Hypothesis Testing Core Concepts | 假设检验核心概念梳理

📚 Hypothesis Testing Core Concepts | 假设检验核心概念梳理

Hypothesis testing is one of the most frequently examined topics in IB Mathematics HL Analysis and Approaches and Applications and Interpretation. This article systematically unpacks the core concepts, terminology, and decision-making logic behind hypothesis testing to help you master the entire process with confidence.

假设检验是IB数学HL分析与方法（AA）以及应用与解释（AI）中考查频率最高的考点之一。本文系统梳理假设检验的核心概念、术语与决策逻辑，帮助你完整掌握这一知识模块，从容应对考试。

1. What Is a Hypothesis Test | 什么是假设检验

A hypothesis test is a formal statistical procedure used to decide whether there is enough evidence in a sample to support a particular claim about a population parameter. It begins with two opposing statements: the null hypothesis and the alternative hypothesis, and uses sample data to determine which statement is more plausible.

假设检验是一种正式的统计推断程序，用于判断样本数据是否提供足够证据来支持关于总体参数的某一主张。整个过程从两个相互对立的陈述出发：零假设和备择假设，并利用样本数据判断哪一个陈述更为合理。

The logic mirrors a criminal trial: the defendant is presumed innocent (null hypothesis) until evidence beyond a reasonable doubt (sample evidence) proves otherwise. You never prove the null hypothesis true; you simply fail to reject it when evidence is insufficient.

其逻辑类似刑事审判：被告在排除合理怀疑的证据出现之前被假定无罪（零假设）。你永远不会“证明”零假设为真；当证据不足时，你只是“无法拒绝”它。

2. The Null Hypothesis \( H_0 \) | 零假设

The null hypothesis, denoted \( H_0 \), is a statement of “no effect,” “no difference,” or “no change.” It always contains equality: =, ≤, or ≥. For example, if a manufacturer claims that the mean weight of a cereal box is 500 g, the null hypothesis would be \( H_0: \mu = 500 \).

零假设，记作 \( H_0 \)，是“无效应”“无差异”或“无变化”的陈述。它总是包含等号：=、≤ 或 ≥。例如，若某制造商声称麦片盒平均重量为500克，则零假设为 \( H_0: \mu = 500 \)。

Why do we always test \( H_0 \) rather than \( H_1 \)? Because assuming the null is true gives us a precise probability distribution for the test statistic, allowing us to calculate how unlikely our sample result would be under that assumption.

为什么我们总是检验 \( H_0 \) 而不是 \( H_1 \)? 因为假设零假设为真，我们就能获得检验统计量的精确概率分布，从而计算在零假设成立下样本结果出现的概率有多小。

3. The Alternative Hypothesis \( H_1 \) | 备择假设

The alternative hypothesis, denoted \( H_1 \), represents what we suspect might be true if the null is false. It can be one-tailed (directional) or two-tailed (non-directional):

备择假设，记作 \( H_1 \)，表示当我们怀疑零假设不成立时可能为真的情况。它可以是单尾（有方向）或双尾（无方向）的：

Two-tailed: \( H_1: \mu \neq 500 \) — the mean is different (either higher or lower).
One-tailed (right): \( H_1: \mu > 500 \) — the mean is higher than 500.
One-tailed (left): \( H_1: \mu < 500 \) — the mean is lower than 500.

双尾检验：\( H_1: \mu \neq 500 \) —— 均值不同（更高或更低）。
单尾检验（右尾）：\( H_1: \mu > 500 \) —— 均值高于500。
单尾检验（左尾）：\( H_1: \mu < 500 \) —— 均值低于500。

In IB exams, read the question carefully: phrases like “has increased” or “has decreased” signal a one-tailed test, while “has changed” or “is different” signal a two-tailed test.

在IB考试中，请仔细读题：“增加了”或“减少了”提示使用单尾检验；而“改变了”或“不同了”提示使用双尾检验。

4. Significance Level \( \alpha \) | 显著性水平

The significance level \( \alpha \) is the threshold probability for rejecting \( H_0 \). It represents the maximum probability of making a Type I error that you are willing to accept. Common values in IB are 10%, 5%, and 1%.

