Confidence Intervals for the Mean: Calculation and Applications | 均值的置信区间计算与应用

📚 Confidence Intervals for the Mean: Calculation and Applications | 均值的置信区间计算与应用

In statistics, a confidence interval (CI) is a range of values, derived from sample data, that is likely to contain the value of an unknown population parameter. For the population mean μ, a confidence interval provides a measure of uncertainty around the sample mean x̄. This article explains how to construct and interpret confidence intervals for the mean, with formulas and real-world applications.

在统计学中，置信区间（CI）是由样本数据计算出的一个数值范围，它很可能包含未知的总体参数值。对于总体均值 μ，置信区间提供了样本均值 x̄ 周围不确定性的度量。本文将解释如何构建和解读均值的置信区间，并给出公式与实际应用。

1. What is a Confidence Interval? | 什么是置信区间？

A confidence interval is an interval estimate of a population parameter. It is constructed so that, if we repeated the sampling process many times and computed intervals each time, a fixed proportion (the confidence level) of those intervals would contain the true parameter.

置信区间是对总体参数的区间估计。其构造方式为：如果我们重复抽样过程很多次并每次都计算区间，这些区间中有固定比例（即置信水平）会包含真实参数值。

For the mean, the interval is centered at the sample mean x̄ and extends a certain number of standard errors on each side. The length of the interval depends on the confidence level, the sample size, and the variability of the data.

对于均值，区间以样本均值 x̄ 为中心，向两侧延伸一定数量的标准误。区间的长度取决于置信水平、样本容量和数据变异性。

2. Point Estimate vs Interval Estimate | 点估计与区间估计

A point estimate is a single value used to estimate a population parameter. For example, the sample mean x̄ is a point estimate of the population mean μ. Because x̄ varies from sample to sample, a point estimate alone does not convey the precision of the estimate.

点估计是用于估计总体参数的单个值。例如，样本均值 x̄ 是总体均值 μ 的点估计。由于 x̄ 会随样本不同而变化，仅有点估计并不能表达估计的精确度。

An interval estimate, such as a confidence interval, accounts for sampling variability by providing a plausible range of values. A 95% confidence interval is narrower when the sample size is large or when the population is less variable.

区间估计（如置信区间）通过提供一个可能的数值范围来考虑抽样变异性。当样本容量大或总体变异性较小时，95%置信区间会更窄。

3. General Structure of a Confidence Interval for the Mean | 均值置信区间的一般结构

All confidence intervals for the mean share a common structure: a point estimate plus or minus a margin of error. The margin of error is the product of a critical value and the standard error of the sample mean.

所有均值置信区间都遵循相同的结构：点估计值加上或减去误差范围（margin of error）。误差范围是临界值与样本均值标准误的乘积。

Confidence Interval = x̄ ± (critical value) × (standard error)

置信区间 = x̄ ± (临界值) × (标准误)

The standard error of the mean is σ/√n when the population standard deviation σ is known, and s/√n when σ is unknown, where s is the sample standard deviation and n is the sample size.

当总体标准差 σ 已知时，均值标准误为 σ/√n；当 σ 未知时，使用样本标准差 s，标准误为 s/√n，其中 n 是样本容量。

4. Using the Normal Distribution (Known σ) | 使用正态分布（σ 已知）

If the population standard deviation σ is known and the data are approximately normal, or the sample size is large enough (n ≥ 30) by the Central Limit Theorem, the sampling distribution of x̄ is approximately normal. We use the z-distribution with critical value z*.

如果总体标准差 σ 已知，且数据近似正态，或样本容量足够大（n ≥ 30），根据中心极限定理，x̄ 的抽样分布近似正态。此时使用 z 分布以及临界值 z*。

x̄ ± z* × (σ/√n)

For common confidence levels, the critical z-values are:

对于常见置信水平，临界 z 值为：

Confidence Level	z*
90%	1.645
95%	1.960
99%	2.576

In practice, σ is rarely known, but this case is still important because it appears in sample-size planning and in some controlled experiments.

在实际中，σ 很少已知，但这种情况仍然重要，因为它出现在样本量规划和某些受控实验中。

5. Using the t-Distribution (Unknown σ) | 使用 t 分布（σ 未知）

In most applications, σ is unknown and must be estimated by the sample standard deviation s. When s is used, the sampling distribution of the standardised statistic follows a Student’s t-distribution with n − 1 degrees of freedom.

在大多数应用中，σ 未知，需要用样本标准差 s 来估计。当使用 s 时，标准化统计量的抽样分布服从自由度为 n − 1 的 Student’s t 分布。

x̄ ± t* × (s/√n)

The t-distribution is similar to the standard normal distribution but has heavier tails. The critical value t* is found from the t-table or a calculator, depending on the confidence level and the degrees of freedom.

t 分布与标准正态分布相似，但尾部更厚。临界值 t* 根据置信水平和自由度从 t 分布表或计算器中获取。

Because s itself varies between samples, using t instead of z gives wider intervals, reflecting additional uncertainty. As n grows, t* approaches z*, and the two methods become nearly identical.

