A-Level Mathematics: Correlation — How to Judge and Interpret | A-Level 数学：相关性的判断与解释

📚 A-Level Mathematics: Correlation — How to Judge and Interpret | A-Level 数学：相关性的判断与解释

In A-Level Mathematics and Statistics, correlation is a measure of the strength and direction of a linear relationship between two variables. It is one of the most frequently tested topics in data interpretation, especially within the context of bivariate data. This article will guide you through the key concepts, formulas, and common pitfalls when judging and interpreting correlation.

在 A-Level 数学与统计学中，相关性是衡量两个变量之间线性关系的强度与方向的重要工具。它也是双变量数据考题中最常出现的知识点之一。本文将系统梳理相关性的核心概念、公式以及判断与解释时的常见误区，帮助你在考试中准确作答。

1. What Is Correlation? | 什么是相关性？

Correlation refers to a statistical association between two quantitative variables. When we plot two variables on a scatter diagram, the pattern of points reveals whether a linear relationship exists. If the points tend to rise together, we say there is a positive correlation; if one increases while the other decreases, the correlation is negative. If the points show no clear pattern, there is little or no correlation.

相关性是指两个定量变量之间存在的统计联系。当我们在散点图上绘制两个变量时，点的分布模式可以揭示它们之间是否存在线性关系。如果点总体上一起上升，则称为正相关；如果一个变量增大而另一个变量减小，则为负相关；如果点没有明确模式，则相关性很弱或不存在。

The most common numerical measure of correlation is Pearson’s product-moment correlation coefficient, denoted by r. It is a dimensionless quantity that always lies between -1 and 1.

最常用的相关性数值度量是皮尔逊积矩相关系数，记作 r。它是一个无量纲的量，取值范围始终在 -1 与 1 之间。

r = Sₓᵧ / √(Sₓₓ × Sᵧᵧ)

where Sₓᵧ is the sum of products of deviations from the means, Sₓₓ is the sum of squared deviations for x, and Sᵧᵧ is the sum of squared deviations for y.

其中 Sₓᵧ 是离差乘积之和，Sₓₓ 是 x 的离差平方和，Sᵧᵧ 是 y 的离差平方和。

2. Properties of the Correlation Coefficient | 相关系数的性质

The correlation coefficient r has several important properties you must remember.

相关系数 r 有几个重要的性质需要牢记。

r is always between -1 and 1. The closer r is to 1, the stronger the positive linear relationship; the closer r is to -1, the stronger the negative linear relationship. A value near 0 indicates a very weak or no linear relationship.
r is unit-free. It does not depend on the units of measurement. For example, changing heights from cm to m does not change r.
r is symmetric. The correlation between x and y is the same as the correlation between y and x, so r(x, y) = r(y, x).
r only measures linear relationships. A strong non-linear relationship may still give a small r value.

r 始终在 -1 与 1 之间。r 越接近 1，正线性关系越强；r 越接近 -1，负线性关系越强；r 接近 0 表示线性关系很弱或不存在。

r 没有单位，不依赖于测量单位。例如，将高度从厘米换成米，r 的值不变。

r 具有对称性。x 与 y 的相关系数等于 y 与 x 的相关系数，即 r(x, y) = r(y, x)。

r 只衡量线性关系。一个很强的非线性关系也可能导致 r 的值很小。

3. Calculating r from Raw Data | 由原始数据计算 r

In exams, you may need to calculate r using the summary statistics provided. The formula below is often given in the formula booklet, but you must know how to apply it correctly.

在考试中，你可能需要利用给出的汇总统计量来计算 r。下面的公式通常出现在公式册中，但你必须知道如何正确使用。

r = (n∑xy − ∑x∑y) / √([n∑x² − (∑x)²][n∑y² − (∑y)²])

where n is the number of data pairs, ∑x is the sum of all x-values, ∑y is the sum of all y-values, ∑xy is the sum of products of paired x and y, ∑x² is the sum of squares of x-values, and ∑y² is the sum of squares of y-values.

其中 n 为数据对的数量，∑x 为所有 x 值之和，∑y 为所有 y 值之和，∑xy 为配对的 x 与 y 乘积之和，∑x² 为 x 值的平方和，∑y² 为 y 值的平方和。

Be careful with the order of operations. Always calculate the components separately and substitute carefully. A common arithmetic error is forgetting to square the sums correctly or mixing up ∑xy with ∑x∑y.