显著性水平 \( \alpha \) 是拒绝 \( H_0 \) 的阈值概率，表示你愿意承担的第一类错误的最大概率。IB考试中常见取值为10%、5%和1%。

If \( \alpha = 0.05 \), it means: “If \( H_0 \) is actually true, I accept a 5% chance of wrongly rejecting it.” The smaller \( \alpha \), the stronger the evidence required to reject \( H_0 \).

若 \( \alpha = 0.05 \)，其含义是：“如果 \( H_0 \) 实际上为真，我接受5%的概率错误地拒绝它。” \( \alpha \) 越小，拒绝 \( H_0 \) 所需的证据就越强。

P(reject \( H_0 \) | \( H_0 \) true) = \( \alpha \)

P(拒绝 \( H_0 \) | \( H_0 \) 为真) = \( \alpha \)

5. p-Value Explained | p值的含义

The p-value is the probability of obtaining a test statistic at least as extreme as the one observed, assuming the null hypothesis is true. It quantifies how surprising the sample result is under the null hypothesis.

p值是在零假设为真的前提下，获得与观测值至少一样极端的检验统计量的概率。它量化了在零假设成立时样本结果有多么“出乎意料”。

Decision rule: if \( p \le \alpha \), reject \( H_0 \); if \( p > \alpha \), fail to reject \( H_0 \). A small p-value indicates the sample evidence is unlikely under the null, thus providing support for \( H_1 \).

决策规则：若 \( p \le \alpha \)，拒绝 \( H_0 \)；若 \( p > \alpha \)，无法拒绝 \( H_0 \)。较小的p值表明样本证据在零假设下不太可能出现，从而支持 \( H_1 \)。

Common misconception: the p-value is NOT the probability that \( H_0 \) is true. This is a frequent IB exam trap; always interpret the p-value conditionally on \( H_0 \) being true.

常见误区：p值不是 \( H_0 \) 为真的概率。这是IB考试中常见的陷阱；务必基于 \( H_0 \) 为真的前提来解释p值。

6. Critical Region and Critical Value | 临界区域与临界值

The critical region (rejection region) is the set of values of the test statistic for which \( H_0 \) is rejected. Its boundary is called the critical value. For a two-tailed test at \( \alpha = 0.05 \) using a z-distribution, the critical values are \( \pm 1.96 \).

临界区域（拒绝域）是检验统计量取值中导致拒绝 \( H_0 \) 的集合，其边界称为临界值。例如，在 \( \alpha = 0.05 \) 下使用z分布进行双尾检验时，临界值为 \( \pm 1.96 \)。

For a one-tailed test at \( \alpha = 0.05 \), the critical value is \( +1.645 \) (right tail) or \( -1.645 \) (left tail). You can find these from the standard normal distribution table or your GDC.

对于 \( \alpha = 0.05 \) 的单尾检验，临界值为 \( +1.645 \)（右尾）或 \( -1.645 \)（左尾）。这些值可通过标准正态分布表或图形计算器（GDC）获取。

Reject \( H_0 \) if \( z \) is in the critical region \( z \ge z_{\alpha} \) for a right-tailed test.

对右尾检验，若 \( z \ge z_{\alpha} \)，则拒绝 \( H_0 \)。

7. Type I and Type II Errors | 第一类与第二类错误

Two types of errors can occur in hypothesis testing. A Type I error is rejecting a true null hypothesis; its probability is exactly \( \alpha \). A Type II error, denoted \( \beta \), is failing to reject a false null hypothesis.

假设检验中可能发生两类错误。第一类错误是拒绝了一个真实的零假设，其概率恰为 \( \alpha \)。第二类错误记作 \( \beta \)，是未能拒绝一个错误的零假设。

Decision \ Reality	\( H_0 \) True	\( H_0 \) False
Reject \( H_0 \)	Type I Error (probability \( \alpha \))	Correct Decision
Fail to reject \( H_0 \)	Correct Decision	Type II Error (probability \( \beta \))

Relationship: \( 1 – \beta \) is called the power of the test.

