📚 GCSE Edexcel Maths: Hypothesis Testing Key Points | GCSE Edexcel 数学：假设检验考点精讲
Hypothesis testing is one of the most distinctive topics in the Edexcel GCSE Statistics and Higher tier Mathematics specification. It gives you a formal framework to decide whether sample data provides enough evidence to challenge an existing belief about a population proportion. This article breaks down every essential concept, from null and alternative hypotheses to critical regions and p-values, using clear bilingual explanations and targeted worked examples.
假设检验是爱德思 GCSE 统计学和高等数学大纲中最具特色的主题之一。它为你提供了一个正式的框架，来判断样本数据是否提供了足够的证据去质疑关于某个总体比例的既有看法。本文通过清晰的双语讲解和有针对性的示范例题，逐一剖析每一个核心概念，包括零假设与备择假设、临界区域以及 p 值法。
1. What Is Hypothesis Testing? | 什么是假设检验？
Hypothesis testing is a statistical decision-making process. You start with an initial assumption about a population parameter (the 'null hypothesis'), collect a random sample, and then calculate how likely it is to observe the sample result if the assumption were true. If the result is very unlikely, you reject the null hypothesis in favour of an alternative claim. At GCSE level, the parameter is almost always a binomial probability p, such as the proportion of defective items or the fairness of a coin.
假设检验是一种统计决策过程。你从一个关于总体参数的初步假设（“零假设”）入手，收集一个随机样本，然后计算在这个假设成立的前提下观察到该样本结果的可能性有多大。如果该结果非常不可能发生，你就拒绝零假设，接受备择主张。在 GCSE 水平，该参数几乎总是一个二项分布的概率 p，例如次品比例或硬币的公平性。
The process mirrors a courtroom trial: the defendant is presumed innocent (null hypothesis) unless the evidence is strong enough to prove guilt beyond reasonable doubt (the significance level). You never claim to 'prove' the alternative; you only decide whether the evidence is strong enough to reject the null.
整个过程类似于法庭审判：被告被假定无罪（零假设），除非证据强到足以排除合理怀疑（显著性水平）地证明有罪。你永远不会声称“证明”了备择假设成立；你只是判断证据是否足够强而可以拒绝零假设。
2. Null and Alternative Hypotheses | 零假设与备择假设
value]. The alternative hypothesis, H₁, specifies what you are testing for: it could be p < claimed value, p > claimed value, or p ≠ claimed value. The choice of H₁ determines whether the test is one-tailed or two-tailed.
零假设，记作 H₀，表明总体参数等于某个声称的值。对于比例，总是写作 H₀: p = [某个值]。备择假设 H₁ 则指明你正在检验的主张：可能是 p < 声称值、p > 声称值，或者 p ≠ 声称值。选择何种 H₁ 决定了检验是单尾还是双尾。
For example, if a factory claims that at most 10% d the alternative reflecting a suspicion that the true rate is higher would be H₁: p > 0.1. Always set up the hypotheses before looking at the data, and express them in terms of the parameter p, never the sample statistic.
例如，如果一家工厂声称其产品中至多有 10% 是次品，设 p 为一件产品是次品的概率。零假设是 H₀: p = 0.1，而反映怀疑真实次品率更高的备择假设则是 H₁: p > 0.1。务必在查看数据之前设立假设，并且始终用参数 p 来表述，而非用样本统计量。
3. Test Statistic and Significance Level | 检验统计量与显著性水平
The test statistic is the number of successes observed in the sample, denoted by X. Under the null hypothesis, X follows a binomial distribution: X ~ B(n, p₀), where n is the sample size and p₀ is the assumed probability of success. All probability calculations are based on this null distribution.
检验统计量是样本中观测到的成功次数，记作 X。在零假设下，X 服从二项分布：X ~ B(n, p₀)，其中 n 是样本量，p₀ 是假设的成功概率。所有的概率计算都基于这个零分布。
The significance level, denoted by alpha (α), is the maximum probability you are willing to accept of making a Type I error — rejecting a true null hypothesis. In GCSE exams, α is usually given as 5% (0.05) or sometimes 1% (0.01). If the probability of obtaining the observed result (or something more extreme) is smaller than α, you reject H₀.
