📚 Confidence Intervals | 置信区间
In A-Level Statistics, we often use sample data to estimate population parameters such as the mean or proportion. A point estimate gives a single value, but it does not communicate the uncertainty inherent in sampling. A confidence interval provides a range of plausible values for the parameter, together with a degree of confidence. This article explains the construction, interpretation and common pitfalls of confidence intervals as required for the Edexcel Mathematics specification.
在A-Level统计中，我们经常利用样本数据来估计总体参数，如均值或比例。点估计只给出一个单一数值，却无法传达抽样本身的不确定性。置信区间则提供了一个包含参数可信值的范围，同时附上了置信程度。本文讲解置信区间的构建、解读以及常见误区，完全符合Edexcel数学考试大纲的要求。
1. What is a Confidence Interval? | 什么是置信区间？
A confidence interval is an interval estimate derived from sample data that is likely to contain the unknown population parameter. It is centred on the sample statistic and extends equally in both directions by a margin of error. The width of the interval reflects the precision of the estimate: narrower intervals indicate more precise estimates.
置信区间是由样本数据得出的一个区间估计，它很可能包含未知的总体参数。区间以样本统计量为中心，并向两侧延伸一个误差范围。区间的宽度反映了估计的精确度：越窄的区间代表越精确的估计。
For example, if we calculate a 95% confidence interval for a population mean as (48.2, 51.8), we are stating that we are 95% confident that the true mean lies somewhere between 48.2 and 51.8. The confidence level quantifies how sure we are that the interval captures the parameter.
例如，若我们算出总体均值的95%置信区间为(48.2, 51.8)，表明我们有95%的信心认为真实的均值落在48.2到51.8之间。置信水平量化了我们对该区间包含参数的把握程度。
2. Confidence Level and Significance Level | 置信水平和显著性水平
ed by 1 - α, is the probability that the interval will contain the parameter under repeated sampling. Common confidence levels are 90%, 95% and 99% .01 respectively. The significance level α is the probability that the interval does not contain the parameter. In Edexcel exams, you must be comfortable converting between a confidence level and a z- or t-critical value.
置信水平通常记为1 - α，表示在重复抽样下，置信区间会包含参数的概率。常见的置信水平有90%、95%和99%，对应的α值分别为0.10、0.05和0.01。显著性水平α是区间不包含参数的概率。在爱德思考试中，你需要熟练掌握置信水平与z临界值或t临界值之间的转换。
The most frequent critical values from the standard normal distribution are:
| Confidence level | α | z-critical value (two-tailed) |
|---|---|---|
| 90% | 0.10 | 1.645 |
| 95% | 0.05 | 1.960 |
| 99% | 0.01 | 2.576 |
标准正态分布最常用的临界值如上表所示。
3. Confidence Interval for Population Mean (Known Variance) | 总体均值的置信区间（方差已知）
When the population variance σ² is known and either the population is normal or the sample size n is large (by the Central Limit Theorem), we use the standard normal distribution. The confidence interval for the population mean μ is given by:
x̄ ± z × (σ/√n)
Here x̄ is the sample mean, σ is the population standard deviation, n is the sample size and z is the critical value from N(0,1) corresponding to the chosen confidence level.
当总体方差σ²已知，且总体服从正态或样本量n足够大（根据中心极限定理），我们使用标准正态分布。总体均值μ的置信区间为：
x̄ ± z × (σ/√n)
其中x̄为样本均值，σ为总体标准差，n为样本量，z为对应置信水平的标准正态分布临界值。
mean lifetime of 1020 hours. Suppose the population standard deviation is 100 hours. A 95% confidence interval for the mean lifetime is 1020 ± 1.96×(100/√50) = 1020 ± 27.7, giving (992.3, 1047.7) hours.
例如，一个容量为50的灯泡样本的平均寿命为1020小时，已知总体标准差为100小时。总体均值95%置信区间为1020 ± 1.96×(100/√50) = 1020 ± 27.7，即(992.3, 1047.7)小时。
4. Confidence Interval for Population Mean (Unknown Variance) – t-distribution | 总体均值的置信区间（方差未知，t分布）
More commonly, the population variance is unknown and must be estimated by the sample variance s². In this case, the sampling distribution of the sample mean follows a t-distribution with n - 1 degrees of freedom, provided the population is approximately normal. The formula becomes:
x̄ ± tn-1 × (s/√n)
更常见的情况是总体方差未知，需要由样本方差s²来估计。此时若总体近似正态，样本均值的抽样分布服从自由度为n - 1的t分布。公式变为：
x̄ ± tn-1 × (s/√n)
The t-distribution is wider than the normal distribution for small sample sizes, reflecting extra uncertainty. As n increases, the t-distribution approaches the standard normal. In Edexcel questions, you will often be given a table of t-critical values or be required to use the provided percentage points.
