r 值（皮尔逊相关系数）只是一个描述性数字，而非因果判定。
tween trade as a share of GDP and average annual growth might be statistically significant at the 5% , it would be reckless to conclude that removing tariffs always boosts growth. The relationship could be spurious, driven by a third factor such as technological advancement. Being critical of globalisation means demanding that advocates show the adjusted R² from a multiple regression, not just a simple bivariate correlation.
例如，计算得出贸易占 GDP 比重与年均增长率之间的 r = 0.68，可能在 5% 显著性水平下是显著的。但若不经进一步分析就贸然得出结论，认为取消关税总能促进增长，便是草率的。这种关系可能是伪相关，由技术进步等第三因素驱动。批判全球化意味着要求其支持者展示多元回归中的调整 R²，而不只是简单的双变量相关。
4. Regression Analysis: Does Globalisation Work for Everyone? | 回归分析：全球化惠及所有人吗？
ess) + b₂×(Education Index) + b₃×(Governance Score) + error. If we find that the coefficient b₁ is negative and significant, it suggests that trade openness is associated with lower poverty after controlling for other factors. However, critics argue that the model often fails to capture the experience of the poorest quintile because national averages hide within-country variation.
一个多元线性回归模型可以检验全球化能减少贫困的说法。我们可以设定模型：贫困率 = a + b₁×(贸易开放度) + b₂×(教育指数) + b₃×(治理得分) + 误差。如果我们发现系数 b₁ 是负的且显著，那就暗示在控制其他因素后，贸易开放与较低的贫困率相关联。然而，批评者认为这类模型常常无法体现最贫困五分之一人口的状况，因为国家平均水平掩盖了国内差异。
When we compute the regression line using least squares estimation, we must check residuals for homoscedasticity. In globalisation data, we often observe heteroscedasticity: variance increases as trade openness increases, because some highly open economies become incredibly rich (e.g., Singapore) while others with similar openness remain poor (e.g., some Sub-Saharan African nations). This violates one of the key assumptions of ordinary least squares regression. Ignoring such model diagnostics leads to overconfident praise of globalisation.
当我们用最小二乘法计算回归直线时，必须检查残差是否具有同方差性。在全球化数据中，我们经常观察到异方差性：方差随着贸易开放度的增加而增大，因为某些高度开放的经济体变得极为富裕（例如新加坡），而其他具有类似开放度的经济体却依然贫穷（例如一些撒哈拉以南非洲国家）。这违反了普通最小二乘回归的关键假设之一。忽视此类模型诊断就会导致对全球化的过高赞誉。
5. Hypothesis Testing: Does Free Trade Reduce Poverty? | 假设检验：自由贸易能减少贫困吗？
beralisation policy. Null hypothesis H₀: μ = 0 (no effect). Alternative hypothesis H₁: μ < 0 (poverty decreased). Using a paired t-test on data from 30 developing countries, we might obtain a sample mean difference of -2.3 percentage points with a sample standard deviation of 4.1. The test statistic t = (x̄ - 0) / (s / √n) = -2.3 / (4.1 / √30) ≈ -3.07. For a one-tailed test at 5% out -1.699. Since -3.07 < -1.699, we reject H₀ and claim poverty rates fell significantly.
我们可以将全球化争论构建为一个正规的假设检验。令 μ 为重大贸易自由化政策实施前后贫困人口比例的平均差值。原假设 H₀：μ = 0（无效果），备择假设 H₁：μ < 0（贫困减少）。利用 30 个发展中国家的数据做配对 t 检验，我们可能得到样本平均差值为 -2.3 个百分点，样本标准差为 4.1。检验统计量 t = (x̄ - 0) / (s / √n) = -2.3 / (4.1 / √30) ≈ -3.07。在 5% 显著性水平的单侧检验下，自由度为 29，临界值约为 -1.699。因为 -3.07 < -1.699，我们拒绝 H₀，并声称贫困率显著下降。
Yet this result can be critiqued. The sample might be biased—perhaps countries that expected gains were more likely to liberalise, a selection effect. Furthermore, a significant test does not measure the size of the effect. A reduction of 2.3 percentage points may be statistically significant but practically negligible if the initial poverty rate was 60%. A more honest assessment would include a confidence interval, e.g., a 95% CI for μ: (-3.7, -0.9). The interval barely avoids zero, indicating that globalisation’s poverty-reducing power is not as robust as promotional literature suggests.
