📚 Hypothesis Testing: IB & WJEC Mathematics Key Points | IB & WJEC 数学：假设检验考点精讲
Hypothesis testing is a cornerstone of inferential statistics, allowing us to evaluate claims about population parameters based on sample data. Mastering this topic is essential for success in IB Mathematics (AI and AA) and WJEC specifications, as it appears frequently in both Paper 2 and Paper 3 across Standard and Higher Level. This article breaks down key concepts, common pitfalls, and exam-ready strategies in a clear, bilingual format.
假设检验是推断统计学的基石，它能让我们根据样本数据评估关于总体参数的声明。掌握这一主题对于在 IB 数学（AI 和 AA）以及 WJEC 考试中取得成功至关重要，因为它经常出现在标准级和高级的试卷二和试卷三中。本文以清晰的双语格式分解关键概念、常见陷阱和备考策略。
1. Introduction to Hypothesis Testing | 假设检验简介
Hypothesis testing provides a formal procedure for deciding whether to believe a claim (a hypothesis) about a population. We use sample evidence to assess the strength of that claim against an opposing viewpoint. The process always begins by identifying the null hypothesis (H₀) and the alternative hypothesis (H₁).
假设检验提供了一个正式的程序，用于决定是否相信有关总体的某个声明（假设）。我们利用样本证据评估该声明与对立观点的强度。过程总是从确定零假设（H₀）和备择假设（H₁）开始。
A test yields a p-value or a comparison with a critical value, and we conclude either to 'reject H₀' in favour of H₁, or to 'fail to reject H₀' due to insufficient evidence. Crucially, we never 'accept' H₀; we simply lack enough evidence to reject it.
检验会产生一个 p 值或与临界值进行比较，我们得出结论：要么“拒绝 H₀”而支持 H₁，要么“未能拒绝 H₀”因为证据不足。关键在于，我们从不“接受”H₀；我们只是没有足够的证据去拒绝它。
2. Null and Alternative Hypotheses (H₀ and H₁) | 零假设与备择假设
ume to be true until evidence suggests otherwise. For example, H₀: μ = 50 or H₀: p = 0.25.
零假设，记作 H₀，是关于无效应、无差异或维持现状的声明。它是我们假定为真、直到有证据表明否则的假设。例如，H₀: μ = 50 或 H₀: p = 0.25。
The alternative hypothesis, H₁ (or Hₐ), represents the claim we are testing for. It contradicts H₀. Depending on the research question, H₁ can be one-sided (H₁: μ > 50 or μ < 50) or two-sided (H₁: μ ≠ 50).
备择假设 H₁（或 Hₐ）代表我们要检验的声明，它与 H₀ 矛盾。根据研究问题，H₁ 可以是单侧的（H₁: μ > 50 或 μ < 50），也可以是双侧的（H₁: μ ≠ 50）。
In IB and WJEC exams, always define both hypotheses clearly in symbols and words, as marks are awarded for precise notation and context.
在 IB 和 WJEC 考试中，务必用符号和文字清晰地定义两个假设，因为准确的符号和上下文会得到分数。
3. Significance Level and p-Value | 显著性水平与 p 值
The significance level, α, is the maximum probability of making a Type I error that the researcher is willing to tolerate. Common values are 0.05, 0.01, or 0.10. It defines the rejection region threshold.
显著性水平 α 是研究者愿意容忍的第一类错误的最大概率。常用值为 0.05、0.01 或 0.10。它定义了拒绝域的阈值。
lly ≤ α) indicates that the observed result is unlikely under H₀, leading us to reject H₀.
p 值是在 H₀ 为真的前提下，得到至少与实际观测一样极端的检验统计量的概率。小的 p 值（通常 ≤ α）表明在 H₀ 下观察到的结果不太可能发生，从而我们拒绝 H₀。
α, reject H₀; if p > α, do not reject H₀. Interpret the result in the context of the problem, not just a bare decision.
务必比较 p 值与 α：若 p ≤ α，拒绝 H₀；若 p > α，不拒绝 H₀。要在问题背景中解释结果，而不仅仅是给出决定。
4. One-Tailed and Two-Tailed Tests | 单尾检验与双尾检验
A one-tailed test examines whether a parameter is significantly greater than or less than a specified value. The p-value is halved when using symmetric two-tailed tables, or technology will compute the correct tail probability directly.
单尾检验考察参数是否显著大于或小于一个指定值。当使用对称的双尾表时，p 值减半，而技术工具会直接计算正确的尾部概率。
A two-tailed test checks for any significant difference, regardless of direction. The critical region is split into both tails. In IB, you must be careful to double the p-value from a one-tail table or let your GDC handle it automatically.
