📚 Rejection of the Null Hypothesis State | 拒绝原假设状态
In Edexcel A‑Level Mathematics, hypothesis testing is a statistical method that allows you to challenge an assumed ‘state’ about a population parameter. That assumed state is called the null hypothesis, and we decide whether to reject it based on sample evidence. Understanding exactly when and why we reject this state is central to the entire topic — it links significance levels, critical regions, p‑values and the types of possible error. This article unpacks every aspect of rejection so you can answer exam questions with confidence.
在爱德思 A‑Level 数学中，假设检验是一种统计方法，用于挑战关于总体参数的某种假定“状态”。这个假定状态就是原假设，我们依据样本证据决定是否拒绝它。准确理解何时以及为何拒绝这一状态，是整个专题的核心 —— 它联系着显著性水平、拒绝域、p 值和可能出现的两类错误。本文将全面拆解拒绝的方方面面，助你自信应对考试。
1. Hypothesis Testing: The Framework | 假设检验的基本框架
Every hypothesis test begins with two competing statements: the null hypothesis (H₀) and the alternative hypothesis (H₁). The null hypothesis represents the currently accepted ‘state’ — often ‘no change’, ‘no difference’ or a specific parameter value. The alternative hypothesis is what you are trying to find evidence for. We collect a random sample, calculate a test statistic and then decide whether the sample data provide sufficient evidence to reject H₀.
每一个假设检验都始于两个对立的陈述：原假设（H₀）和备择假设（H₁）。原假设代表当前被接受的“状态”—— 通常是“无变化”、“无差异”或某个特定参数值。备择假设则是你试图寻找证据支持的主张。我们收集随机样本，计算检验统计量，然后判断样本数据是否提供了足够的证据来拒绝 H₀。
If the test statistic falls inside a pre‑determined rejection region, we reject the null hypothesis state; otherwise, we do not reject it. Critically, we never ‘accept’ H₀ — we simply lack enough evidence to throw it away, much like a jury returning a ‘not guilty’ verdict rather than proving innocence.
如果检验统计量落入预先确定好的拒绝域内，我们就拒绝原假设状态；否则，我们不拒绝它。关键点在于，我们从不“接受”H₀ —— 只是没有足够的证据把它推翻，就像陪审团给出“无罪”判决，而并非证明清白。
2. Null and Alternative Hypotheses: Defining the State | 原假设与备择假设：定义状态
The null hypothesis always carries an equals sign: H₀: μ = 50, H₀: p = 0.3, or H₀: μ₁ – μ₂ = 0. This is the ‘ e to be true unless the data strongly contradict it. The alternative hypothesis can be one‑sided (μ > 50, μ < 50) or two‑ whether the rejection region lies in one tail or both tails of the sampling distribution.
原假设总是带有等号：H₀: μ = 50，H₀: p = 0.3，或 H₀: μ₁ – μ₂ = 0。这就是我们假定为真、除非数据强烈反驳的“状态”。备择假设可以是单侧的（μ > 50，μ < 50）或双侧的（μ ≠ 50）。H₁ 的选择直接决定了拒绝域落在抽样分布的单尾还是双尾。
In Edexcel exam questions, you will often be asked to write the hypotheses in words and in symbols. Always define your parameter clearly — ‘Let μ be the mean length …’ or ‘Let p be the proportion of defective items …’ — so the state under test is unambiguous.
在爱德思的考题中，常常要求你用文字和符号写出假设。一定要清晰地定义参数 —— “令 μ 为平均长度……”或“令 p 为次品比例……”—— 使被检验的状态没有歧义。
3. What Is a Rejection Region? | 什么是拒绝域？
A rejection region (also called the critical region) is the set of values of the test statistic for which the null hypothesis is rejected. It is designed so that, if H₀ is true, the probability of obtaining a test statistic in that region is exactly equal to the significance level α (typically 0.05 or 0.01).
拒绝域（也叫临界域）是检验统计量取值的集合，当统计量落入该集合时，原假设被拒绝。它的设计保证：如果 H₀ 为真，获得落入该区域的检验统计量的概率恰好等于显著性水平 α（通常为 0.05 或 0.01）。
Think of the rejection region as a ‘rare event’ zone under the null hypothesis state. When our observed statistic lands there, we interpret it as strong evidence against H₀, and we reject the state. The boundary values that mark the start of the rejection region are called critical values.
