📚 IB Mathematics: Fundamentals of Hypothesis Testing | IB数学：假设检验的基本原理
Hypothesis testing is one of the most powerful and frequently examined topics in the IB Mathematics curriculum. It provides a formal framework for making decisions about a population based on sample data, and it bridges the gap between probability theory and real-world statistical inference.
假设检验是IB数学课程中最强大且最常被考查的主题之一。它提供了一个基于样本数据对总体作出决策的正式框架，将概率论与现实世界的统计推断紧密联系起来。
In this article, you will learn the fundamental concepts that underpin every hypothesis test: null and alternative hypotheses, test statistics, p-values, significance levels, errors, critical regions, and the power of a test. We will also show you how to apply these ideas step by step to IB-style questions.
本文将带你学习支撑每一次假设检验的基本概念：原假设与备择假设、检验统计量、p值、显著性水平、错误类型、拒绝域以及检验的功效。我们还将逐步展示如何将这些思想应用到IB风格的题目中。
1. What Is Hypothesis Testing? | 什么是假设检验？
Hypothesis testing is a procedure in statistics that uses sample data to evaluate a claim or hypothesis about a population parameter. The goal is not to prove a hypothesis but to decide whether the data provide enough evidence to reject the claim.
假设检验是统计学中利用样本数据来评估关于总体参数的声明或假设的过程。其目标不是“证明”某个假设，而是判断数据是否提供了足够的证据来拒绝这个声明。
In IB Mathematics, you will meet hypothesis tests for the population mean μ, the population proportion p, and the correlation coefficient ρ. In each case, the logic is identical: assume a default position, collect data, and decide whether the data contradict that default position strongly enough.
在IB数学中，你会遇到关于总体均值 μ、总体比例 p 和相关系数 ρ 的假设检验。每种情况下的逻辑都是一致的：先假设一个默认立场，收集数据，然后判断数据是否足够强烈地反驳该默认立场。
Think of a criminal trial: the defendant is presumed innocent until evidence proves otherwise. Similarly, in hypothesis testing we assume the null hypothesis is true until sample evidence suggests otherwise. This asymmetry is central to statistical thinking.
可以把假设检验比作刑事审判：在证据确凿之前，被告被假定为无罪。类似地，在假设检验中，我们假定原假设为真，直到样本证据表明并非如此。这种不对称性是统计思维的核心。
2. Null and Alternative Hypotheses | 原假设与备择假设
Every hypothesis test begins with two competing statements about the population parameter. These are called the null hypothesis, written H₀, and the alternative hypothesis, written H₁ or Hₐ.
每一次假设检验都始于关于总体参数的两个相互对立的陈述。它们分别被称为原假设，记作 H₀，以及备择假设，记作 H₁ 或 Hₐ。
s is the claim we want to find evidence for, and it can be two-sided (≠) or one-sided (> or <).
原假设是被检验的陈述，它总是包含等号：=、≤ 或 ≥。例如，H₀ : μ = 10。备择假设是我们希望找到证据支持的声明，它可以是双尾的（≠）或单尾的（> 或 <）。
H₀ : μ = μ₀ versus H₁ : μ ≠ μ₀
It is essential to set up the hypotheses carefully before performing any calculation. In IB exams, you may be asked to state the null and alternative hypotheses for a given context, so make sure your notation matches the parameter in question.
在进行任何计算之前，严谨地写出假设至关重要。在IB考试中，你可能会被要求根据给定情境写出原假设和备择假设，所以务必确保你的符号与所讨论的参数一致。
ll hypothesis would then be H₀ : μ ≤ μ₀ or simply H₀ : μ = μ₀, depending on the test used.