显著性水平，记作 α，是你愿意承担的犯第一类错误——即错误地拒绝了一个真实的零假设——的最大概率。在 GCSE 考试中，α 通常取 5%（0.05），有时取 1%（0.01）。如果得到观测结果（或更极端结果）的概率小于 α，就拒绝 H₀。
A lower significance level makes it harder to reject H₀, reducing the chance of a false positive but increasing the chance of missing a real effect. The choice of α is a balance between these two types of error and is always stated at the start of the test.
更低的显著性水平使拒绝 H₀ 变得更难，从而降低了假阳性的机会，但增加了漏掉真实效应的机会。α 的选择是对这两类错误的权衡，并总是在检验开始时明确给出。
4. One-tailed and 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: H₁: p < value (left-tailed) or H₁: p > value (right-tailed). The whole of the significance level is placed in that single tail of the distribution. A two-tailed test is used for H₁: p ≠ value, splitting the significance level equally between both tails.
当备择假设指明了方向时，采用单尾检验：H₁: p < 某值（左尾检验）或 H₁: p > 某值（右尾检验）。整个显著性水平都放在分布的那一个尾部。对于 H₁: p ≠ 某值，则使用双尾检验，并将显著性水平均分到左右两个尾部。
f α = 0.05). Testing whether a new drug increases the recovery rate leads to a right-tailed test with H₁: p > claimed proportion.
例如，检验一枚硬币是否存在偏差（可能偏重于任何一面）需要使用双尾检验，H₁: p ≠ 0.5，且若 α = 0.05，每尾各含 α/2 = 0.025。检验一种新药是否提高了康复率则导致右尾检验，H₁: p > 声称比例。
The direction of the test must be determined by the wording of the claim. Words like 'increase', 'improve', 'more than' point to a right-tailed test; 'decrease', 'less than' suggest a left-tailed test; 'change', 'different', 'not equal' indicate a two-tailed test.
检验的方向必须根据主张的措辞来确定。类似 'increase'、'improve'、'more than' 等词指向右尾检验；'decrease'、'less than' 则暗示左尾检验；'change'、'different'、'not equal' 则表明是双尾检验。
5. Critical Region and Critical Value | 临界区域与临界值
The critical region (or rejection region) is the set of values of the test statistic X that lead to rejection of H₀. The boundary of this region is called the critical value. To find the critical region, you use the binomial distribution under H₀ to identify the range of X for which the cumulative probability just exceeds the significance level in the tail(s).
临界区域（或拒绝域）是导致拒绝 H₀ 的检验统计量 X 的取值集合。该区域的边界值称为临界值。为了找到临界区域，你需要利用 H₀ 下的二项分布，找出使尾部累积概率刚好不超过显著性水平的 X 的范围。
e largest value c such that P(X ≤ c) ≤ 0.05; critical region is X ≤ c. In a two-tailed test at α = 0.05, find two values c₁ and c₂ such that P(X ≤ c₁) ≤ 0.025 and P(X ≥ c₂) ≤ 0.025, and the critical region is X ≤ c₁ or X ≥ c₂.
对于 α = 0.05 的右尾检验，找出使 P(X ≥ c) ≤ 0.05 的最小整数值 c；临界区域为 X ≥ c。对于左尾检验，找出使 P(X ≤ c) ≤ 0.05 的最大整数值 c；临界区域为 X ≤ c。对于 α = 0.05 的双尾检验，找出两个值 c₁ 和 c₂ 使得 P(X ≤ c₁) ≤ 0.025 且 P(X ≥ c₂) ≤ 0.025，临界区域为 X ≤ c₁ 或 X ≥ c₂。
Once the critical region is determined, you simply check whether the observed test statistic falls in that region. If it does, reject H₀; otherwise, do not reject H₀. This method is widely used in exams alongside the p-value approach.
一旦确定临界区域，只需检查观测到的检验统计量是否落在该区域内。如果在，则拒绝 H₀；否则不拒绝 H₀。该方法与 p 值法一起被广泛应用于考试中。
6. p-Value Approach | p 值法
ailed test with observed x₀, the p-value is P(X ≥ x₀). For a left-tailed test, it is P(X ≤ x₀). For a two-tailed test, you usually double the smaller tail probability to get the p-value, reflecting both directions of extremeness.
p 值是在零假设成立的前提下，得到至少与实际观测结果同样极端的检验结果的概率。对于观测值为 x₀ 的右尾检验，p 值是 P(X ≥ x₀)；对于左尾检验，是 P(X ≤ x₀)；对于双尾检验，通常取较小一尾的概率乘以 2 作为 p 值，以反映两个方向的极端性。
If the p-value is less than or equal to the significance level α, you reject H₀. If p-value > α, you do not reject H₀. The smaller the p-value, the stronger the evidence against H₀. Many students find the p-value method more intuitive because it gives a direct measure of how surprising the data are under H₀.