可以把拒绝域想象成原假设状态下的“稀有事件”区。当我们的观察统计量落入其中时，我们将其解读为反对 H₀ 的有力证据，并拒绝该状态。标记拒绝域起点的边界值称为临界值。
4. Significance Level and Critical Values | 显著性水平与临界值
true. It is chosen by the researcher before the test is carried out. Common choices are 5% (α = 0.05) and 1% smaller α makes it harder to reject the null state, reducing the risk of a false positive.
显著性水平 α 是当 H₀ 实际为真时拒绝它的概率。由研究者在检验前选定。常用取值是 5% (α = 0.05) 和 1% (α = 0.01)。α 越小，拒绝原假设状态就越困难，从而降低假阳性的风险。
f the test statistic. For a normal distribution with known variance, critical values are z‑scores. For a one‑tailed test at α = 0.05, the critical z‑value might be 1.645; for a two‑tailed test, ±1.96. These values split the rejection region from the non‑rejection region.
临界值由 α 以及检验统计量的分布决定。对于方差已知的正态分布，临界值为 z 分数。例如，单尾检验 α = 0.05 时，临界 z 值为 1.645；双尾检验为 ±1.96。这些值将拒绝域与非拒绝域分开。
If Z > 1.645, reject H₀ (right‑tailed test).
如果 Z > 1.645，则拒绝 H₀（右尾检验）。
5. One‑tailed vs Two‑tailed Tests | 单尾检验与双尾检验
The alternative hypothesis dictates the shape of the rejection region. For H₁: μ > 50, the rejection region lies entirely in the right tail. For H₁: μ < 50, it lies in the left tail. These are one‑tailed tests. For H₁: μ ≠ 50, the rejection region is split equally between both tails — a two‑tailed test. Each tail contains area α/2.
备择假设决定了拒绝域的形状。对于 H₁: μ > 50，拒绝域完全位于右尾。对于 H₁: μ < 50，位于左尾。这些都是单尾检验。对于 H₁: μ ≠ 50，拒绝域被平等地分到双尾 —— 双尾检验。每尾包含的面积是 α/2。
In Edexcel problems, you must be careful: if the question hints at a direction (‘more than’, ‘increased’, ‘higher’), a one‑tailed test is appropriate. If it simply asks if there is a ‘difference’ or ‘change’, use a two‑tailed test. Choosing the wrong side will lead to an incorrect rejection region and lost marks.
在处理爱德思考题时，务必小心：如果问题暗示了方向（“多于”、“增加”、“更高”），适用单尾检验。如果只是问是否存在“差异”或“变化”，则使用双尾检验。选错尾部会导致错误的拒绝域，从而丢分。
6. Calculating the Rejection Region for a Normal Distribution | 正态分布下拒绝域的计算
When the population variance is known (or the sample is large, n > 30), the test statistic follows a standard normal distribution after standardisation:
Z = (x̄ – μ₀) / (σ / √n)
当总体方差已知（或大样本 n > 30），检验统计量经标准化后服从标准正态分布：
Z = (x̄ – μ₀) / (σ / √n)
he null state. The rejection region is then defined by comparing Zcalc with the critical z‑value from tables. For a right‑tailed test at α = 0.05, reject H₀ if Zcalc > 1.645. For a two‑tailed test, reject if |Zcalc| > 1.96.
这里的 μ₀ 是原假设状态下的假设均值。然后通过比较计算出的 Zcalc 与查表得到的临界 z 值来确定拒绝域。在 α = 0.05 的右尾检验中，若 Zcalc > 1.645 则拒绝 H₀。在双尾检验中，若 |Zcalc| > 1.96 则拒绝。
metimes the question asks you to find the critical value in terms of the original units (x̄crit) rather than the z‑score. You can rearrange the standardisation formula: x̄crit = μ₀ + z × (σ / √n). Any sample mean greater than this value would cause rejection of the null state.
有时题目要求你求出以原始单位（x̄crit）表示的临界值，而不是 z 分数。你可以对标准化公式变形：x̄crit = μ₀ + z × (σ / √n)。任何比这个值大的样本均值都会导致拒绝原假设状态。
7. Using p‑values to Make Decisions | 利用 p 值做决策
A p‑value is the probability, under the null hypothesis state, of obtaining a test statistic at least as extreme as the one observed. It provides an alternative way to decide whether to reject. If the p‑value is less than or equal to α, we reject H₀. If the p‑value > α, we do not reject H₀.
p 值是在原假设状态下，获得至少与观测统计量一样极端的结果的概率。它提供了另一种决策方式。若 p 值 ≤ α，拒绝 H₀；若 p 值 > α，不拒绝 H₀。
Edexcel often requires you to state a conclusion like ‘Reject H₀ because p (0.021) < 0.05.’ The p‑value method is especially common in questions involving binomial hypothesis tests, where exact probabilities are calculated.