注意，备择假设才是研究问题。如果一位生物专业学生声称一种新药会延长反应时间，那么备择假设应为 H₁ : μ > μ₀。此时原假设可以是 H₀ : μ ≤ μ₀，或者简单写为 H₀ : μ = μ₀，这取决于所用的检验方法。
3. Test Statistic and Sampling Distribution | 检验统计量与抽样分布
Once the hypotheses are stated, we calculate a test statistic from the sample data. The test statistic is a numerical value that measures how far the sample result is from the null hypothesis value, measured in units of standard error.
假设写出之后，我们从样本数据中计算一个检验统计量。检验统计量是一个数值，它衡量样本结果与零假设值之间的距离，以标准误为单位。
For a population mean with known variance, the test statistic is the z-score:
对于方差已知的总体均值，检验统计量是z分数：
z = (x̄ − μ₀) / (σ / √n)
When the population variance is unknown, we use the sample standard deviation s and the t-statistic:
当总体方差未知时，我们使用样本标准差 s 和t统计量：
t = (x̄ − μ₀) / (s / √n), df = n − 1
Each test statistic follows a known sampling distribution under H₀. For example, z follows the standard normal distribution, whereas t follows a t-distribution with n − 1 degrees of freedom. This distribution allows us to compute probabilities associated with the observed result.
每一个检验统计量在原假设 H₀ 下都服从已知的抽样分布。例如，z服从标准正态分布，而t服从自由度为 n − 1 的t分布。这一分布使我们能够计算与观察结果相关的概率。
In IB, you are expected to use your GDC (graphical display calculator) to compute test statistics and associated probabilities. However, you must understand which formula your calculator is using and how the degrees of freedom affect the p-value.
在IB中，你被期望利用图形计算器（GDC）来计算检验统计量和相关概率。然而，你必须理解计算器使用的是哪个公式，以及自由度如何影响p值。
4. Significance Level and p-value | 显著性水平与p值
before the data are collected. Common choices in IB are α = 0.05, α = 0.01, and α = 0.10.
显著性水平，记作 α，是当原假设实际为真时拒绝原假设的概率。它由研究者在收集数据之前确定。IB中常见的选择有 α = 0.05、α = 0.01 和 α = 0.10。
The p-value is the probability of obtaining a test statistic at least as extreme as the one observed, assuming that H₀ is true. It is not the probability that H₀ is true. This distinction is subtle but vital for correct interpretation.
p值是在假定 H₀ 为真的条件下，得到至少与观测值一样极端的检验统计量的概率。它并不是 H₀ 为真的概率。这一区别虽然微妙，但对正确解读至关重要。
If p-value ≤ α → reject H₀ | If p-value > α → fail to reject H₀
A small p-value indicates that the observed data would be very unlikely if H₀ were true. This provides evidence against H₀. Conversely, a large p-value means the data are consistent with H₀, so we do not reject it.
较小的p值表明，如果 H₀ 为真，观测到的数据将是非常不可能的。这为反对 H₀ 提供了证据。反之，较大的p值意味着数据与 H₀ 相符，因此我们不拒绝它。
Many IB students write "accept H₀" when the p-value is large. This is incorrect. We can only say "fail to reject H₀" because a lack of evidence against H₀ does not prove that H₀ is true.
许多IB学生在p值较大时写“接受 H₀”，这是不正确的。我们只能说“不能拒绝 H₀”，因为缺乏反对 H₀ 的证据并不等于证明 H₀ 为真。
5. One-tailed vs Two-tailed Tests | 单尾检验与双尾检验
The alternative hypothesis determines whether a test is one-tailed or two-tailed. If H₁ uses ≠, the test is two-tailed, because extreme values on either side of the null value provide evidence against H₀.
备择假设决定了检验是单尾还是双尾。如果 H₁ 使用 ≠，则该检验是双尾的，因为位于原假设取值两侧的极端值都可以提供反对 H₀ 的证据。
If H₁ uses >, the test is a right-tailed test, because only unusually large sample values can support the alternative. If H₁ uses <, the test is a left-tailed test, because only unusually small values can support it.