关系式：\( 1 – \beta \) 称为检验的势（功效）。

Note that reducing \( \alpha \) generally increases \( \beta \), and vice versa. The only way to reduce both is to increase the sample size n.

注意，减小 \( \alpha \) 通常会增大 \( \beta \)，反之亦然。唯一能同时减小两者的方法是增大样本量 n。

8. Test Statistic for the Mean \( \mu \) | 均值的检验统计量

When testing a hypothesis about a population mean \( \mu \) and the population standard deviation \( \sigma \) is known, the test statistic is:

当检验总体均值 \( \mu \) 且总体标准差 \( \sigma \) 已知时，检验统计量为：

\( z = \frac{\bar{x} – \mu_0}{\sigma / \sqrt{n}} \)

Here \( \bar{x} \) is the sample mean, \( \mu_0 \) is the value claimed under \( H_0 \), \( \sigma \) is the population standard deviation, and n is the sample size. This statistic follows a standard normal distribution when n is large or the population is normal.

其中 \( \bar{x} \) 为样本均值，\( \mu_0 \) 是 \( H_0 \) 中声称的数值，\( \sigma \) 是总体标准差，n 为样本量。当n较大或总体服从正态分布时，该统计量服从标准正态分布。

If \( \sigma \) is unknown (which is common in IB), we use the sample standard deviation s and the t-distribution with n − 1 degrees of freedom: \( t = \frac{\bar{x} – \mu_0}{s / \sqrt{n}} \).

若 \( \sigma \) 未知（IB考试中常见），则使用样本标准差 s 和自由度为 n − 1 的t分布：\( t = \frac{\bar{x} – \mu_0}{s / \sqrt{n}} \)。

9. Hypothesis Test for Proportions | 比例的假设检验

When testing a population proportion p, the test statistic is based on the normal approximation to the binomial distribution. For a claimed proportion \( p_0 \):

检验总体比例 p 时，检验统计量基于二项分布的正态近似。对于声称的比例 \( p_0 \)：

\( z = \frac{\hat{p} – p_0}{\sqrt{\frac{p_0(1 – p_0)}{n}}} \)

Here \( \hat{p} = \frac{x}{n} \) is the sample proportion, x is the number of successes, and n is the sample size. This approximation is valid when \( n p_0 \ge 10 \) and \( n(1 – p_0) \ge 10 \).

其中 \( \hat{p} = \frac{x}{n} \) 为样本比例，x 为成功次数，n 为样本量。该近似在 \( n p_0 \ge 10 \) 且 \( n(1 – p_0) \ge 10 \) 时有效。

In IB Applications and Interpretation, this is particularly relevant for surveys and opinion polls. Always check the validity conditions before applying the normal approximation.

在IB应用与解释课程中，该检验常应用于调查和民意测验。使用正态近似前务必检查有效性条件。

10. Confidence Intervals and Hypothesis Testing | 置信区间与假设检验的联系

There is a beautiful duality between confidence intervals and two-tailed hypothesis tests. A two-tailed test at significance level \( \alpha \) rejects \( H_0 \) if and only if the value \( \mu_0 \) lies outside the \( (1 – \alpha) \times 100\% \) confidence interval for \( \mu \).

置信区间与双尾假设检验之间存在优美的对偶性。在显著性水平 \( \alpha \) 下，双尾检验拒绝 \( H_0 \) 当且仅当 \( \mu_0 \) 落在 \( \mu \) 的 \( (1 – \alpha) \times 100\% \) 置信区间之外。

For example, if a 95% confidence interval for the mean is (45.2, 52.8), then a two-tailed test of \( H_0: \mu = 50 \) cannot be rejected at the 5% level, but \( H_0: \mu = 42 \) would be rejected at the same level.

例如，若均值的95%置信区间为(45.2, 52.8)，则在5%显著性水平下无法拒绝双尾检验 \( H_0: \mu = 50 \)，但 \( H_0: \mu = 42 \) 将被拒绝。

This relation helps you verify your answers using GDC: compute both the confidence interval and the test statistic to cross-check consistency.