如果 p 值小于或等于显著性水平 α，就拒绝 H₀；如果 p 值 > α，就不拒绝 H₀。p 值越小，反对 H₀ 的证据越强。许多学生觉得 p 值法更直观，因为它直接给出了在 H₀ 下数据令人惊讶的程度。
In Edexcel GCSE questions, you will often be given binomial probability tables or be expected to use a calculator. You can then compare the p-value against the stated α and write a conclusion in context.
在爱德思 GCSE 考题中，通常会提供二项分布概率表，或期望你使用计算器。然后你可以将 p 值与给定的 α 进行比较，并在上下文中写下结论。
7. Steps for Carrying Out a Hypothesis Test | 进行假设检验的步骤
Following a consistent structure is essential to gain full marks on exam questions. Here is a reliable step-by-step guide:
按照一个一致的结构进行是考试拿高分的关键。以下是一个可靠的逐步指南：
Step 1: Define the random variable X and state the null and alternative hypotheses clearly in terms of p.
步骤 1：定义随机变量 X，并用 p 清楚地陈述零假设和备择假设。
Step 2: Note the significance level α and state the distribution of X under H₀, e.g. X ~ B(20, 0.5).
步骤 2：记录显著性水平 α，并陈述 H₀ 下 X 的分布，例如 X ~ B(20, 0.5)。
Step 3: Calculate the test statistic (the observed number of successes) from the given data.
步骤 3：根据给出的数据计算检验统计量（观测到的成功次数）。
Step 4: Find either the critical region or the p-value, depending on which method you prefer.
步骤 4：找出临界区域或计算 p 值，取决于你偏好哪种方法。
Step 5: Compare the test statistic with the critical region, or compare the p-value with α.
步骤 5：将检验统计量与临界区域比较，或将 p 值与 α 比较。
Step 6: Make a decision: reject H₀ or do not reject H₀, and write a conclusion in the context of the original problem. Never say 'accept H₀' – only 'there is insufficient evidence to reject H₀'.
步骤 6：做出决策：拒绝 H₀ 或不拒绝 H₀，并在原始问题背景下写出结论。永远不要说“接受 H₀”——只能说“没有足够证据拒绝 H₀”。
Being methodical not only avoids arithmetic errors but also ensures that your reasoning is clear to the examiner. Practice writing these steps out in full until they become second nature.
有条不紊不仅能避免算术错误，也能确保你的推理过程对考官清晰可见。反复练习完整写出这些步骤，直到它们成为你的第二天性。
8. Worked Example 1: Testing a Proportion (Two-tailed) | 示例 1：检验比例（双尾）
A student suspects a coin is biased. She tosses it 20 times and obtains 15 heads. Test, at the 5% significance level, whether the coin is fair.
一名学生怀疑一枚硬币有偏差。她抛掷该硬币 20 次，得到 15 次正面。在 5% 的显著性水平下，检验这枚硬币是否公平。
H₁: p ≠ 0.5. This is a two-tailed test with α = 0.05, so each tail has 0.025. Under H₀, X ~ B(20, 0.5). The observed test statistic is x = 15.
设 p 为得到正面的概率。假设为：H₀: p = 0.5，H₁: p ≠ 0.5。这是一个双尾检验，α = 0.05，因此每尾占 0.025。在 H₀ 下，X ~ B(20, 0.5)。观测到的检验统计量为 x = 15。
e p-value is P(X ≥ 15) + P(X ≤ 5) = 0.0207 + 0.0207 = 0.0414, which is less than 0.05, confirming rejection. Conclusion: there is sufficient evidence at the 5% level to suggest that the coin is biased.