爱德思常常要求你给出这样的结论：“因为 p (0.021) < 0.05，拒绝 H₀。” p 值法在涉及二项假设检验的题目中尤为常见，因为那里会计算精确概率。
p‑value = P(Z > zobs | H₀ true) for a right‑tailed test.
右尾检验中，p 值 = P(Z > zobs | H₀ 为真)。
8. Type I and Type II Errors | 第一类错误和第二类错误
Whenever we reject a null state, there is a risk of making a mistake. A Type I error occurs when we reject H₀ but H₀ is actually true. Its probability is exactly α, the significance level. A Type II error occurs when we fail to reject H₀, but H₀ is false. Its probability is denoted by β, and (1 – β) is called the power of the test — the probability of correctly rejecting a false null state.
每当我们拒绝一个原假设状态，就有犯错的風險。第一类错误发生在 H₀ 为真却被拒绝时。其概率就是显著性水平 α。第二类错误发生在 H₀ 为假但我们未能拒绝它时。它的概率记为 β，而 (1 – β) 称为检验的功效 —— 即正确拒绝一个错误原假设状态的概率。
In Edexcel contexts, reducing α makes it harder to reject H₀, which decreases the chance of a Type I error but increases the chance of a Type II error (unless the sample size is increased). Conversely, a larger sample size reduces both error probabilities for the same α.
在爱德思的语境下，减小 α 会使拒绝 H₀ 更难，从而降低第一类错误的风险，但同时会增加第二类错误的风险（除非增加样本容量）。反之，增大样本容量，在相同 α 下，可同时降低两类错误的概率。
9. Rejection Region for Binomial Hypothesis Tests | 二项分布假设检验的拒绝域
. Because the binomial distribution is discrete, the actual significance level may be slightly below α.
当检验总体比例 p 时，检验统计量 X（成功次数）在原假设状态下服从二项分布 B(n, p₀)。拒绝域通过找出那些累积概率 ≤ α 的极端 X 值来确定。由于二项分布是离散的，实际显著性水平可能略低于 α。
For a right‑ p₀, you locate the smallest r such that P(X ≥ r) ≤ α, and the rejection region is {r, r+1, …, n}. Edexcel marks critical‑ th the significance level. Sometimes you are asked to find the critical region explicitly.
对于右尾检验 H₁: p > p₀，你需要找到最小的 r 使得 P(X ≥ r) ≤ α，那么拒绝域就是 {r, r+1, …, n}。爱德思注重临界思维：你必须展示 P(X ≥ 观察值) 并与显著性水平比较。有时会要求明确求出拒绝域。
If X ~ B(20, 0.3), H₁: p > 0.3, α = 0.05, find r: P(X ≥ r) ≤ 0.05.
如果 X ~ B(20, 0.3)，H₁: p > 0.3，α = 0.05，求 r：P(X ≥ r) ≤ 0.05。
10. Summary and Exam Tips | 总结与考试技巧
Rejecting the null hypothesis state is not a casual decision — it must be justified by a test statistic lying in the appropriate rejection region, or by a p‑value smaller than α. Always state your conclusion in context: ‘There is sufficient evidence at the 5% level to reject the claim that …’ or ‘There is insufficient evidence to reject H₀.’ Never say ‘prove H₁’.
拒绝原假设状态并非随意的决定 —— 必须由落入相应拒绝域内的检验统计量，或者小于 α 的 p 值来提供正当理由。结论一定要结合语境陈述：“在5%水平上有充分证据拒绝……的主张”，或“没有足够证据拒绝 H₀”。绝不可以说“证明 H₁”。
Key exam tips: (1) Always define the test statistic and state its distribution under H₀. (2) Clearly mark the significance level. (3) Draw a sketch of the rejection region – this helps avoid one‑tailed vs two‑tailed confusion. (4) When using p‑values, show the probability calculation and comparison with α. (5) For binomial tests, state the actual significance level if it differs from α. (6) Always write a full, contextual conclusion.
关键的考试技巧：(1) 始终定义检验统计量，并说明其在 H₀ 下的分布。(2) 清楚标明显著性水平。(3) 画出拒绝域的示意图 —— 这有助于避免单尾与双尾的混淆。(4) 使用 p 值时，展示概率计算并与 α 比较。(5) 对于二项检验，若实际显著性水平与 α 不同，要说明。(6) 务必写出完整的、结合上下文的结论。
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