利用这一关系可帮助你用GDC验证答案：同时计算置信区间和检验统计量，检查结果是否一致。

11. Step-by-Step Procedure in IB Exams | IB考试中的完整步骤

In IB exam papers, marker schemes reward a structured approach. Follow these six steps to maximise marks:

在IB考试卷中，评分方案奖励结构清晰的解答。遵循以下六个步骤以最大化得分：

Define the parameter and write down \( H_0 \) and \( H_1 \) clearly.
State the significance level \( \alpha \) and the type of test (one- or two-tailed).
Calculate the test statistic using the appropriate formula, showing all working.
Find the critical value or the p-value from your GDC or statistical tables.
Compare: reject \( H_0 \) if the test statistic lies in the critical region, or if \( p \le \alpha \).
Write a conclusion in the context of the question, using formal language.

定义参数并清晰写出 \( H_0 \) 和 \( H_1 \)。
说明显著性水平 \( \alpha \) 和检验类型（单尾或双尾）。
使用相应公式计算检验统计量，展示全部计算过程。
使用GDC或统计用表查找临界值或p值。
比较判断：若检验统计量落入临界区域，或 \( p \le \alpha \)，则拒绝 \( H_0 \)。
结合题目情境写出结论，使用规范统计语言。

For the conclusion, use one of these phrases exactly: “There is sufficient evidence at the \( \alpha\% \) significance level to reject \( H_0 \), and conclude that \( H_1 \) is supported.” or “There is insufficient evidence to reject \( H_0 \).” Avoid saying “accept \( H_0 \)” — this is a common mark deduction.

写结论时请使用规范表述：“在 \( \alpha\% \) 显著性水平下有足够证据拒绝 \( H_0 \)，认为 \( H_1 \) 成立。”或“没有足够证据拒绝 \( H_0 \)。”切忌说“接受 \( H_0 \)”——这是常见的失分点。

12. Common Pitfalls and Exam Tips | 常见陷阱与考试建议

Here are the most frequent mistakes students make in IB hypothesis testing questions, and how to avoid them:

以下是IB学生在假设检验题目中最常犯的错误及规避方法：

Mixing up hypotheses: Always write \( H_0 \) first with the equality sign, and write \( H_1 \) according to the direction stated in the problem.
Using the wrong tail: Read the question for keywords like “greater than” (right-tailed), “less than” (left-tailed), or “different from” (two-tailed).
Misinterpreting the p-value: The p-value is calculated under \( H_0 \) being true, not the probability that \( H_0 \) is true.
Ignoring the conditions: For t-tests, the sample should be approximately normally distributed or n should be large; for proportion tests, check \( np_0 \ge 10 \).
Weak conclusion: Always connect your conclusion to the real-world context of the question, not just a statistical statement.

混淆假设：先写含等号的 \( H_0 \)，再根据题干方向写出 \( H_1 \)。
尾端选择错误：关注关键词：“大于”（右尾）、“小于”（左尾）、“不同”（双尾）。
误解p值：p值是在 \( H_0 \) 为真的前提下计算的，并非 \( H_0 \) 为真的概率。
忽略条件：t检验要求样本近似正态或n较大；比例检验要求 \( np_0 \ge 10 \)。
结论随意：务必结合题目情境写出结论，而不仅是统计表述。

Finally, always set your GDC to use the correct distribution (normal or t) and ensure you know how to compute p-values directly using the appropriate GDC function. In the new IB syllabus, GDC proficiency is essential for maximizing efficiency in hypothesis testing questions.

最后，务必确保GDC使用正确的分布（正态或t），并熟练掌握用GDC功能直接计算p值。在新IB大纲中，熟练使用GDC是在假设检验题目中提高效率的关键。

Published by TutorHao | Mathematics Revision Series | aleveler.com

更多咨询请联系16621398022（同微信）

Hypothesis Testing Core Concepts | 假设检验核心概念梳理