📚 Percentage points of the F-distribution | F分布的百分位点
The F-distribution is a fundamental probability distribution in A-Level Statistics, especially when comparing two population variances using an F-test. Understanding its percentage points – more commonly called critical values – is essential for hypothesis testing. This article explains how to read and apply these percentage points from standard F-tables, covering upper and lower tails, relationship formulas, and practical techniques such as interpolation. By the end, you will be confident in tackling any Edexcel question involving the F-distribution.
F分布是A-Level统计中一种基础的概率分布，尤其在利用F检验比较两个总体方差时至关重要。理解它的百分位点——通常被称为临界值——是进行假设检验的核心。本文将讲解如何从标准F分布表中读取并应用这些百分位点，涵盖上尾、下尾、换位公式以及插值等实用技巧。学完本文，你将能自信地解答Edexcel考试中涉及F分布的任何题目。
1. What is the F-distribution? | 什么是F分布？
two independent chi-squared random variables, each divided by its degrees of freedom. If U ~ χ²(v₁) and V ~ ws an F-distribution with v₁ numerator degrees of freedom and v₂ denominator degrees of freedom, written as F(v₁, v₂). It is the theoretical model for the ratio of two sample variances, making it central to variance comparison tests.
当我们考虑两个独立的卡方随机变量的比值（各自除以其自由度）时，就得到了F分布。若U ~ χ²(v₁) 且 V ~ χ²(v₂) 相互独立，则随机变量F = (U/v₁) / (V/v₂) 服从自由度为 v₁（分子）和 v₂（分母）的F分布，记作F(v₁, v₂)。它是两个样本方差之比的理论模型，因此是方差比较检验的核心工具。
2. Shape and properties | 形状与性质
The F-distribution is continuous, non-negative, and right-skewed. Its shape depends on the two degrees of freedom parameters. For small v₁ and v₂ the skew is pronounced, but as both degrees of freedom increase, the distribution becomes more symmetric and approaches a normal shape. The total area under any F-curve equals 1, and the distribution has a mean of v₂/(v₂ - 2) for v₂ > 2. All probability is concentrated on positive real numbers.
F分布是连续、非负且右偏的。其形状由两个自由度参数决定：当v₁和v₂都较小时，偏态非常明显；但随着两个自由度的增大，分布趋于对称并向正态形式靠拢。任何F曲线下的总面积均为1，分布的均值在v₂ > 2时为 v₂/(v₂ - 2)。所有概率都集中在正实数上。
3. Degrees of freedom explained | 自由度的含义
zes of the samples whose variances are being compared. The first parameter, v₁, is associated with the numerator variance estimate, and the second, v₂, with the denominator. Their order is crucial because the F-distribution is not symmetric in its degrees of freedom.
自由度直接来源于方差检验中的样本容量。在双样本F检验中，v₁ = n₁ - 1，v₂ = n₂ - 1，其中n₁和n₂分别是比较方差的两个样本的容量。第一个参数v₁对应分子方差估计，第二个v₂对应分母。它们的顺序至关重要，因为F分布关于自由度并不对称。
4. Definition of percentage points | 百分位点的定义
A percentage point of the F-distribution is a value Fv₁,v₂,α such that the probability of the random variable exceeding that value is exactly α. In symbols, P(F(v₁, v₂) > Fv₁,v₂,α e 0.05 or 0.01, corresponding to the significance level of a test. These points are also called upper critical values. Because α is the area in the right tail, the percentage is often expressed as 100(1 - α)%, so a 5% upper point corresponds to the 95th percentile.
F分布的百分位点是一个数值 Fv₁,v₂,α，该随机变量超过它的概率恰好为α。用符号表示即 P(F(v₁, v₂) > Fv₁,v₂,α) = α。这里的α是一个较小的概率，如0.05或0.01，对应检验的显著性水平。这些点也称为上临界值。因为α是右尾的面积，百分位通常以100(1 - α)%的形式表述，因此5%的上侧分位数对应的是第95百分位数。
5. Structure of F-distribution tables | F分布表的结构
nized with v₁ (numerator degrees of freedom) as columns and v₂ (denominator degrees of freedom) as rows. To find the 5% r F(4, 10), locate the column for v₁ = 4 and the row for v₂ = 10; the intersection gives F4,10,0.05 = 3.48 (example value). Always check the correct table for the required α.
Edexcel提供的F分布表通常覆盖常用的上尾概率，一般为 α = 0.05、0.025、0.01 和 0.005。表格以v₁（分子自由度）为列、v₂（分母自由度）为行。要查找 F(4, 10) 的5%临界值，先找到 v₁ = 4 的列和 v₂ = 10 的行，交叉处即为 F4,10,0.05 = 3.48（示例值）。务必核对所查表格对应的 α 值。
6. Upper-tail percentage points | 上尾百分位点
ther. For example, if H₁: σ₁² > σ₂², then the test statistic F = s₁²/s₂² is compared with Fv₁,v₂,α. If the calculated F exceeds this tabulated value, we reject H₀ at significance level α. The table only provides upper-tail points, which is why understanding how to obtain lower-tail values is essential.