双尾检验检查任何显著差异，不论方向。拒绝域被分拆到两个尾部。在 IB 中，你必须小心地将单尾表中的 p 值加倍，或者让 GDC 自动处理。
Exam tip: base your choice of test on the wording of H₁. Words like 'higher' or 'increased' indicate a one-tailed test; 'changed' or 'different' suggest a two-tailed test.
考试技巧：根据 H₁ 的措辞选择检验方向。像“更高”或“增加”表明单尾检验；“改变”或“不同”则提示双尾检验。
5. Test Statistic and Critical Region | 检验统计量与拒绝域
A test statistic is a standardised value computed from sample data, used to determine whether to reject H₀. Examples include z, t, χ², or the number of successes in a binomial test. Its sampling distribution under H₀ must be known.
检验统计量是由样本数据计算出的标准化值，用于决定是否拒绝 H₀。例子包括 z、t、χ²，或者二项检验中的成功次数。它在 H₀ 下的抽样分布必须是已知的。
The critical region (or rejection region) is the set of values of the test statistic for which H₀ is rejected. Critical values are the boundaries of this region, determined by α. If the test statistic falls in the critical region, reject H₀.
临界域（或拒绝域）是导致拒绝 H₀ 的检验统计量值的集合。临界值是这个区域的边界，由 α 决定。如果检验统计量落入拒绝域，则拒绝 H₀。
d test at α = 0.05, the critical value is z = 1.645. Learn these commonly used thresholds.
对于 α = 0.05 的双侧 z 检验，临界值为 z = ±1.96。对于 α = 0.05 的右侧单尾检验，临界值为 z = 1.645。记住这些常用的阈值。
6. t-Test for a Population Mean | 总体均值的 t 检验
When the population standard deviation σ is unknown (the real-life scenario), we use the t-test. The test statistic is calculated as:
当总体标准差 σ 未知（现实场景），我们使用 t 检验。检验统计量计算公式为：
t = (x̄ − μ₀) / (s / √n)
where x̄ viation, and n is the sample size. The statistic follows a t-distribution with df = n − 1.
其中 x̄ 是样本均值，s 是样本标准差，n 为样本量。该统计量服从自由度为 df = n − 1 的 t 分布。
In your IB AI or WJEC exam, you will typically use a GDC to perform a 't-test' from a list of data or summary statistics. The calculator provides the p-value directly; compare it to α and write a concluding statement that references the mean.
在 IB AI 或 WJEC 考试中，你通常会使用 GDC 对数据列表或汇总统计执行“t 检验”。计算器直接给出 p 值；将其与 α 比较，并撰写引用均值的结论陈述。
7. z-Test for a Population Proportion | 总体比例的 z 检验
ormal approximation to the binomial when np₀ ≥ 10 and n(1−p₀) ≥ 10. The test statistic is:
对于分类数据，检验总体比例 p 时，在 np₀ ≥ 10 且 n(1−p₀) ≥ 10 条件下使用二项分布的正态近似。检验统计量为：
z = (p̂ − p₀) / √[p₀(1−p₀) / n]
where p̂ is the sample proportion. The p-value is then found from the standard normal distribution.
其中 p̂ 为样本比例。随后从标准正态分布中查找 p 值。
IB Mathematics: Applications and Interpretation HL specifically tests proportions using both binomial exact tests and normal approximations. For small n, exact binomial probabilities must be used, and you may be asked to compute the p-value by summing relevant binomial probabilities.
IB 数学：应用与解释 HL 专门用二项精确检验和正态近似测试比例。在小样本下，必须使用精确的二项概率，并且你可能会被要求通过求和相关的二项概率来计算 p 值。
8. Chi-Squared (χ²) Test for Independence | 卡方独立性检验
The χ² test assesses whether two categorical variables are associated. The hypotheses are H₀: the variables are independent, versus H₁: they are not independent. Expected frequencies are computed under the assumption of independence:
χ² 检验评估两个分类变量是否有关联。假设为 H₀：变量独立，对应 H₁：变量不独立。在独立假设下计算期望频数：
Expected = (row total × column total) / grand total
is χ² = Σ (O − E)² / E, with degrees of freedom (r−1)(c−1) for an r × c contingency table.
检验统计量为 χ² = Σ (O − E)² / E，对于一个 r × c 列联表，自由度为 (r−1)(c−1)。
Conditions: all expected frequencies must be at least 5. If any E < 5, you must combine categories. Your GDC will perform the χ²-test quickly; remember to state the test statistic, p-value, degrees of freedom, and conclusion in terms of independence.