如果 H₁ 使用 >，则该检验是右尾检验，因为只有异常大的样本值才能支持备择假设。如果 H₁ 使用 <，则该检验是左尾检验，因为只有异常小的值才能支持它。
| Alternative Hypothesis | Type of Test |
| H₁ : μ ≠ μ₀ | Two-tailed test |
| H₁ : μ > μ₀ | Right-tailed test |
| H₁ : μ < μ₀ | Left-tailed test |
For a two-tailed test, the p-value is the sum of the probabilities in both tails beyond the absolute value of the test statistic. Most GDCs compute this automatically, but you should understand why the p-value doubles when using a standard normal table.
对于双尾检验，p值是检验统计量绝对值之外两个尾部的概率之和。大多数计算器会自动计算，但你应该理解为什么使用标准正态表时p值需要乘以2。
values are z = −z(α/2) and z = z(α/2). For a one-tailed test, the critical value is z(α).
关键在于，决策规则取决于尾部的结构。对于显著性水平为 α 的双尾检验，临界值为 z = −z(α/2) 和 z = z(α/2)。对于单尾检验，临界值为 z(α)。
6. Critical Regions and Critical Values | 拒绝域与临界值
The critical region, also called the rejection region, is the set of values of the test statistic that lead to rejection of H₀. Its boundary is called the critical value. The critical region is determined by the significance level α.
临界区域，也称为拒绝域，是导致拒绝 H₀ 的检验统计量的取值集合。其边界称为临界值。临界区域由显著性水平 α 决定。
stic is greater than 1.645, we reject H₀. If it is less than or equal to 1.645, we fail to reject H₀.
例如，在 α = 0.05 的右尾z检验中，临界值为 z = 1.645。如果计算出的检验统计量大于1.645，则拒绝 H₀。如果小于或等于1.645，则不拒绝 H₀。
When working with your GDC, you may not need to compute critical values directly because the p-value method is usually preferred. However, IB questions sometimes ask you to find the critical value or to state whether a test statistic lies in the critical region.
使用计算器时，你可能不需要直接计算临界值，因为通常更推荐使用p值方法。然而，IB题目有时要求你求临界值，或判断检验统计量是否落在临界区域内。
The critical value divides the sample space into two regions: the rejection region and the non-rejection region. The interpretation of the test result depends on which region contains the observed test statistic.
临界值将样本空间划分为两个区域：拒绝区域和非拒绝区域。检验结果的解释取决于观测到的检验统计量落在哪个区域内。
7. Type I and Type II Errors | 第一类错误与第二类错误
No hypothesis test is perfect. Two possible mistakes can be made when drawing a conclusion. These are called Type I and Type II errors.
没有任何假设检验是完美的。在得出结论时可能犯两种错误，分别称为第一类错误和第二类错误。
A Type I error occurs when we reject H₀ even though H₀ is true. The probability of a Type I error is exactly the significance level α, which the researcher controls in advance.
第一类错误发生在 H₀ 实际为真时我们却拒绝了 H₀。犯第一类错误的概率恰好等于显著性水平 α，这是研究者事先控制的。
A Type II error occurs when we fail to reject H₀ even though H₀ is false. The probability of a Type II error is denoted by β. Unlike α, β depends on the true value of the parameter and on the sample size.
第二类错误发生在 H₀ 实际为假时我们却不拒绝 H₀。犯第二类错误的概率记作 β。与 α 不同，β 取决于参数的真实值和样本容量。
| Reject H₀ | Fail to reject H₀ | |
| H₀ is true | Type I error (probability α) | Correct decision |
| H₀ is false | Correct decision | Type II error (probability β) |
In IB exams, a common question asks you to interpret a Type I or Type II error in context. For example, if H₀ states that the mean concentration is safe, rejecting H₀ when it is actually safe would be a Type I error: declaring the water unsafe when it is safe.