上尾百分位点直接用于单尾检验，当备择假设表明一个方差大于另一个时。例如，若 H₁: σ₁² > σ₂²，检验统计量 F = s₁²/s₂² 将与 Fv₁,v₂,α 比较。若计算出的F值大于表中查得的数值，则在显著性水平 α 下拒绝H₀。F表通常只提供上尾分位数，因此掌握如何获取下尾值至关重要。
7. Lower-tail points and the reciprocal relation | 下尾点与倒数关系
Lower-tail percentage points are not tabulated directly, but they can be obtained from a simple relationship: Fv₁,v₂,1-α = 1 / Fv₂,v₁,α. This means the lower 5% point (α = 0.05 in the left tail) equals the reciprocal of the upper 5% point with the degrees of freedom swapped. For instance, the lower 5% critical value for F(4, 10) is 1 / F10,4,0.05. This formula is a direct consequence of the definition of the F random variable as a ratio.
下尾百分位点并不直接列表给出，但可以通过一个简单关系求得：Fv₁,v₂,1-α = 1 / Fv₂,v₁,α。这意味着左侧5%的分位数（α = 0.05）等于分子分母自由度互换后的上侧5%分位数的倒数。例如，F(4, 10) 的下5%临界值为 1 / F10,4,0.05。这个公式直接来源于F随机变量的比值定义。
8. Applying percentage points in two-tailed tests | 双尾检验中的应用
². Both very small and very large values of the ratio s₁²/s₂² would cast doubt on H₀. The critical region uses two percentage points: the upper Fv₁,v₂,α/2 and the lower Fv₁,v₂,1-α/2. The lower point is found via the reciprocal relation as 1 / Fv₂,v₁,α/2. If the test statistic falls below the lower point or above the upper point, H₀ is rejected.
在双尾F检验中，假设为 H₀: σ₁² = σ₂²，备择假设 H₁: σ₁² ≠ σ₂²。比值 s₁²/s₂² 无论过小还是过大都将对H₀产生怀疑。拒绝域使用两个百分位点：上侧 Fv₁,v₂,α/2 和下侧 Fv₁,v₂,1-α/2。下侧点通过倒数关系求得，即 1 / Fv₂,v₁,α/2。若检验统计量落在下侧点以下或上侧点以上，则拒绝H₀。
9. Interpolating between tabulated values | 在表值之间插值
lues for v₂ = 30 and v₂ = 40, you can interpolate: F(v₁, 33) ≈ F(v₁, 30) + (3/10)[F(v₁, 40) - F(v₁, 30)]. While not perfectly linear, this approximation is acceptable in Edexcel exams unless stated otherwise. Ensure you state the formula used and keep calculations precise to 3 or 4 decimal places.
通常，表中并未列出准确自由度值，这时需要线性插值。例如，需要 v₂ = 33 的值而表中只有 v₂ = 30 和 v₂ = 40 时，可按如下插值：F(v₁, 33) ≈ F(v₁, 30) + (3/10)[F(v₁, 40) - F(v₁, 30)]。虽然F分布并非完全线性，但在Edexcel考试中，除非另有说明，这种近似均被接受。请写明所用公式并保持小数点后3至4位的精度。
10. Worked example: testing equality of variances | 例题：检验方差齐性
₁² > σ₂². Here v₁ = 14, v₂ = 19. The test statistic F = 12.5/5.8 ≈ 2.155. From the upper 5% table, F14,19,0.05 needed). Since 2.155 < 2.26, we do not reject H₀; there is insufficient evidence at the 5% arger. This classic application illustrates how percentage points are used in decision making.
假设两个样本的方差分别为 s₁² = 12.5（n₁ = 15）和 s₂² = 5.8（n₂ = 20）。在 α = 0.05 下检验 σ₁² > σ₂² 是否成立。此处 v₁ = 14，v₂ = 19。检验统计量 F = 12.5/5.8 ≈ 2.155。查上5%表，F14,19,0.05 ≈ 2.26（必要时使用插值）。由于 2.155 < 2.26，我们不拒绝H₀；在5%显著性水平下，没有足够证据表明第一个方差更大。这个经典应用展示了百分位点如何用于决策。
11. Common pitfalls and exam tips | 常见误区与应考提示
Beware of mixing up the order of degrees of freedom; the critical value for F(5, 8) is not the same as for F(8, 5). Always label v₁ and v₂ clearly and verify the correct table. For lower-tail values, do not forget to swap the degrees of freedom before taking the reciprocal. When using the reciprocal relation, write down the full expression to avoid arithmetic mistakes. In two-tailed tests, remember to halve the significance level. Finally, show all working with table lookups and interpolation steps to secure full method marks.
务必注意不要混淆自由度的顺序；F(5, 8) 的临界值与 F(8, 5) 不同。始终清晰标注 v₁ 和 v₂，并核对正确的表。求下尾值时，不要忘记先互换自由度再取倒数。使用倒数关系时，写出完整表达式以避免计算错误。在双尾检验中，记住将显著性水平减半。最后，展示查表和插值的所有步骤，以便获得完整的过程分。
12. Summary | 总结
The percentage points of the F-distribution are essential tools for testing hypotheses about variances. Mastery lies in confidently reading tables for upper-tail values, applying the reciprocal rule for lower-tail values, and correctly handling one-tailed and two-tailed tests. Keep a tidy approach to interpolation and always double-check your degrees of freedom. With practice, these critical values become a natural part of your statistical toolkit for Edexcel A-Level Mathematics.
F分布的百分位点是进行方差假设检验的必备工具。掌握它们的关键在于：自信地查表读取上尾值，用倒数规则求取下尾值，并正确处理单尾与双尾检验。保持插值过程的整洁，并反复核验自由度。通过练习，这些临界值将成为你应对Edexcel A-Level数学统计问题的自然利器。
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