注意运算顺序。务必分别计算各个分量，再小心代入。常见计算错误包括忘记正确计算平方和，或者混淆 ∑xy 与 ∑x∑y。

4. Drawing and Interpreting Scatter Diagrams | 绘制与解读散点图

A scatter diagram is the graphical representation of bivariate data. It is often the first step in judging correlation. Each point on the diagram represents one data pair (x, y).

散点图是双变量数据的图形表示。它通常是判断相关性的第一步。图中的每个点代表一个数据对 (x, y)。

Strong positive correlation: points cluster closely around a rising straight line.
Strong negative correlation: points cluster closely around a falling straight line.
Weak positive correlation: points show an upward trend but are widely scattered.
No correlation: points are scattered randomly with no visible trend.

强正相关：点紧密聚集在一条上升的直线附近。

强负相关：点紧密聚集在一条下降的直线附近。

弱正相关：点呈现上升趋势但分布较为分散。

无相关：点随机分布，没有可见的趋势。

When interpreting a scatter diagram, remember that the pattern must be approximately linear for r to be meaningful. If the points form a curve, such as a parabola, r will underestimate the strength of the relationship.

解读散点图时，必须记住只有当分布模式近似线性时，r 才有意义。如果点形成曲线，例如抛物线，r 会低估关系的强度。

5. Correlation vs. Causation | 相关性与因果关系的区别

This is one of the most important interpretive points in A-Level statistics: correlation does not imply causation. Two variables may be strongly correlated simply by coincidence, or because both are related to a third variable, called a confounding variable or lurking variable.

这是 A-Level 统计学中最重要的解释性要点之一：相关性并不等于因果关系。两个变量可能仅仅因为巧合而高度相关，也可能因为二者都与第三个变量（称为混杂变量或潜在变量）有关而呈现相关。

For example, ice cream sales and drowning incidents may be positively correlated, but eating ice cream does not cause drowning. The lurking variable is hot weather, which increases both ice cream consumption and swimming activity.

例如，冰淇淋销量与溺水事件可能呈正相关，但吃冰淇淋并不会导致溺水。潜在变量是炎热的天气，因为炎热天气既增加了冰淇淋的消费，也增加了游泳活动。

In exam questions, you may be asked to comment on whether a correlation suggests causation. The correct response should always mention that correlation alone is not sufficient evidence for causation, and that a controlled experiment or consideration of confounding variables is needed.

在考试题目中，你可能会被要求评论相关性是否暗示因果关系。正确的回答应始终提及仅凭相关性不足以证明因果关系，还需要受控实验或考虑混杂变量的影响。

6. The Coefficient of Determination r² | 决定系数 r²

Although r measures the strength of a linear relationship, squaring it gives the coefficient of determination, r². This value represents the proportion of the total variation in y that is explained by the linear relationship with x.

虽然 r 衡量线性关系的强度，但对其平方得到决定系数 r²。该值表示 y 的总变异中可以被 x 的线性关系所解释的比例。

r² = explained variation / total variation

For example, if r = 0.8, then r² = 0.64. This means that 64% of the variation in y can be explained by the linear relationship with x; the remaining 36% is due to other factors. If r = 0.5, then r² = 0.25, so only 25% of the variation is explained, leaving 75% unexplained.

例如，若 r = 0.8，则 r² = 0.64。这意味着 y 的变异中有 64% 可以由 x 的线性关系解释；其余 36% 来源于其他因素。如果 r = 0.5，则 r² = 0.25，即只能解释 25% 的变异，剩余 75% 无法解释。

The coefficient of determination is particularly useful when comparing models. A model with r² closer to 1 explains more variability and is generally preferred.

决定系数在比较模型时尤其有用。r² 越接近 1 的模型解释的变异性越多，通常更受青睐。

7. Spearman’s Rank Correlation Coefficient | 斯皮尔曼等级相关系数

In addition to Pearson’s r, A-Level syllabuses often include Spearman’s rank correlation coefficient, denoted by rₛ. It is used when the data are ranks, or when the relationship is monotonic but not necessarily linear. The formula is based on the differences between ranks.

