📚 Hypothesis Testing for IB and Edexcel Maths | IB与爱德思数学：假设检验考点精讲
Hypothesis testing is a cornerstone of statistical inference, integral to both the IB Diploma Programme (Analysis & Approaches and Applications & Interpretation) and the Edexcel A Level Mathematics specification. This revision guide covers all essential concepts, from formulating null and alternative hypotheses to interpreting p-values and avoiding common errors. You will learn how to apply hypothesis tests using binomial and normal distributions, understand critical regions, and distinguish between one-tailed and two-tailed tests, with a focus on what examiners expect in both syllabi.
假设检验是统计推断的基石，在IB文凭项目（分析与方法、应用与解释）和爱德思A Level数学中都不可或缺。本考点精讲涵盖从建立原假设与备择假设到解释p值、避免常见错误等所有核心内容。你将学会如何运用二项分布与正态分布进行假设检验，理解临界区域，并能区分单尾与双尾检验，同时紧扣两大课程体系中考官的要求。
1. What is Hypothesis Testing? | 什么是假设检验？
Hypothesis testing is a statistical procedure that uses sample data to evaluate a claim about a population parameter. It begins with a null hypothesis, representing the status quo or a statement of no effect, and then assesses whether the observed sample provides sufficient evidence to reject this null hypothesis in favour of an alternative hypothesis. The process involves calculating a test statistic, determining a p-value or finding a critical region, and drawing a conclusion that refers back to the original claim.
假设检验是一种利用样本数据评估关于总体参数的声明的统计方法。它从一个表示现状或无效声明的原假设开始，然后评估观察到的样本是否提供了足够证据来拒绝原假设并支持备择假设。这个过程包括计算检验统计量、确定p值或找到临界区域，并最终得出联系原始声明的结论。
In both IB and Edexcel, hypothesis testing is not just a mechanical procedure; it requires correct interpretations in context. For instance, failing to reject the null hypothesis does not prove it true—it simply means the evidence is insufficient to reject it.
在IB和爱德思考试中，假设检验不仅仅是机械的步骤；它要求结合背景进行正确解读。例如，未能拒绝原假设并不证明它为真——仅仅意味着证据不足以拒绝它。
2. Null and Alternative Hypotheses | 原假设与备择假设
herwise. The alternative hypothesis, H₁ or Hₐ, represents what we aim to support; it specifies that the parameter differs from the null value in some way. For a binomial test about a proportion p, a typical null might be H₀: p = 0.5, while alternatives could be H₁: p < 0.5, p > 0.5, or p ≠ 0.5 depending on the research question.
原假设，记作H₀，是关于总体参数取某个特定值的声明，在证据表明其不成立之前被假定为真。备择假设，记作H₁或Hₐ，代表我们试图支持的结论，它指明参数以某种方式不同于原假设的值。对于关于比例p的二项检验，典型的原假设可能是H₀: p = 0.5，而根据研究问题，备择假设可以是H₁: p < 0.5、p > 0.5 或 p ≠ 0.5。
ls, never sample statistics. IB and Edexcel examiners penalise vague wording; always state 'H₀: μ = 100' instead of 'the mean is 100'.
书写假设时必须使用总体参数符号，绝不能使用样本统计量。IB和爱德思考官对措辞不严谨会扣分；务必写成'H₀: μ = 100'，而不是'the mean is 100'。
3. Significance Level and Critical Region | 显著性水平与临界区域
The significance level, α, is the probability of rejecting H₀ when it is actually true (Type I error). Common choices are 0.05 (5%) or 0.01. The critical (or rejection) region is the set of values of the test statistic for which the null hypothesis is rejected. For a given α, the critical value(s) are the boundary of this region. If the test statistic falls in the critical region, H₀ is rejected.
显著性水平α是当H₀为真时拒绝它的概率（第一类错误）。常见取值为0.05（5%）或0.01。临界区域（或拒绝域）是检验统计量的取值集合，当统计量落入该区域时拒绝原假设。对于给定的α，临界值是该区域的边界。若检验统计量落入临界区域，则拒绝H₀。
In IB exams, you are often asked to find the critical region explicitly for a binomial distribution, while Edexcel questions may require you to compare the test statistic with a critical value obtained from tables.
在IB考试中，常要求明确求出二项分布的临界区域，而爱德思的题目可能要求你将检验统计量与从表中查得的临界值进行比较。
4. One-tailed vs Two-tailed Tests | 单尾检验与双尾检验
A one-tailed test is used when the alternative hypothesis specifies a direction (e.g., H₁: p < 0.3 or H₁: μ > 50). The critical region occupies one tail of the distribution. A two-tailed test is used when H₁ states that the parameter is simply different from the null value (e.g., H₁: p ≠ 0.5). In a two-tailed test at significance level α, the critical region is split symmetrically between both tails, each with area α/2.
