IB Mathematics: Hypothesis Testing for a Mean with Unknown Variance | IB数学：未知方差的均值假设检验

📚 IB Mathematics: Hypothesis Testing for a Mean with Unknown Variance | IB数学：未知方差的均值假设检验

In statistics, we often need to test a claim about a population mean (μ) when the population standard deviation (σ) is not known. In this situation, we cannot use the normal distribution-based z-test directly. Instead, we rely on the sample standard deviation (s) and a distribution that accounts for this extra uncertainty: the t-distribution. This article explains the full process of performing a hypothesis test for a mean with unknown variance, as required in IB Mathematics.

在统计学中，我们常常需要在总体标准差（σ）未知的情况下检验关于总体均值（μ）的某个主张。此时我们无法直接使用基于正态分布的 z 检验。我们转而依靠样本标准差（s）和一种能体现这种额外不确定性的分布：t 分布。本文详细解说 IB 数学中在未知方差下进行均值假设检验的完整流程。

1. Why a Different Test? | 为何需要不同的检验？

When the population variance is known, the sample mean x̄ is normally distributed (for large samples or normal populations), and we can standardise it using the population standard deviation to get a z-score. However, when σ is unknown and we estimate it with the sample standard deviation s, the ratio (x̄ − μ) / (s / √n) follows a t-distribution rather than a normal distribution. The t-distribution is similar to the normal but has heavier tails, reflecting the extra variability introduced by estimating σ.

当总体方差已知时，样本均值 x̄ 服从正态分布（对大样本或正态总体），我们可以用总体标准差将其标准化得到 z 分数。然而当 σ 未知，我们用样本标准差 s 去估计它时，比值 (x̄ − μ) / (s / √n) 遵循的是 t 分布而不是正态分布。t 分布与正态分布形状相似，但尾部更厚，体现了估计 σ 所带来的额外变异性。

The t-distribution was developed by William Sealy Gosset under the pseudonym “Student”, hence it is often called Student’s t-distribution. Its shape depends on the degrees of freedom, which are linked to the sample size. With a larger sample, the t-distribution approaches the standard normal distribution.

t 分布由威廉·西利·戈塞特以笔名”Student”提出，因此通常称为 Student t 分布。其形状依赖于自由度，自由度与样本大小有关。随着样本量增大，t 分布逐渐逼近标准正态分布。

2. Setting Up Hypotheses | 建立假设

Every hypothesis test starts with two opposing hypotheses. The null hypothesis, written H₀, always states that the population mean equals a specific value μ₀. For a mean test, H₀: μ = μ₀. The alternative hypothesis, written H₁ or Hₐ, can be one-sided or two-sided. For a two-tailed test, H₁: μ ≠ μ₀; for a one-tailed test, we write H₁: μ > μ₀ or H₁: μ < μ₀ depending on the direction of interest.

每一个假设检验都从两个对立的假设开始。零假设，记作 H₀，总是说明总体均值等于某个特定值 μ₀。对均值检验而言，H₀: μ = μ₀。备择假设，记作 H₁ 或 Hₐ，可以是单侧或双侧。对双侧检验，H₁: μ ≠ μ₀；对单侧检验，我们根据关注方向写成 H₁: μ > μ₀ 或 H₁: μ < μ₀。

The choice of the alternative hypothesis depends on the research question or the claim being tested. It should be set before the data are collected, not after looking at the data. In IB exam questions, the wording often hints at whether the test should be one-tailed or two-tailed, for example, “increased”, “reduced”, or “changed”.

备择假设的选择取决于研究问题或待检验的主张。它应该在收集数据之前设定，而不能在看过数据之后更改。在国际文凭考试题中，措辞通常会提示检验是单侧还是双侧，例如”增加”、”减少”或”改变”。

3. Assumptions and Conditions | 假设与条件

To use the t-test for a population mean, the following conditions must be satisfied:

要使用针对总体均值的 t 检验，必须满足以下条件：

Random sampling: The data must be collected from a random sample, so that it is representative of the population.
随机抽样：数据必须来自随机样本，以使其能代表总体。
Independent observations: Each observation should be independent of the others. In practice, this is satisfied when sampling is done with replacement or when the sample size is less than 10% of a finite population.
观测独立：各观测值应相互独立。实际操作中，当有放回抽样或样本量小于有限总体 10% 时，该条件得到满足。
Normality or large sample size: The population should be approximately normal, especially for small samples. If the sample size is large (often n ≥ 30), the Central Limit Theorem ensures that the sampling distribution of x̄ is approximately normal even if the population is not. For small samples, graphical checks (histogram or normal probability plot) or prior knowledge are used to assess normality.
正态性或大样本：总体应近似正态，尤其在小样本时。若样本量较大（通常 n ≥ 30），中心极限定理保证即使总体非正态，x̄ 的抽样分布也近似正态。对小样本，可用图表（直方图或正态概率图）或先验知识来评估正态性。