对于小样本，t分布比正态分布更扁平、尾部更厚，反映了额外的不确定性。随着n增大，t分布趋近于标准正态。在爱德思试题中，通常会给出t临界值表或要求使用提供的百分位数。
Suppose eight measurements of the concentration of a solution yield x̄ = 4.8 mg/L and s = 0.25 mg/L. For a 99% f freedom, the t-critical value is approximately 3.499. The interval is 4.8 ± 3.499×(0.25/√8) = 4.8 ± 0.309, giving (4.491, 5.109) mg/L.
假设对某溶液浓度进行8次测量，得x̄ = 4.8 mg/L，s = 0.25 mg/L。对于自由度为7的99%置信区间，t临界值约为3.499。区间为4.8 ± 3.499×(0.25/√8) = 4.8 ± 0.309，即(4.491, 5.109) mg/L。
5. Confidence Interval for a Population Proportion | 总体比例的置信区间
When estimating a population proportion p, we use the sample proportion p̂ ber of successes in n independent trials. Provided the sample is large enough such that np̂ > 5 and n(1 - p̂) > 5, the sampling distribution of p̂ is approximately normal with mean p and standard error √[p̂(1 - p̂)/n]. The confidence interval is:
p̂ ± z × √[ p̂(1 - p̂) / n ]
在估计总体比例p时，我们使用样本比例p̂ = x/n，其中x为n次独立试验中的成功次数。只要样本足够大，使得np̂ > 5且n(1 - p̂) > 5，p̂的抽样分布就近似正态，均值为p，标准误为√[p̂(1 - p̂)/n]。置信区间为：
p̂ ± z × √[ p̂(1 - p̂) / n ]
For example, in a survey of 200 voters, 112 support a candidate. Then p̂ = 112/200 = 0.56. A 95% confidence interval for the true proportion is 0.56 ± 1.96 × √0.56×0.44/200 = 0.56 ± 0.0688, so the interval is (0.491, 0.629). We would report that we are 95% confident the true support lies between 49.1% and 62.9%.
例如，一项对200名选民的调查中，112人支持某候选人。则p̂ = 112/200 = 0.56。真实支持比例的95%置信区间为0.56 ± 1.96 × √0.56×0.44/200 = 0.56 ± 0.0688，区间为(0.491, 0.629)。我们会报告有95%的信心认为真实支持率在49.1%到62.9%之间。
6. Margin of Error | 误差范围
int estimate to form the confidence interval. For a mean with known variance it is z × (σ/√n); for an unknown variance it is t × (s/√n); for a proportion it is z × √[p̂(1 - p̂ of error depends on the confidence level, the variability of the data and the sample size.
误差范围是在点估计基础上加减以形成置信区间的量。对于方差已知的均值，它是z × (σ/√n)；对于方差未知的情况，它是t × (s/√n)；对于比例，它是z × √[p̂(1 - p̂)/n]。误差范围依赖于置信水平、数据的变异程度和样本量。
Increasing the confidence level widens the interval, meaning we sacrifice precision for greater certainty. Increasing the sample size reduces the standard error, narrowing the interval and giving a more precise estimate. Understanding this relationship helps you explain why larger samples lead to more reliable conclusions.
提高置信水平会拉宽区间，意味着我们牺牲精确度以换取更大的把握。增大样本量会减小标准误，从而缩窄区间，得到更精确的估计。理解这种关系有助于你解释为什么更大的样本能得出更可靠的结论。
7. Determining Sample Size | 确定样本量
When designing a study, researchers often decide the required sample size to achieve a desired margin of error E at a given confidence level. For a population mean (known σ), the necessary sample size is:
n = (z × σ / E)²
在规划一项研究时，研究者常常需要决定，要在给定的置信水平下达到指定的误差范围E，所需的最小样本量。对于总体均值（σ已知），所需样本量为：
n = (z × σ / E)²
p is available, we use p = 0.5 to maximise the variance, giving the conservative formula:
n = (z² × 0.25) / E²
对于比例的情况，若没有p的先验估计，我们使用p = 0.5以最大化方差，得到保守公式：
n = (z² × 0.25) / E²
/ E². Always round the calculated n up to the next integer. An Edexcel question might ask: 'Find the minimum sample size needed to estimate a proportion to within 3%, with 95% confidence.' p=0.5 gives n = (1.96² × 0.25) / 0.03² ≈ 1067.1, so at least 1068 respondents are needed.