由于 s 本身随样本变化，使用 t 而不是 z 会给出更宽的区间，以反映额外的不确定性。随着 n 增大，t* 趋近 z*，两种方法变得几乎相同。

6. Degrees of Freedom | 自由度

Degrees of freedom (df) for a one-sample t-interval is n − 1. The df controls the exact shape of the t-distribution. With small df, the distribution has more spread and larger critical values.

单样本 t 区间的自由度（df）为 n − 1。自由度控制 t 分布的确切形状。当自由度较小时，分布更分散，临界值更大。

For example, with a 95% confidence level and 9 degrees of freedom, t* ≈ 2.262, whereas with 30 degrees of freedom, t* ≈ 2.042, and the standard normal value is 1.960.

例如，对于 95% 置信水平和 9 个自由度，t* ≈ 2.262；而 30 个自由度时，t* ≈ 2.042；标准正态值为 1.960。

df = n − 1

When using a statistical calculator or software, always check whether the reported interval is based on t or z, as using the wrong one can lead to incorrect conclusions.

使用统计计算器或软件时，务必检查报告的区间是基于 t 还是 z，因为使用错误会导致错误结论。

7. Margin of Error | 误差范围

The margin of error (ME) is half the width of the confidence interval. It is defined as the critical value multiplied by the standard error of the mean.

误差范围（ME）是置信区间宽度的一半。其定义为临界值乘以均值标准误。

ME = z* × (σ/√n) or ME = t* × (s/√n)

ME = z* × (σ/√n) 或 ME = t* × (s/√n)

Increasing the sample size n reduces the standard error, making the interval narrower. Increasing the confidence level increases the critical value, making the interval wider. Thus there is a trade-off between precision and confidence.

增大样本容量 n 会减小标准误，使区间更窄。提高置信水平会增大临界值，使区间更宽。因此，在精确度和置信度之间存在权衡。

For example, a 99% confidence interval is wider than a 95% confidence interval for the same data.

例如，对于同一组数据，99%置信区间比95%置信区间更宽。

8. Determining Sample Size | 样本容量的确定

Before collecting data, researchers often calculate the sample size needed to achieve a desired margin of error E. When σ is known or a reasonable estimate σ₀ is available, the required sample size is:

在收集数据之前，研究人员通常计算达到期望误差范围 E 所需的样本容量。当 σ 已知或有合理估计值 σ₀ 时，所需样本容量为：

n = (z* × σ / E)²

This formula assumes a z-interval and a large-sample approximation. If using a t-interval, the sample size calculation is slightly more complex and may require iteration.

该公式假设使用 z 区间和大样本近似。如果使用 t 区间，样本量计算稍复杂，可能需要迭代。

To halve the margin of error, the sample size must be quadrupled, because n appears under a square root in the standard error.

若要将误差范围减半，样本容量必须扩大为原来的四倍，因为 n 在标准误中处于根号下。

9. Interpreting Confidence Intervals | 置信区间的正确解读

The correct interpretation of a 95% confidence interval is: if we repeatedly sampled from the same population and constructed intervals in exactly the same way, approximately 95% of those intervals would contain the true mean μ.

95%置信区间的正确解读是：如果我们反复从同一总体中抽样，并以完全相同的方式构造区间，那么大约95%的区间会包含真实均值 μ。

It is not correct to say that “there is a 95% probability that μ lies in this particular interval.” In frequentist statistics, the true mean is fixed, not random. The interval is random because it depends on the sample.

说“μ 落在某个特定区间内的概率为95%”是不正确的。在频率统计学中，真实均值是固定的，而不是随机的。区间是随机的，因为它取决于样本。

95% CI = x̄ ± t* × (s/√n)

Instead, we say: “We are 95% confident that this interval captures μ.” Confidence refers to the method’s long-run success rate, not to any particular interval.

相反，我们说：“我们有95%的置信度认为该区间包含了 μ。”置信度指的是方法的长期成功率，而不是某个特定区间。

10. Applications in Science, Business and Policy | 在科学、商业和政策中的应用

Confidence intervals are widely used to report the precision of estimates from sample data.

置信区间广泛用于报告样本数据估计的精确度。

Clinical trials: Researchers compute confidence intervals for the mean effect of a drug to assess whether the effect is statistically and practically significant.

临床试验：研究人员计算药物平均效应的置信区间，以评估效应是否具有统计显著性和实际意义。
Quality control: Manufacturers use intervals for the mean diameter or weight of a product to ensure that the process remains within specification limits.

质量控制：制造商使用产品平均直径或重量的置信区间，确保生产过程保持在规格限度内。
Public opinion polling: Pollsters report the margin of error for the percentage of voters supporting a candidate, which is a type of confidence interval for a proportion.