4. The Test Statistic | 检验统计量

When the population variance is unknown, the appropriate test statistic is the one-sample t-statistic. It is calculated by taking the difference between the sample mean and the hypothesised population mean, and then dividing by the standard error of the mean.

当总体方差未知时，合适的检验统计量是单样本 t 统计量。它的计算方法是：用样本均值与假设的总体均值之差除以均值的标准误。

t = (x̄ − μ₀) / (s / √n)

Here, x̄ is the sample mean, μ₀ is the mean under the null hypothesis, s is the sample standard deviation (computed with n − 1 in the denominator), and n is the sample size. The quantity s / √n is the standard error of the sample mean, which estimates how far the sample mean is likely to be from the population mean.

这里，x̄ 是样本均值，μ₀ 是零假设下的均值，s 是样本标准差（计算时用 n − 1 作为分母），n 是样本量。量 s / √n 是样本均值的标准误，它估计了样本均值可能偏离总体均值的程度。

The degrees of freedom for this statistic are df = n − 1. This comes from the fact that we have n independent observations but one parameter (the mean) is estimated from the sample, leaving n − 1 independent deviations for estimating the variance.

该统计量的自由度为 df = n − 1。这是因为我们有 n 个独立观测值，但有一个参数（均值）是从样本中估计的，因此剩下 n − 1 个独立残差用于估计方差。

5. The t-Distribution and Degrees of Freedom | t 分布与自由度

The t-distribution is a symmetric, bell-shaped distribution with a peak at zero, but it has heavier tails than the standard normal distribution. The exact shape is governed by the degrees of freedom (df). For a given df, the probability density function describes how likely different t-values are.

t 分布是一种对称的钟形分布，峰值在零处，但其尾部比标准正态分布更重。其具体形状由自由度（df）决定。对给定的 df，其概率密度函数描述了不同 t 值出现的可能性有多大。

df = n − 1: For a one-sample t-test, the degrees of freedom equal the sample size minus one.
df = n − 1：对于单样本 t 检验，自由度等于样本量减一。
Effect of increasing df: As n grows, df increases, and the t-distribution becomes closer to the standard normal distribution N(0, 1). Practically, for n ≥ 30, the critical values of t are very close to those of z.
自由度增大的影响：当 n 增大时，df 增加，t 分布越来越接近标准正态分布 N(0, 1)。实际中，当 n ≥ 30 时，t 的临界值与 z 的临界值非常接近。
Critical values: For a significance level α, the critical region is determined by the (1 − α) quantile or tail probabilities from the t-table. For a two-tailed test with α = 0.05, for example, we look for the value t* such that P(T > t*) = 0.025 and P(T < −t*) = 0.025.
临界值：对于显著性水平 α，临界区域由 t 分布表中的 (1 − α) 分位数或尾部概率决定。例如对 α = 0.05 的双侧检验，我们寻找满足 P(T > t*) = 0.025 和 P(T < −t*) = 0.025 的 t*。

6. Critical Value Approach | 临界值方法

There are two common ways to make a decision in a t-test: the critical value (classical) approach and the p-value approach. In the critical value approach, we compare the calculated test statistic with a critical value from the t-table.

做 t 检验决策有两种常用方法：临界值（经典）方法和 p 值方法。在临界值方法中，我们将计算出的检验统计量与 t 表中查到的临界值进行比较。

The decision rule is:

决策规则如下：

For a two-tailed test (H₁: μ ≠ μ₀): reject H₀ if |t| ≥ t_{α/2, n−1}.
对双侧检验（H₁: μ ≠ μ₀）：若 |t| ≥ t_{α/2, n−1}，则拒绝 H₀。
For a one-tailed test (H₁: μ > μ₀): reject H₀ if t ≥ t_{α, n−1}.
对右侧单侧检验（H₁: μ > μ₀）：若 t ≥ t_{α, n−1}，则拒绝 H₀。
For a one-tailed test (H₁: μ < μ₀): reject H₀ if t ≤ −t_{α, n−1}.
对左侧单侧检验（H₁: μ < μ₀）：若 t ≤ −t_{α, n−1}，则拒绝 H₀。

Here t_{α, n−1} represents the critical value such that the area to its right under the t-distribution with n − 1 degrees of freedom equals α. Remember to draw a t-distribution diagram and shade the rejection region; this helps avoid sign errors.