在IB考试中，常见的问题要求你结合情境解释第一类或第二类错误。例如，如果 H₀ 表示平均浓度是安全的，当它实际上安全时却拒绝 H₀ 就是第一类错误：在安全的情况下宣布水质不安全。
Reducing α typically increases β, and reducing β typically requires a larger sample size. This trade-off is a fundamental feature of statistical inference.
降低 α 通常会增大 β，而降低 β 通常需要更大的样本量。这种权衡是统计推断的基本特征。
8. Power of a Test | 检验的功效
alse. In symbols, Power = 1 − β. A high-power test is more likely to detect a real effect.
假设检验的功效是当 H₀ 为假时检验正确拒绝 H₀ 的概率。用符号表示，功效 = 1 − β。高功效的检验更有可能检测到真实效应。
Power depends on several factors: the true value of the parameter, the sample size n, the significance level α, and the variability of the data. In general, increasing the sample size increases the power of a test.
功效取决于多个因素：参数的真实值、样本量 n、显著性水平 α 以及数据的变异性。一般来说，增大样本量会提高检验的功效。
Power = P(reject H₀ | H₀ is false)
For IB students, power is more conceptual than computational. You should be able to explain why a larger sample size improves power and why a smaller significance level reduces power, other things being equal.
对于IB学生，功效更多是概念性的而非计算性的。你应该能够解释为什么在其它条件相同的情况下，更大的样本量能提高功效，而更小的显著性水平会降低功效。
In practice, power analysis helps researchers design experiments with enough observations to have a reasonable chance of detecting meaningful differences. This idea connects hypothesis testing with experimental design.
在实践中，功效分析帮助研究者设计具有足够观测值的实验，以便有合理的机会检测到有意义的差异。这一思想将假设检验与实验设计联系起来。
9. Step-by-Step Procedure for Hypothesis Testing | 假设检验的标准步骤
IB questions often ask you to perform a full hypothesis test. It is helpful to follow a consistent, clearly written sequence of steps so that examiners can award method marks even if a numerical error occurs.
IB题目通常要求你完成一次完整的假设检验。遵循一个一致、书写清晰的步骤序列是很有帮助的，这样即使出现数值错误，考官也能给出方法分。
Step 1: Write down the null and alternative hypotheses in terms of the population parameter.
第一步：用总体参数写出原假设和备择假设。
Step 2: Choose the appropriate test statistic and check any conditions required, such as a normal approximation or known variance.
第二步：选择合适的检验统计量，并检查所要求的条件，例如正态近似或已知方差。
Step 3: Use your GDC to calculate the test statistic and the p-value.
第三步：使用图形计算器计算检验统计量和p值。
Step 4: Compare the p-value with the significance level α, or compare the test statistic with the critical value.
第四步：将p值与显著性水平 α 比较，或将检验统计量与临界值比较。
Step 5: Write a conclusion in the context of the problem. Avoid vague statements like "reject H₀" without explaining what H₀ means in the original situation.
第五步：结合具体问题写出结论。避免写出诸如“拒绝 H₀”这样模糊的表述，而不解释 H₀ 在原始情境中的含义。
For example: "There is sufficient evidence at the 5% significance level to conclude that the mean waiting time has increased." This is the type of conclusion that earns full marks in IB exams.
例如：“在5%显著性水平下，有足够证据得出结论：平均等待时间增加了。”这就是在IB考试中获得满分的结论写法。
10. Hypothesis Tests in the IB Syllabus | IB考纲中的假设检验
In IB Mathematics: Analysis and Approaches (AA), hypothesis testing appears at Higher Level and includes testing for a population mean using the t-distribution and for a proportion using the normal approximation.
在IB数学：分析与方法（AA）中，假设检验出现在高级水平（HL），包括使用t分布检验总体均值以及使用正态近似检验总体比例。
In IB Mathematics: Applications and Interpretation (AI), hypothesis testing is also part of Higher Level, with a stronger focus on interpreting p-values and making decisions in real-world contexts.