除了皮尔逊相关系数 r，A-Level 大纲通常还包括斯皮尔曼等级相关系数，记作 rₛ。它适用于数据为等级形式，或关系为单调但不一定是线性的情形。该公式基于等级之间的差值。

rₛ = 1 − (6∑d²) / (n(n² − 1))

where d is the difference between the ranks for each pair of observations, and n is the number of pairs.

其中 d 为每个观察对的等级之差，n 为数据对的数量。

Spearman’s rₛ also lies between -1 and 1. It is less sensitive to outliers and can detect monotonic relationships, which makes it a robust alternative when the data are not normally distributed or when extreme values are present.

斯皮尔曼 rₛ 同样在 -1 与 1 之间。它对异常值不那么敏感，并且可以检测单调关系。因此，当数据不服从正态分布或存在极端值时，它是一个稳健的替代选择。

8. Tied Ranks | 相同等级的处理

When two or more observations have the same value, they are said to be tied. In Spearman’s rank calculation, tied ranks are replaced by the average of the ranks they occupy.

当两个或多个观测值相等时，称为并列等级。在斯皮尔曼等级计算中，并列等级用它们所占位置的平均等级代替。

For example, if the second and third values are equal, they both receive the average rank 2.5. The sum of the ranks remains unchanged, but the formula becomes slightly more complicated. In most A-Level exam questions, ties are avoided or easy to handle.

例如，如果第二个和第三个值相等，则它们都取平均等级 2.5。等级总和保持不变，但公式会稍微复杂一些。在大多数 A-Level 考题中，并列情况要么被避免，要么容易处理。

When ties occur, the simplified formula for rₛ is not exact. However, many exam boards accept it as an approximation and mark accordingly. Always check your syllabus for the expected treatment.

当出现并列时，简化公式并非精确值。然而，许多考试局接受其作为近似值并相应给分。请务必查阅你的教学大纲，了解考试局期望的处理方式。

9. Common Mistakes in Interpreting Correlation | 解读相关性时的常见错误

Students often lose marks in this topic by making a few repeated mistakes. Let’s look at them explicitly.

学生在此话题中常因反复出现的一些错误而失分。让我们明确地看一看这些错误。

Confusing strength with slope. A steep regression line does not necessarily indicate a strong correlation. The strength depends on how closely the points fit a line, not on the gradient of the line.
Claiming causation. Stating that x causes y without additional evidence is a serious error.
Ignoring the scatter diagram. A high r value may be misleading if the data contain outliers or a non-linear pattern.
Using r for non-linear relationships. r is only valid for linear relationships; for curved patterns, r can be close to zero even when a strong relationship exists.
Forgetting that r is dimensionless. Never assign units to r.

把强度与斜率混淆。回归线陡峭并不一定表示相关性很强。强度取决于点与直线的拟合程度，而不是直线斜率的大小。

声称因果关系。在没有额外证据的情况下断言 x 导致 y 是一个严重错误。

忽略散点图。如果数据包含异常值或非线性模式，高 r 值可能具有误导性。

对非线性关系使用 r。r 仅适用于线性关系；对于曲线模式，即使存在强关系，r 也可能接近零。

忘记 r 没有单位。绝不要给 r 赋予单位。

10. Correlation and Outliers | 相关性’与异常值

Outliers can have a dramatic effect on Pearson’s correlation coefficient. A single extreme point can inflate or deflate r, depending on its position relative to the rest of the data.

异常值对皮尔逊相关系数的影响可能非常显著。单个极端点可能抬高或压低 r 的值，具体取决于该点相对于其余数据的位置。

Consider a small cluster of points with a weak positive relationship. If one point lies far to the upper right, it can make the correlation appear much stronger than it actually is. Similarly, a point far to the lower right can weaken a positive correlation or even turn it negative.

考虑一小簇呈弱正相关的点。如果一个点位于右上方远处，它可能使相关性看起来比实际强得多。类似地，一个位于右下方远处的点可能削弱正相关，甚至将其变为负相关。

Therefore, whenever you calculate r, always look at the scatter diagram first. If an outlier is present, you should mention its influence and, in some cases, consider removing it, justifying your decision clearly.