条件：所有期望频数必须至少为 5。若有任何 E < 5，必须合并类别。你的 GDC 会快速执行 χ² 检验；记得陈述检验统计量、p 值、自由度，以及关于独立性的结论。
9. Type I and Type II Errors | 第一类错误和第二类错误
Understanding errors is a mandatory part of the syllabus. The following table summarises the four possible outcomes of hypothesis testing:
理解错误是课程大纲的必考部分。下表总结了假设检验的四种可能结果：
| Decision / Truth | H₀ True | H₀ False |
|---|---|---|
| Reject H₀ | Type I Error (α) | Correct Decision (Power) |
| Fail to Reject H₀ | Correct Decision | Type II Error (β) |
A Type I error (false positive) occurs when we reject a true H₀. The probability of this error is α, the significance level. A Type II error (false negative) occurs when we fail to reject a false H₀; its probability is β. The power of a test, 1 − β, is the probability of correctly rejecting a false H₀.
第一类错误（假阳性）发生在拒绝了一个真实的 H₀ 时。这类错误的概率是 α，即显著性水平。第二类错误（假阴性）发生在未能拒绝一个错误的 H₀ 时；其概率为 β。检验功效 1 − β 是正确拒绝错误 H₀ 的概率。
10. Confidence Intervals and Hypothesis Testing | 置信区间与假设检验
There is a direct duality between two-tailed hypothesis tests and confidence intervals. If a (1−α)100% confidence interval for a parameter excludes the null value μ₀, then the two-tailed test at level α will reject H₀. Conversely, if the interval contains μ₀, we do not reject H₀.
双侧假设检验与置信区间之间存在直接的对偶关系。如果一个参数的 (1−α)100% 置信区间不包括零假设值 μ₀，那么水平为 α 的双侧检验将拒绝 H₀。反之，若区间包含 μ₀，我们不拒绝 H₀。
For example, a 95% hip is tested in IB exams and helps you check your conclusion or avoid calculation errors.
例如，μ 的 95% 置信区间为 (8.2, 9.8) 不包括 10，因此 H₀: μ = 10 在 α = 0.05 下会被拒绝。IB 考试会考查这一关系，它帮助你验证结论或避免计算错误。
11. Using Technology: GDC Tips | 技术使用：GDC 技巧
for a 1-proportion z-test, 1-PropZTest; for χ² test, χ²-Test (enter observed matrix and expected will be stored). Always check that the p-value is displayed.
IB 和 WJEC 都要求熟练掌握图形计算器（GDC）。了解正确的菜单：t 检验通常为 [STAT] → TESTS → T-Test；单比例 z 检验为 1-PropZTest；χ² 检验为 χ²-Test（输入观测矩阵，期望值将被存储）。务必确认 p 值已显示。
th α, and state your conclusion. When using the critical value approach, report the test statistic and critical value. Full marks depend on showing the comparison and contextual conclusion.
如果你选择 p 值法，写出 p = ...（来自 GDC），与 α 比较，并陈述结论。当使用临界值法时，报告检验统计量和临界值。完全展示比较过程和语境化结论才能获得满分。
12. Common Pitfalls and Exam Strategy | 常见陷阱与考试策略
Misinterpreting the p-value: the p-value is not the probability that H₀ is true; it is the probability of observing data as extreme as ours given H₀ is true. Avoid phrasing like 'the chance the null is correct'.
误读 p 值：p 值不是 H₀ 为真的概率；它是在 H₀ 为真的条件下观察到极端数据的概率。避免使用“零假设正确的机会”这类表述。
Ignoring conditions: mly selected and the population approximately normal if n is small. For χ² tests, confirm E ≥ 5 and independent observations.
忽略条件：在检验之前务必检查假设条件（正态性、独立性、期望频数）。对于 t 检验，样本应随机选取，若 n 较小，总体应近似正态。对于 χ² 检验，确认 E ≥ 5 且观测值独立。
Conclusion in context: a bare 'reject H₀' loses marks. Write 'There is sufficient evidence at the 5% level to suggest that the mean height has increased.' (or similar). Tie the conclusion back to the original problem statement.
上下文中的结论：仅仅写“拒绝 H₀”会丢分。要写“在 5% 显著性水平下，有充分证据表明平均身高增加了”（或类似）。将结论与原始问题陈述联系起来。
Double p-value for two-tailed when using tables: if you obtain a one-tailed probability from a table, double it for a two-tailed test unless your GDC gives the two-tailed p-value directly.
使用表格时双尾 p 值加倍：如果你从表格中得到的是单尾概率，对于双尾检验必须加倍，除非你的 GDC 直接给出双尾 p 值。
Master these nuances, practise official past papers, and you'll handle hypothesis testing questions with confidence.
掌握这些细微差别，练习官方历年真题，你将自信应对假设检验问题。
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