然而这一结论可以受到批判。样本可能存在偏差——或许那些预期获益的国家更可能实施自由化，这是一种选择效应。此外，显著的检验结果并不能衡量效应的大小。如果初始贫困率为 60%，2.3 个百分点的降低在统计上显著，但在实际上可能微不足道。更诚实的评估应当包含置信区间，例如 μ 的 95% 置信区间：（-3.7，-0.9）。该区间刚刚避开零值，这表明全球化减少贫困的效力并不像宣传材料所说的那样稳健。
6. Outliers and the Celebration of 'Success Stories' | 异常值与“成功故事”的欢呼
Globalisation advocates often point to a few spectacular success stories—South Korea, China, Vietnam—as proof that integration into global markets lifts nations out of poverty. From a statistical viewpoint, these countries are outliers in any regression model linking trade liberalisation to growth. Removing them from a dataset can dramatically alter the slope of a trend line or change a significant p-value to non-significant. In A-level terms, these data points have high leverage and large influence on the coefficients.
全球化的倡导者常常举出几个耀眼的成功故事——韩国、中国、越南——作为融入全球市场能让国家摆脱贫困的证据。从统计学的角度来看，在任何将贸易自由化与增长联系起来的回归模型中，这些国家都是异常值。将它们从数据集中剔除会显著改变趋势线的斜率，或者将显著的 p 值变为不显著。用 A-level 的话说，这些数据点具有高杠杆率且对系数有很大影响。
Consider a dataset of 50 developing countries with trade openness as the explanatory variable and growth as the response. The regression line with all points may have a positive slope. But if we calculate Cook’s distance for each point, we find that South Korea and China have values exceeding 1, indicating substantial influence. Without them, the remaining 48 countries might show a flat or even negative relationship. This suggests that globalisation’s alleged benefits are not universally accessible; they are concentrated in a few specific historical‑political contexts.
考虑一个包含 50 个发展中国家的数据集，以贸易开放度为解释变量，以增长为响应变量。将所有点纳入后，回归线可能具有正的斜率。但是，如果我们计算每个点的库克距离，会发现韩国和中国的值超过 1，显示出重大影响。如果不包含它们，剩下的 48 个国家可能会表现出平缓甚至负向的关系。这表明全球化所谓的益处并非普遍可得，而是集中在少数特定的历史政治背景中。
7. Globalisation and Environmental Degradation: A Time Series Perspective | 全球化与环境退化：时间序列的视角
Globalisation accelerates the movement of goods across the planet, which increases carbon emissions from transport. A bivariate time series plot of world trade volume against atmospheric CO₂ concentration since 1980 shows a strong upward trend in both. However, time series data often suffer from non-stationarity, which can create spurious regressions. Two series with deterministic trends can appear highly correlated (high R²) even when they are completely unrelated in a causal sense.
全球化加速了货物在全球范围内的流动，从而增加了运输产生的碳排放。自 1980 年以来，世界贸易量与大气中 CO₂ 浓度的双变量时间序列图显示出两者均呈现强劲的上升趋势。然而，时间序列数据往往存在非平稳性，这会导致伪回归。两个具有确定性趋势的数列可以表现出很高的相关性（R² 很大），即便它们在因果意义上毫不相干。
O₂ as r = 0.95, seemingly damning for globalisation. But first differencing the series (Δyt = yt - yt-1 series comparisons, often used by environmental activists to attack globalisation, can be just as misleading as the correlations touted by pro-globalisation economists. Statistics demands rigour on both sides of the debate.
一项 A-level 式的调查可能计算出贸易量与 CO₂ 之间的相关系数 r = 0.95，似乎对全球化极为不利。但如果对数列进行一阶差分（Δyt = yt - yt-1）以消除趋势，就会显示出弱得多的相关性，或许只有 r = 0.2。这告诉我们，环境活动者常常用来抨击全球化的那种天真的时间序列比较，可能与亲全球化经济学家所吹嘘的相关性一样具有误导性。统计在辩论双方都要求严谨。
8. Sampling Bias: Who Gets Counted in Globalisation Data? | 抽样偏差：全球化数据算上了谁？
Much of the data underlying globalisation indices comes from surveys that are costly to administer. Wealthy countries collect extensive data, while low-income nations often lack reliable statistical infrastructure. Consequently, global averages are disproportionately influenced by the experience of richer economies. This sampling bias creates a picture of globalisation that overstates its benefits because the voices of those left behind are literally uncounted.