在IB数学：应用与解释（AI）中，假设检验也是高级水平（HL）的一部分，更侧重于解释p值并在现实情境中作出决策。
Both courses require students to use technology efficiently. You should be familiar with the following calculator functions: one-sample t-test, z-test, and one-proportion z-test. Practice entering sample data or summary statistics correctly.
两门课程都要求学生高效地使用技术工具。你应该熟悉以下计算器功能：单样本t检验、z检验以及单比例z检验。练习正确输入原始数据或汇总统计量。
Common IB questions may also involve probability questions connected to errors, such as finding P(Type I error) or P(Type II error) for a simple binomial test. Make sure you understand how to set these up from first principles.
IB中常见的问题还可能涉及与错误相关的概率题，例如求简单二项检验中 P(第一类错误) 或 P(第二类错误)。请务必理解如何从基本原理出发进行求解。
11. Common Pitfalls and Exam Tips | 常见误区与考试建议
One frequent mistake is swapping the hypotheses. Remember that the null hypothesis always contains the equality, and the alternative hypothesis reflects the claim being tested. Write H₀ and H₁ before doing any calculation.
一个常见错误是写反假设。记住原假设总是包含等号，而备择假设反映被检验的声明。在进行任何计算之前先写出 H₀ 和 H₁。
df = n − 1. Using the wrong df can change the p-value and lead to an incorrect conclusion.
另一个错误是在t检验中使用错误的自由度。对于单样本t检验，df = n − 1。使用错误的自由度会改变p值，并导致错误的结论。
Students also often misinterpret the p-value as the probability that the null hypothesis is true. The p-value is computed under the assumption that H₀ is true; it is not P(H₀ | data). In IB Paper 2, such conceptual questions may be included.
学生也经常将p值误解为原假设为真的概率。p值是在假定 H₀ 为真的条件下计算的；它不是 P(H₀ | 数据)。在IB Paper 2中，可能会出现这样的概念性问题。
When the sample size is small, the normal approximation may not be valid for testing proportions. IB examinations usually specify when a normal approximation is acceptable, so read the question carefully.
当样本量较小时，正态近似可能不适用于比例检验。IB考试通常会说明何时可以使用正态近似，所以请仔细阅读题目。
Finally, always state your conclusion in context. Write "the data provide sufficient evidence to conclude that..." or "there is insufficient evidence to conclude that..." using the parameters and units from the question.
最后，始终结合情境陈述结论。写“数据提供了足够的证据表明……”或“没有足够的证据表明……”，并引用题目中的参数和单位。
12. Conclusion | 总结
Hypothesis testing is a logical blend of probability, statistics, and decision-making. Understanding the principles behind H₀, H₁, p-values, errors, and significance levels will not only help you score well on IB exams but also build a solid foundation for university-level statistics.
假设检验是概率、统计和决策制定的逻辑融合。理解 H₀、H₁、p值、错误类型和显著性水平背后的原理，不仅帮助你在IB考试中取得好成绩，也为大学阶段的统计学打下坚实基础。
The key to success is practice. Work through IB past-paper questions, and for every question, write down the hypotheses, identify the correct test, use your GDC carefully, compare the p-value with α, and write a contextual conclusion.
成功的关键在于练习。认真做IB历年真题，并且对每道题都写出假设、识别正确的检验、谨慎使用计算器、将p值与 α 比较，并写出结合情境的结论。
Remember that hypothesis testing does not prove anything with absolute certainty. It is a disciplined way of weighing evidence. As long as you respect the logic of the method, you will be able to interpret statistical results correctly in exams and beyond.
请记住，假设检验并不能绝对确定地证明任何事物。它是一种严谨的衡量证据的方式。只要你尊重这一方法的逻辑，你就能在考试及以后的学习生活中正确解读统计结果。
Published by TutorHao | IB 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