民意调查：调查机构报告支持某候选人的投票者百分比的误差范围，这本质上是比例的置信区间。
Economics: Policy analysts use confidence intervals for the mean household income or expenditure to guide government decisions.

经济学：政策分析人员使用家庭平均收入或支出的置信区间来指导政府决策。

In each case, the confidence interval communicates both the best estimate and the uncertainty of that estimate better than a single number alone.

在每种情况下，置信区间都比单一数字更好地传达最佳估计值及其不确定性。

11. Common Misconceptions | 常见误解

Several misconceptions about confidence intervals can lead to misinterpreting results.

关于置信区间的一些常见误解可能导致错误解读结果。

Misconception 1: “The interval contains 95% of the sample data.” Actually, the interval is about the population mean, not about individual data values.

误解1：“该区间包含95%的样本数据。”实际上，该区间是关于总体均值的，而非关于个体数据值。
Misconception 2: “There is a 95% chance that μ is inside this interval.” As explained above, μ is fixed; the probability statement applies to the method, not to the interval.

误解2：“μ 在这个区间内的概率是95%。”如上所述，μ 是固定的；概率陈述适用于方法，而非该区间。
Misconception 3: “Narrower intervals are always better regardless of confidence level.” A narrow interval at 90% confidence may not provide the level of certainty required by the decision-maker.

误解3：“无论置信水平如何，区间越窄越好。”在90%置信度下的窄区间可能无法提供决策者所需的确定程度。
Misconception 4: “Two intervals that overlap mean there is no significant difference.” Overlap does not automatically rule out a statistically significant difference between two means.

误解4：“两个区间重叠意味着没有显著差异。”重叠并不自动排除两个均值之间存在统计显著差异。

Understanding these misconceptions helps students and practitioners use confidence intervals responsibly.

理解这些误解有助于学生和实践者负责任地使用置信区间。

12. Worked Examples | 例题分析

Example 1: Known σ A manufacturer knows that the population standard deviation of screw lengths is σ = 0.2 cm. A random sample of 36 screws has mean x̄ = 4.8 cm. Construct a 95% confidence interval for the population mean screw length.

例1：σ 已知 某制造商知道螺丝长度的总体标准差为 σ = 0.2 cm。随机抽取 36 颗螺丝，样本均值为 x̄ = 4.8 cm。构造总体平均螺丝长度的95%置信区间。

Since σ is known, use z* = 1.960. The standard error is σ/√n = 0.2/√36 = 0.2/6 = 0.0333. The margin of error is 1.960 × 0.0333 ≈ 0.0653.

由于 σ 已知，使用 z* = 1.960。标准误为 σ/√n = 0.2/√36 = 0.2/6 = 0.0333。误差范围为 1.960 × 0.0333 ≈ 0.0653。

95% CI = 4.8 ± 0.0653 = (4.7347, 4.8653) cm

We are 95% confident that the true mean length lies between 4.73 cm and 4.87 cm.

我们有95%的信心认为真实平均长度位于 4.73 cm 到 4.87 cm 之间。

Example 2: Unknown σ Ten students were asked how many hours they study per week. The sample mean is x̄ = 12.5 hours and the sample standard deviation is s = 3.2 hours. Construct a 90% confidence interval for the mean study time.

例2：σ 未知 随机询问10名学生每周学习时长。样本均值为 x̄ = 12.5 小时，样本标准差为 s = 3.2 小时。构造平均学习时间的90%置信区间。

With n = 10, df = 9. For a 90% confidence level, t* ≈ 1.833. The standard error is s/√n = 3.2/√10 ≈ 1.012. The margin of error is 1.833 × 1.012 ≈ 1.855.

由 n = 10，自由度 df = 9。对于90%置信水平，t* ≈ 1.833。标准误为 s/√n = 3.2/√10 ≈ 1.012。误差范围为 1.833 × 1.012 ≈ 1.855。

90% CI = 12.5 ± 1.855 = (10.645, 14.355) hours

Thus the interval is (10.6, 14.4) hours. The wider interval compared with a z-interval reflects the extra uncertainty from estimating σ with s.

因此区间为 (10.6, 14.4) 小时。与 z 区间相比，该区间更宽，反映了用 s 估计 σ 带来的额外不确定性。

In summary, confidence intervals for the mean are essential tools for statistical inference. Whether using the z-distribution with known σ or the t-distribution with unknown σ, the key is to combine a point estimate with a margin of error that reflects sampling variability. Correct interpretation and careful attention to assumptions allow these intervals to support decision-making across science, business, and policy.

总之，均值的置信区间是统计推断中的重要工具。无论是使用已知 σ 的 z 分布，还是使用未知 σ 的 t 分布，关键都是将点估计与反映抽样变异性的误差范围相结合。正确的解读和对假设的仔细关注，使这些区间能够支持科学、商业和政策中的决策。

Published by TutorHao | IB Math Revision Series | aleveler.com

更多咨询请联系16621398022（同微信）

Confidence Intervals for the Mean: Calculation and Applications | 均值的置信区间计算与应用