支撑全球化指数的许多数据来自费用高昂的调查。富裕国家收集详尽的数据，而低收入国家往往缺乏可靠的统计基础设施。因此，全球均值不成比例地受到较富裕经济体状况的影响。这种抽样偏差描绘出一幅高估全球化好处的图景，因为那些被落下的人的声音事实上未被计入。
In A-level sampling terms, this is a classic case of under-coverage. If we imagine a stratified sample designed to capture globalisation’s effects, the stratum of subsistence farmers in sub-Saharan Africa has a very low sampling fraction, while the stratum of European consumers has a high fraction. Any estimate of 'average global welfare gain' calculated from such a sample is biased upward. A more critical approach would require weighting the responses in inverse proportion to the sampling fraction, but this is rarely done in glossy international reports.
用 A-level 抽样的术语来说，这是典型的覆盖不足。如果设想一个旨在捕捉全球化效应的分层样本，撒哈拉以南非洲的自给农民这一层的抽样比很低，而欧洲消费者这一层的抽样比却很高。从这种样本计算出的任何“全球平均福利收益”估计值都会偏高。更具批判性的做法需要按照抽样比的倒数进行加权处理，但这种做法在光鲜的国际报告中鲜少使用。
9. Composite Indices and the Art of Hiding Disparity | 综合指数与隐藏差距的艺术
Think tanks often construct a 'Globalisation Index' by combining scores on trade, capital flows, migration, and technology transfer. The aggregation method—typically a simple arithmetic mean of sub-scores—can mask huge variation. A country might have a high globalisation score because it is a financial hub attracting massive capital inflows, yet its domestic population may experience wage stagnation and cultural homogenisation. The index provides a neat number that policymakers can cite, but it loses vital information about the distribution of gains.
智库常常通过将贸易、资本流动、移民和技术转让等方面的得分进行整合，构建出一个“全球化指数”。这种加总方法——通常是各子项得分的简单算术平均——可能掩盖巨大的差异。一个国家可能因作为金融中心吸引了大量资本流入而获得高全球化得分，但其国内民众可能正经历着工资停滞和文化同质化。这一指数提供了一个政策制定者可以引用的简洁数字，却丢失了关于利益分配的关键信息。
Students studying measures of location and spread will recognise that a single composite score tells us nothing about the range or interquartile range of its components. Suppose a nation gets its high index score mainly from financial openness (10/10) but scores only 2/10 on labour rights protection. Reporting only the average (6/10) conceals the exploitation that globalisation critics highlight. A responsible analyst would present a radar chart or a table of sub-scores, not just a headline figure.
学习过位置度量和离散度量的学生会认识到，单一的综合分数并不能告诉我们各组成部分的极差或四分位距。假设一个国家主要因金融开放度得到满分（10/10），而在劳工权利保护上只得到 2/10。仅报告平均分（6/10）就掩盖了全球化批评者所强调的剥削。一个负责任的分析者应当提供雷达图或子项得分表格，而不是单单一个醒目的数字。
Globalisation Index₁ = (Trade + Finance + Technology + Migration) / 4
全球化指数₁ = (贸易 + 金融 + 技术 + 移民) / 4
10. The Fair Trade Premium: A t-Test for Two Independent Samples | 公平贸易溢价：两个独立样本的 t 检验
The fair trade movement claims that it offers a superior model of globalisation by guaranteeing a minimum price to producers. We can test whether fair trade certification significantly raises the income of coffee farmers compared to conventional farmers. Let group F be fair trade farmers and group C be conventional. Using a two-sample t-test assuming unequal variances, we have: nF = 25, x̄_F = £2,850, sF = £520; nC = 30, x̄_C = £2,450, sC = £480.
公平贸易运动声称，它通过保证生产者最低价格，提供了一种更优越的全球化模式。我们可以检验公平贸易认证是否显著提高了咖啡农的收入。令 F 组为公平贸易农民，C 组为传统农民。使用假设方差不等的两个独立样本 t 检验，我们有：nF = 25，x̄_F = 2,850 英镑，sF = 520 英镑；nC = 30，x̄_C = 2,450 英镑，sC = 480 英镑。
The test statistic is t = (2850 - 2450) / √520²/25 + 480²/30 ≈ 400 / √10816 + 7680 ≈ 400 / √18496 = 400 / 136.0 ≈ 2.94. The degrees of freedom are approximated by the Welch–Satterthwaite formula, roughly 48. The critical t-value for a one-tailed test at 1% emium is statistically significant. This seems to support globalisation with a human face.