这里 t_{α, n−1} 表示在自由度为 n − 1 的 t 分布下，右侧面积为 α 的临界值。记得画出 t 分布图并标出拒绝区域，这样有助于避免符号错误。

7. p-Value Approach | p 值方法

The p-value is the probability of obtaining a test statistic at least as extreme as the one actually observed, assuming that the null hypothesis is true. A small p-value indicates that the observed data are unlikely under H₀, which leads us to reject the null hypothesis.

p 值是在零假设成立的前提下，获得一个至少与观测到的检验统计量一样极端的检验统计量的概率。p 值越小，说明观测数据在 H₀ 下越不可能出现，从而促使我们拒绝零假设。

Using a graphical display calculator (GDC), we can directly find the p-value for a t-test. For example, on most IB-approved calculators, selecting a “t-test” option yields both the t-statistic and the p-value. Alternatively, if you only have access to a t-table, you can bound the p-value between two tabulated probabilities.

计算可以用图形计算器（GDC）直接得出 t 检验的 p 值。例如，在大多数国际文凭认可的计算机上，选择”t 检验”选项会同时给出 t 统计量和 p 值。如果只有 t 表，你也可以把 p 值确定在两个表列概率之间。

The decision rule is simple: compare the p-value with the significance level α. If p-value ≤ α, reject H₀; if p-value > α, do not reject H₀. In the IB examination, reporting both the test statistic and the p-value is essential for full marks, along with a clear conclusion in the context of the question.

决策规则很简单：将 p 值与显著性水平 α 比较。若 p 值 ≤ α，则拒绝 H₀；若 p 值 > α，则不拒绝 H₀。在国际文凭考试中，为了获得满分，必须同时报告检验统计量和 p 值，并在问题的语境中给出清晰结论。

8. Worked Example | 实例演算

A company claims that its bags of coffee weigh an average of 500 g. A consumer group doubts this and suspects the average weight is less. They take a random sample of 10 bags and find a sample mean of 492 g and a sample standard deviation of 15 g. Test at the 5% significance level whether the true mean weight is less than 500 g.

某公司声称其咖啡袋平均重量为 500 克。一个消费者组织对此表示怀疑，认为平均重量偏低。他们随机抽取 10 袋，测得样本均值为 492 克，样本标准差为 15 克。在 5% 显著性水平下检验真实平均重量是否小于 500 克。

Step 1: State the hypotheses. Since the suspicion is “less than”, we use a one-tailed test.

第一步：陈述假设。由于怀疑”小于”，我们使用单侧检验。

H₀: μ = 500, H₁: μ < 500

Step 2: Identify the significance level. α = 0.05.

第二步：确定显著性水平。α = 0.05。

Step 3: Calculate the test statistic. Here n = 10, x̄ = 492, s = 15, μ₀ = 500.

第三步：计算检验统计量。这里 n = 10，x̄ = 492，s = 15，μ₀ = 500。

t = (492 − 500) / (15 / √10) = −8 / 4.743 = −1.686

Step 4: Find the critical value or p-value. Degrees of freedom = 10 − 1 = 9. For a one-tailed test with α = 0.05 and df = 9, the critical value is t* = −1.833. The t-statistic −1.686 is greater than −1.833, so it does not fall in the critical region. Alternatively, the p-value is about 0.063, which is greater than 0.05.

第四步：查找临界值或 p 值。自由度 = 10 − 1 = 9。对 α = 0.05、df = 9 的单侧检验，临界值为 t* = −1.833。统计量 −1.686 大于 −1.833，因此不在拒绝区域内。或者，p 值约为 0.063，大于 0.05。

Step 5: Draw a conclusion. Since the p-value is greater than α (or the test statistic is not in the critical region), we do not reject H₀. There is insufficient evidence to support the claim that the true mean weight is less than 500 g.

第五步：得出结论。由于 p 值大于 α（或检验统计量不在拒绝区域），我们不拒绝 H₀。没有足够证据支持真实平均重量小于 500 克的主张。

9. Common Errors and Exam Advice | 常见错误与考试建议

Hypothesis testing is a rich source of marks, but students often lose marks on small details. Below are common pitfalls and how to avoid them.