当备择假设指明方向时（如H₁: p < 0.3或H₁: μ > 50），使用单尾检验。临界区域位于分布的一个尾部。当H₁仅指出参数不同于原假设值（如H₁: p ≠ 0.5）时，使用双尾检验。在显著性水平α的双尾检验中，临界区域对称地分配于两侧尾部，每侧面积为α/2。
| Test Type | 检验类型 | Alternative Hypothesis | 备择假设 | Critical Region | 临界区域 |
|---|---|---|
| One-tailed (left) | 单尾左 | H₁: parameter < value | Lower tail | 下侧尾部 |
| One-tailed (right) | 单尾右 | H₁: parameter > value | Upper tail | 上侧尾部 |
| Two-tailed | 双尾 | H₁: parameter ≠ value | Both tails, each α/2 | 两侧尾部，各α/2 |
5. The p-value Approach | p值法
ates strong evidence against the null hypothesis, leading to rejection. In IB Mathematics, the p-value method is the primary approach, and you are expected to compute p-values using binomial or normal distributions and then compare them with the significance level.
p值是在H₀为真的条件下，观察到至少与当前样本统计量一样极端的检验统计量的概率。当p值很小（≤ α）时，表明有强证据反对原假设，从而拒绝它。在IB数学中，p值法是主要方法，要求使用二项分布或正态分布计算p值，然后与显著性水平比较。
For a binomial test with test statistic X, a one-tailed left-side p-value is P(X ≤ x | prefer the method of comparing a one-tail probability with α/2 when using critical values.
对于检验统计量为X的二项检验，左侧单尾的p值为P(X ≤ x | H₀)，右侧为P(X ≥ x)。在双尾检验中，p值通常为较小尾概率的两倍，然而IB和爱德思在用临界值法时更倾向于将单尾概率与α/2比较。
6. Critical Value Method | 临界值法
ly. For a normal distribution test, the critical values for α = 0.05 in a two-tailed test are often ±1.96 (using Z ~ N(0,1)). If the test statistic falls beyond these values, H₀ is rejected.
不计算p值，你也可以找到定义拒绝域的临界值，并直接将观察到的检验统计量与之比较。对于正态分布检验，在α = 0.05的双尾检验中临界值通常为±1.96（使用Z ~ N(0,1)）。若检验统计量超出这些值，则拒绝H₀。
Edexcel exam papers frequently require candidates to look up critical values from statistical tables and state a conclusion in context. Ensure you round correctly and never truncate prematurely when working with discrete distributions such as the binomial.
爱德思的试卷常要求考生从统计表中查出临界值并在上下文中陈述结论。在处理离散分布（如二项分布）时，务必正确取整，切勿过早截断。
7. Hypothesis Testing with Binomial Distribution | 二项分布假设检验
: p = p₀. The test statistic is the number of successes X in n independent trials, where X ~ ernative hypothesis, you calculate the probability of obtaining X or a more extreme value.
当使用二项分布检验总体比例p时，原假设为H₀: p = p₀。检验统计量为n次独立试验中的成功次数X，在H₀下X ~ B(n, p₀)。根据备择假设，计算得到X或更极端值的概率。
tossed 50 times, yielding 33 heads. Test whether the coin is biased towards heads at the 5% significance level. H₀: p = 0.5, H₁: p > 0.5 (one-tailed). Under H₀, X ~ 0164 < 0.05, we reject H₀ and conclude there is evidence the coin is biased towards heads.
例题：一枚硬币抛掷50次，出现33次正面。在5%显著性水平下检验该硬币是否偏向正面。H₀: p = 0.5, H₁: p > 0.5（单尾）。在H₀下X ~ B(50, 0.5)。p值 = P(X ≥ 33) = 1 - P(X ≤ 32)。利用二项分布表或技术工具，该p值约等于0.0164。由于0.0164 < 0.05，拒绝H₀，结论为有证据表明硬币偏向正面。
ves the critical region X ≥ 32. Since 33 falls in this region, the conclusion is the same.
在爱德思试卷中，可能要求找出临界区域。对于上述检验，使P(X ≥ k) ≤ 0.05的最小k给出临界区域X ≥ 32。因为33落入该区域，结论一致。
8. Hypothesis Testing with Normal Distribution | 正态分布假设检验
rge sample size justifies the use of the normal distribution. The test statistic is Z = (x̄ - μ₀) / (σ / √n), where x̄ is the sample mean, μ₀ the hypothesised mean, σ the population standard deviation, and n the sample size.