检验统计量 t = (2850 - 2450) / √520²/25 + 480²/30 ≈ 400 / √10816 + 7680 ≈ 400 / √18496 ≈ 400 / 136.0 ≈ 2.94。自由度通过 Welch–Satterthwaite 公式近似为 48。在 1% 显著性水平的单侧检验下，临界 t 值约为 2.41。由于 2.94 > 2.41，我们拒绝无差异的原假设，得出公平贸易溢价在统计上是显著的。这似乎支持了以人为本的全球化。
However, critics would point out that the sample sizes are small and may be self-selected. Farmers who join fair trade cooperatives may already be more skilled or politically connected. Moreover, the premium might not be sustainable if fair trade demand plateaus. The statistically significant result does not automatically validate the entire fair trade system as a scalable solution to globalisation’s flaws.
然而，批评者会指出样本量较小，且可能存在自选择。加入公平贸易合作社的农民可能本就技能更强或政治联系更好。此外，如果公平贸易的需求趋于平稳，溢价可能无法持续。具有统计显著性的结果并不会自动证明整个公平贸易体系就是可扩展的全球化缺陷解决之道。
11. The Standard Deviation Story: Winners and Losers within Countries | 标准差的故事：一国之中的赢家与输家
Globalisation can increase the standard deviation of incomes within a previously homogenous region. Consider a town where a multinational firm sets up a factory. Some skilled workers receive high wages (say £40,000) while those in traditional crafts lose their livelihoods (falling to £8,000). The mean income might rise from £18,000 to £20,000, but the standard deviation jumps sharply. Reporting only the modest increase in mean hides the fact that the spread of incomes has widened, creating a more fractured community.
全球化可能加大原本同质化地区内部收入的标准差。设想一个小镇，一家跨国公司在此设立了工厂。一部分技术工人获得高工资（比如 40,000 英镑），而传统手工业者则失去了生计（收入跌至 8,000 英镑）。平均收入可能从 18,000 英镑上升到 20,000 英镑，但标准差却急剧上升。只报告均值的温和增长，掩盖了收入差距正在拉大以及社区正在变得更加割裂的事实。
atistics exercise, you might calculate the variance before and after: before, s² = 2,500 (£²); after, s² = 12,000 (£²). The coefficient of variation (CV = s / x̄ oubling relative variability. This measure better captures the human cost of globalisation—the simultaneous creation of a small number of winners and a large number of losers—that the headline growth figure disguises.
在 A-level 统计练习中，你可以计算前后的方差：之前 s² = 2,500（英镑²），之后 s² = 12,000（英镑²）。变异系数 (CV = s / x̄) 从 0.087 升至 0.173，相对变异几乎翻了一番。这一度量更好地捕捉了全球化的人力成本——它同时创造了一小部分赢家与大量输家——而这一现象却被标题式的增长数字所掩盖。
12. Conclusion: Statistical Literacy as a Shield Against Hype | 结论：统计素养是抵御浮夸的盾牌
Mathematics cannot tell us whether globalisation is inherently good or bad, but it equips us to interrogate the data behind the claims. By scrutinising the choice of averages, testing for causality, examining residuals, and acknowledging sample biases, we see that the story is far more nuanced than either side of the debate often admits. For an A-level student, learning to criticise the use of statistics is just as important as learning to compute them.
数学并不能告诉我们全球化本质上是好是坏，但它赋予我们审视主张背后数据的能力。通过仔细检查平均数的选择、检验因果关系、审视残差并正视样本偏差，我们就会发现，现实远比争论双方的常见说辞更为微妙。对于一名 A-level 学生而言，学会批判性看待统计的运用，和学习如何计算它们同等重要。
The next time you encounter a headline like 'Globalisation lifts 500 million out of poverty', ask yourself: which measure of poverty? Over what time period? Which countries are included, and who is omitted? This statistical curiosity is the true spirit of A-level Mathematics, and it will serve you well in a world awash with data. The critique of globalisation is, at its heart, a lesson in not taking numbers at face value.
下次你看到诸如“全球化使 5 亿人摆脱贫困”这样的标题时，问问自己：采用了哪种贫困衡量指标？哪个时间段？纳入了哪些国家，又遗漏了谁？这种统计上的好奇心才是 A-level 数学的真正精神，它将助力你在一个数据泛滥的世界里游刃有余。对全球化的批判，归根结底是一堂不要轻信数字表面价值的课。
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