假设检验是得分点很多的板块，但学生常常因细节失分。以下是常见陷阱及避免方法。

Confusion between one-tailed and two-tailed tests: Read the wording carefully. “Greater”, “less”, “increased”, “decreased” indicate one-tailed; “different”, “changed”, “not equal” indicate two-tailed.
单侧与双侧检验混淆：仔细阅读措辞。”大于”、”小于”、”增加”、”减少”表示单侧；”不同”、”改变”、”不等于”表示双侧。
Using σ instead of s: If the population variance is unknown, you must use the sample standard deviation s and the t-distribution. Only if σ is given can you use the z-test.
误用 σ 代替 s：若总体方差未知，必须使用样本标准差 s 和 t 分布。只有给出 σ 时才能使用 z 检验。
Incorrect degrees of freedom: Always use df = n − 1, not n. This is a common error, especially under exam pressure.
自由度错误：始终使用 df = n − 1 而不是 n。这是常见的错误，尤其在考试压力下。
Forgetting to state the conclusion in context: The final conclusion should mention the variable and the claimed mean, not just “reject H₀” or “do not reject H₀”. For example: “There is insufficient evidence to conclude that the average bag weight is less than 500 g.”
忘记结合语境写结论：最终结论应提及变量和主张的均值，而不只是”拒绝 H₀”或”不拒绝 H₀”。例如：”没有足够证据说明咖啡袋平均重量小于 500 克。”
Mixing up p-value and α: The p-value is a calculated probability; α is the pre-chosen significance level. Reject H₀ only if p-value ≤ α.
混淆 p 值和 α：p 值是计算得到的概率；α 是事先设定的显著性水平。只有在 p 值 ≤ α 时才拒绝 H₀。

In the IB exam, present your work in clear steps: hypotheses, significance level, test statistic, p-value or critical value, and a written conclusion. Show the formula you used, and round appropriately, usually to three decimal places for final answers unless otherwise instructed.

在国际文凭考试中，请分步清晰作答：假设、显著性水平、检验统计量、p 值或临界值，以及文字结论。写出所用公式，最终答案通常四舍五入保留三位小数（除非另有要求）。

10. Summary and Connections | 总结与联系

To summarise, a hypothesis test for a mean with unknown variance is performed using a one-sample t-test. The test statistic is t = (x̄ − μ₀) / (s / √n), with n − 1 degrees of freedom. We compare this value with a critical value from the t-distribution or assess its p-value against the significance level α. The t-distribution accounts for the extra uncertainty caused by estimating the population standard deviation.

总结一下，未知方差的均值假设检验使用单样本 t 检验。检验统计量为 t = (x̄ − μ₀) / (s / √n)，自由度为 n − 1。我们将其与 t 分布表中的临界值比较，或用 p 值与显著性水平 α 对比。t 分布考虑了估计总体标准差所带来的额外不确定性。

This test is closely related to other procedures you will encounter in IB Mathematics. When the sample size is large or σ is known, the test becomes a z-test, using the normal distribution. The t-test also appears in comparing two means, in the form of the matched-pairs t-test and the two-sample t-test. Understanding the one-sample t-test lays a solid foundation for these further topics.

该检验与 IB 数学中你还会遇到的其他方法紧密相连。当样本量较大或 σ 已知时，检验成为使用正态分布的 z 检验。t 检验也出现在比较两个均值的情形中，例如配对 t 检验和双样本 t 检验。理解单样本 t 检验为这些后续主题奠定了坚实基础。

Above all, remember that hypothesis testing is a decision-making tool under uncertainty. It does not prove that a hypothesis is true or false; instead, it measures the strength of evidence against the null hypothesis. A significance level of 0.05 means that if H₀ is actually true, we would reject it about 5% of the time. Always interpret your results with this probabilistic meaning in mind.

最重要的是，假设检验是在不确定性下做决策的工具。它并不证明假设真伪；而是衡量反对零假设的证据强度。显著性水平 0.05 意味着如果 H₀ 确实为真，我们大约有 5% 的次数会拒绝它。解读结果时，务必牢记这种概率意义。

Published by TutorHao | Mathematics Revision Series | aleveler.com

更多咨询请联系16621398022（同微信）

IB Mathematics: Hypothesis Testing for a Mean with Unknown Variance | IB数学：未知方差的均值假设检验