如果有一个先验估计p*，则使用n = (z² × p*(1-p*)) / E²。计算出的n总是要向上取整。爱德思考试可能会问：“要估计比例使其误差不超过3%，置信水平95%，求最小样本量。” 使用p=0.5得到n = (1.96² × 0.25) / 0.03² ≈ 1067.1，因此至少需要1068名受访者。
8. Interpreting Confidence Intervals Correctly | 置信区间的正确解读
One of the most common exam pitfalls is misinterpreting a confidence interval. A 95% confidence interval does not mean there is a 95% probability that the population parameter lies in that specific interval. The parameter is a fixed but unknown constant. Instead, the confidence level refers to the long-run success rate of the method: if we repeated the sampling procedure many times, about 95% of the intervals constructed would contain the true parameter.
考试中最常见的陷阱之一就是误读置信区间。一个95%置信区间并不意味着总体参数落在该具体区间内的概率为95%。参数是一个固定但未知的常数。相反，置信水平指的是这种方法的长期成功率：如果多次重复抽样过程，大约95%构造出来的区间会包含真实参数。
Correct wording: 'We are 95% confident that the interval (48.2, 51.8) captures the population mean.' Incorrect: 'The probability that the mean is between 48.2 and 51.8 is 0.95.' Exam mark schemes are strict on this distinction.
正确表述：“我们有95%的信心认为区间(48.2, 51.8)包含了总体均值。”错误表述：“均值落在48.2到51.8之间的概率是0.95。” 考试评分方案对这一区别要求严格。
9. Confidence Intervals and Hypothesis Testing | 置信区间与假设检验
Confidence intervals are deeply connected to two-tailed hypothesis tests. For a significance level α, a 100(1-α)% confidence interval can be used to perform a two-tailed test: if the hypothesised parameter value lies inside the confidence interval, we do not reject H₀; if it lies outside, we reject H₀. This link offers an alternative way to draw conclusions, and many Edexcel questions will expect you to use it.
置信区间与双侧假设检验有着深刻的联系。在显著性水平α下，一个100(1-α)%的置信区间可用于执行双侧检验：如果假设的参数值落在置信区间内，不拒绝原假设H₀；如果落在区间外，则拒绝H₀。这一联系提供了一个替代的推断方法，许多爱德思考题都会要求你运用它。
For instance, a 95% , since 23 is not within the interval, we reject H₀. If the hypothesised value were 20, it would be inside, so we would not reject. This visual interpretation can be very powerful.
例如，一个均值的95%置信区间为(18.2, 22.5)。检验H₀: μ = 23，显著性水平α = 0.05，由于23不在区间内，我们拒绝H₀。若假设值为20，它在区间内，则不拒绝原假设。这种可视化解读非常强大。
10. Summary and Exam Tips | 总结与考试技巧
Confidence intervals are a fundamental part of inferential statistics. Remember to:
- Identify whether the variance is known or unknown, to decide between z and t procedures.
- Check conditions for normality (sample size or population distribution) and for a proportion the success-failure condition np̂ > 5, n(1-p̂) > 5.
- Use the correct critical value from the normal or t-table, and if using a t-distribution, state the degrees of freedom.
- Interpret the interval in context, avoiding probabilistic language about the parameter.
- When calculating sample size, always round up.
置信区间是推断统计的核心内容之一。牢记以下要点：
- 判断方差已知还是未知，从而决定使用z分布还是t分布。
- 检查正态性条件（样本量大小或总体分布），对于比例则要满足成功-失败条件np̂ > 5, n(1-p̂) > 5。
- 使用正确的正态或t分布临界值，若使用t分布，须注明自由度。
- 结合背景解读区间，要避免使用关于参数的概率性语言。
- 计算样本量时，一定要向上取整。
In the Edexcel exam, you may be asked to construct an interval, interpret it, compare intervals, or relate it to a hypothesis test. Practising with past papers will build your confidence – no pun intended – in handling these questions quickly and accurately.
在爱德思考试中，你可能会被要求构建区间、解读区间、比较区间或者与假设检验联系起来。通过练习历年试题，你可以熟练并准确地解答这些题目，并对自己的实力充满“信心”。
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