许多IB和爱德思的题目涉及已知总体方差σ²或大样本可引用正态分布时的总体均值μ的检验。检验统计量为 Z = (x̄ - μ₀) / (σ / √n)，其中x̄为样本均值，μ₀为假设均值，σ为总体标准差，n为样本量。
Z = (x̄ - μ₀) / (σ / √n)
standard deviation of 8 g. A sample of 40 packs has a mean weight of 497 g. Test at the 1% than claimed. H₀: μ = 500, H₁: μ < 500 (one-tailed, left). Calculate Z = (497 - 500) / (8/√40 r p-value < 0.01, we reject H₀. There is evidence that the mean weight is less than 500 g.
例题：某生产商宣称方糖包平均重量为500克，标准差为8克。随机抽取40包样本的平均重量为497克。在1%显著性水平下检验平均重量是否低于宣称值。H₀: μ = 500, H₁: μ < 500（单尾左）。计算Z = (497 - 500) / (8/√40) ≈ -2.372。α=0.01左侧临界Z值为-2.326（或p值≈0.0089）。由于-2.372 < -2.326，或p值 < 0.01，拒绝H₀。有证据表明平均重量低于500克。
In IB, questions may involve the t-distribution for smaller samples when σ is unknown, but the structure remains similar. Always remember to use the appropriate distribution and state the degrees of freedom if necessary.
在IB中，若σ未知且样本量较小，可能涉及t分布，但结构类似。务必使用合适的分布，并在必要时注明自由度。
9. Type I and Type II Errors | 第一类与第二类错误
A Type I error occurs when a true null hypothesis is rejected. Its probability is exactly the significance level α. A Type II error happens when a false null hypothesis is not rejected; its probability is denoted β. The power of a test is 1 - β, the probability of correctly rejecting a false H₀.
第一类错误发生在原假设为真却被拒绝时，其概率正是显著性水平α。第二类错误发生在原假设为假却未被拒绝时，其概率记为β。检验的功效为1 - β，即正确拒绝错误H₀的概率。
| Truth \ Decision | 真实情况 \ 判定 | Reject H₀ | 拒绝H₀ | Do not reject H₀ | 不拒绝H₀ |
|---|---|---|
| H₀ true | H₀为真 | Type I error (α) | 第一类错误 | Correct decision | 正确决策 |
| H₀ false | H₀为假 | Correct decision (power) | 正确决策（功效） | Type II error (β) | 第二类错误 |
IB questions may ask you to explain the errors in context. For instance, in the coin example, a Type I error would be concluding the coin is biased when it is actually fair. A Type II error would be failing to detect a genuine bias.
IB题目可能要求结合背景解释错误。例如，在硬币例子中，第一类错误是硬币实际上公平却推断它有偏差；第二类错误是未能检测出真正的偏差。
10. Tips for IB & Edexcel Exams | IB与爱德思考试技巧
Always define your hypotheses using correct notation (H₀ and H₁) and state them clearly at the start. In contextual questions, link your conclusion to the scenario: 'There is sufficient evidence, at the 5% significance level, to suggest that the new drug reduces blood pressure.' Never write 'accept H₀' – instead use 'do not reject H₀'.
始终使用正确符号（H₀和H₁）定义假设并在开头明确陈述。在情境题中，将结论与场景关联：“在5%显著性水平下，有充分证据表明新药降低了血压。”切勿写“接受H₀”，而应使用“不拒绝H₀”。
smaller one-tail probability; check whether the question asks for a critical region instead. Edexcel candidates should be comfortable using cumulative distribution tables and writing critical regions as X ≤ c₁ or X ≥ c₂.
对于二项检验，在累加概率时要严格使用不等号。对于双尾检验，IB期望p值为较小侧概率的两倍；注意题目是否要求求出临界区域。爱德思考生应熟练使用累积分布表，并将临界区域写成X ≤ c₁ 或 X ≥ c₂。
In normal distribution problems, always standardise correctly and round critical values to the required number of decimal places. Show all steps: hypotheses, calculation of test statistic, comparison with critical value or p-value, and conclusion in context. Diagrams can help but are not mandatory unless requested.
在正态分布问题中，始终正确标准化并将临界值四舍五入到指定小数位数。展示所有步骤：假设、检验统计量的计算、与临界值或p值的比较，以及基于背景的结论。画图有助于理解，但除非题目要求，并非必须。
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