因此，每当你计算 r 时，务必先查看散点图。如果存在异常值，你应该提及它的影响，在某些情况下还可以考虑删除它，并清楚说明理由。

11. Extrapolation and Interpolation | 外推与内插

After calculating the regression line, you may be asked to estimate the value of y for a given x. If the x-value lies within the range of the observed data, this is called interpolation, and the estimate is generally reliable. If it lies outside the range, it is called extrapolation, and the estimate is unreliable because the linear relationship may not hold beyond the data range.

在计算回归线之后，你可能会被要求估计给定 x 对应的 y 值。如果 x 值位于观测数据范围内，则称为内插，估计通常较为可靠。如果 x 值位于范围之外，则称为外推，估计不可靠，因为线性关系在数据范围之外可能不再成立。

For example, suppose data are collected for x between 10 and 20. Estimating y at x = 15 is interpolation. Estimating at x = 30 is extrapolation, and the result should be treated with caution.

例如，假设数据收集于 x 在 10 到 20 之间。在 x = 15 处估计 y 属于内插。在 x = 30 处估计则属于外推，结果应谨慎对待。

In exams, you may be asked to comment on the reliability of an estimate. State clearly whether the value is within the data range and avoid making predictions far outside it.

在考试中，你可能被要求评价估计值的可靠性。请明确指出该值是否位于数据范围内，并避免对范围之外的远距离值进行预测。

12. Exam-Style Example and Commentary | 典型例题与解析

Let us work through a short exam-style question to consolidate the ideas.

让我们通过一道简短的例题来巩固以上内容。

Question: A teacher records the number of hours, x, that 8 students spend revising and their exam scores, y. The summary statistics are: n = 8, ∑x = 60, ∑y = 640, ∑x² = 500, ∑y² = 52,000, ∑xy = 5,100. Calculate Pearson’s r and interpret the result.

题目：一位老师记录了 8 名学生复习的小时数 x 和他们的考试成绩 y。汇总统计量为：n = 8，∑x = 60，∑y = 640，∑x² = 500，∑y² = 52,000，∑xy = 5,100。计算皮尔逊相关系数 r，并解释结果。

r = (8 × 5,100 − 60 × 640) / √([8 × 500 − 60²][8 × 52,000 − 640²])

First calculate the numerator: 8 × 5,100 = 40,800; 60 × 640 = 38,400; difference = 2,400.

先计算分子：8 × 5,100 = 40,800；60 × 640 = 38,400；差值为 2,400。

Now calculate the denominator. For x: 8 × 500 = 4,000; 60² = 3,600; difference = 400. For y: 8 × 52,000 = 416,000; 640² = 409,600; difference = 6,400.

再计算分母。对于 x：8 × 500 = 4,000；60² = 3,600；差值为 400。对于 y：8 × 52,000 = 416,000；640² = 409,600；差值为 6,400。

Thus r = 2,400 / √(400 × 6,400) = 2,400 / √(2,560,000) = 2,400 / 1,600 = 1.5. This is impossible! Wait — let’s check the numbers.

因此 r = 2,400 / √(400 × 6,400) = 2,400 / √(2,560,000) = 2,400 / 1,600 = 1.5。这是不可能的！等等——让我们检查一下数据。

Indeed, with these summary values, r exceeds 1, which indicates a calculation or data error. This example highlights the importance of checking plausibility: r must always be between -1 and 1. If you ever obtain a value outside this range, revisit your arithmetic or the original data.

确实，使用这些汇总值时，r 超过了 1，这表明计算或数据存在错误。这个例子强调了检查合理性的重要性：r 必须始终在 -1 与 1 之间。如果你得到超出该范围的值，应重新检查你的运算或原始数据。

Summary | 总结

Judging and interpreting correlation is a core skill in A-Level Mathematics. You must understand the meaning of Pearson’s r and Spearman’s rₛ, know how to calculate them, and, most importantly, interpret them appropriately. Always plot a scatter diagram, check for outliers and non-linear patterns, and remember that correlation does not imply causation.

判断与解释相关性是 A-Level 数学中的核心技能。你必须理解皮尔逊 r 与斯皮尔曼 rₛ 的含义，掌握它们的计算方法，并且最重要的是，能够恰当解释它们。务必绘制散点图，检查异常值和非线性模式，并记住相关性不等于因果关系。

Published by TutorHao | Mathematics Revision Series | aleveler.com

更多咨询请联系16621398022（同微信）

A-Level Mathematics: Correlation — How to Judge and Interpret | A-Level 数学：相关性的判断与解释