使用二项分布表，P(X ≤ 5) = 0.0207，P(X ≥ 15) = 0.0207。临界区域为 X ≤ 5 或 X ≥ 15。由于 15 落在临界区域内，我们拒绝 H₀。p 值为 P(X ≥ 15) + P(X ≤ 5) = 0.0207 + 0.0207 = 0.0414，小于 0.05，证实拒绝。结论：在 5% 的显著性水平下，有足够证据表明这枚硬币有偏差。
be in the critical region, and we would not reject H₀. Precise use of boundaries is crucial.
请注意，如果我们得到 14 次正面，P(X ≥ 14) ≈ 0.0577 > 0.025，因此 14 不在临界区域内，我们就不会拒绝 H₀。精确使用边界值至关重要。
9. Worked Example 2: Testing a Claim with Binomial Distribution (One-tailed) | 示例 2：使用二项分布检验声明（单尾）
A manufacturer claims that at most 10% of the light bulbs produced are defective. A quality inspector tests a random sample of 20 bulbs and finds 4 defective. Test at the 5% significance level whether there is evidence that the proportion of defective bulbs exceeds 10%.
一家制造商声称其生产的灯泡中至多有 10% 是次品。一名质检员随机抽取了 20 个灯泡进行检测，发现有 4 个是次品。在 5% 的显著性水平下，检验是否有证据表明次品比例超过 10%。
y a bulb is defective. H₀: p = 0.1, H₁: p > 0.1 (right-tailed test). α = 0.05. Under H₀, X ~ B(20, 0.1). Observed test statistic x = 4.
设 p 为一个灯泡是次品的概率。H₀: p = 0.1，H₁: p > 0.1（右尾检验）。α = 0.05。在 H₀ 下，X ~ B(20, 0.1)。观测统计量 x = 4。
lternatively, find the critical region: P(X ≥ 5) ≈ 0.0432 ≤ 0.05, so critical region is X ≥ 5. Here 4 is not in the critical region.)
我们计算 p 值：P(X ≥ 4) = 1 − P(X ≤ 3)。查表得 P(X ≤ 3) ≈ 0.8670，因此 p 值 ≈ 1 − 0.8670 = 0.1330。由于 0.1330 > 0.05，我们不拒绝 H₀。（或找出临界区域：P(X ≥ 5) ≈ 0.0432 ≤ 0.05，因此临界区域为 X ≥ 5。此处 4 不在临界区域内。）
Conclusion: There is insufficient evidence at the 5% significance level to reject the manufacturer's claim. The observed number of defectives is not significantly higher than expected under the 10% claim.
结论：在 5% 显著性水平下，没有足够证据拒绝制造商的声称。观测到的次品数量并未显著高于 10% 声称下的期望值。
10. Common Mistakes to Avoid | 常见错误要避免
Many marks are lost through small but preventable errors. Here are the most frequent one to watch out for:
许多分数因小却可避免的错误而丢失。以下是最常见的陷阱：
Using the sample proportion as the parameter: Always write hypotheses in terms of p, not p̂ or the sample proportion.
将样本比例当作参数：始终用 p 来写假设，而不是用 p̂ 或样本比例。
Mixing up one-tailed and two-tailed tests: Read the wording carefully. Words like 'changed' or 'different' imply a two-tailed test, whereas 'increased' or 'improved' point to one side.
混淆单尾与双尾检验：仔细阅读措辞。诸如 'changed' 或 'different' 等词暗示双尾检验，而 'increased' 或 'improved' 则指向单侧。
Forgetting to halve the significance level in two-tailed tests: When finding critical values or comparing p-values, remember that a 5% two-tailed test uses 0.025 in each tail.
双尾检验中忘记将显著性水平减半：在找临界值或比较 p 值时，记住 5% 的双尾检验每尾用 0.025。
Stating 'accept H₀': In this framework, you never accept the null hypothesis; you only fail to reject it. Always phrase your conclusion as 'do not reject H₀' or 'insufficient evidence to reject H₀'.
声称“接受 H₀”：在这个框架中，你从不接受零假设；你只是未能拒绝它。结论的措辞永远应是“不拒绝 H₀”或“没有足够证据拒绝 H₀”。
babilities: Ensure you use the correct inequality direction. For X ≥ c use 1 − P(X ≤ c−1).
错误地计算累积概率：确保使用正确的不等号方向。对于 X ≥ c，要用 1 − P(X ≤ c−1)。
By being mindful of these pitfalls, you can turn a good answer into an excellent one.
Published by TutorHao | GCSE 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