Confidence Interval for the Mean with Unknown Variance | 未知方差下均值的置信区间

📚 Confidence Interval for the Mean with Unknown Variance | 未知方差下均值的置信区间

When we estimate a population mean from a sample, we rarely know the true population variance. This situation demands a different approach: the t-distribution confidence interval. In IB Mathematics, understanding how to construct a confidence interval for the mean with unknown variance is essential for statistical inference. This article guides you through the logic, formula, and application step by step.

当我们从样本中估计总体均值时，几乎从来不知道真实的总体方差。这种情况需要使用另一种方法：基于 t 分布的置信区间。在 IB 数学中，理解如何在方差未知时构建均值的置信区间是统计推断的核心。本文将逐步带你掌握其背后的逻辑、公式和应用。

1. The Challenge of Unknown Variance | 方差未知带来的挑战

In basic confidence interval for the mean, we assume population variance σ² is known, using the z-interval: x̄ ± z˰ᵠ˰ × (σ/√n). In practice, σ is unknown, so we estimate it with the sample standard deviation s. This introduces extra uncertainty, especially in small samples, because s varies from sample to sample. The z-distribution no longer accurately captures the variability of the estimator.

在基本的均值置信区间中，我们假设总体方差 σ² 已知，使用 z 区间：x̄ ± z˰ᵠ˰ × (σ/√n)。但在实际中，σ 是未知的，因此我们用样本标准差 s 来估计它。这就引入了额外的不确定性，尤其在小样本中，s 随样本不同而变化。z 分布不再能准确刻画估计量的变异性。

---

2. The t-Distribution | t 分布

To account for using s instead of σ, William Gosset derived the t-distribution (Student's t). The t-distribution is symmetric, bell-shaped, but has heavier tails than the normal distribution. The shape depends on the degrees of freedom (df). As df increases, the t-distribution approaches the standard normal distribution.

为了反映用 s 替代 σ 所引入的误差，威廉·戈塞特推导出了 t 分布（学生 t 分布）。t 分布是对称的、钟形，但尾部比正态分布更厚。其形状取决于自由度（df）。随着自由度增加，t 分布趋近于标准正态分布。

---

3. Degrees of Freedom (n-1) | 自由度 (n-1)

n is sample size. We lose one degree of freedom because we estimate the population mean x&# te s. This adjustment ensures the t-distribution correctly reflects the extra uncertainty.

单样本 t 区间的自由度为 df = n - 1，其中 n 是样本容量。因为我们在计算 s 时用同一组数据估计了均值 x̄，所以损失了一个自由度。这一调整确保 t 分布能够正确反映额外的变异性。

---

4. Confidence Interval Formula | 置信区间公式

The confidence interval for the population mean μ (with σ unknown) is given by:

总体均值 μ（σ 未知）的置信区间公式为：

x̄ ± t* × (s / √n)

where t* is the critical value from the t-distribution with degrees of freedom n-1 corresponding to the desired confidence level, and s/√n is the standard error of the mean.

其中 t* 是从自由度为 n-1 的 t 分布中查得的临界值，对应所需的置信水平，s/√n 是均值的标准误。

---

5. Assumptions and Conditions | 假设与条件

For the t-interval to be valid, certain conditions should hold: (1) the sample is randomly selected from the population; (2) the data come from a normally distributed population OR the sample size is large enough for the Central Limit Theorem to apply (commonly n ≥ 30). If the population is strongly skewed and sample size is small, results may be unreliable. Always check for outliers.

要使 t 区间有效，需满足以下条件：(1) 样本是从总体中随机抽取的；(2) 数据来自正态分布的总体，或样本容量足够大使得中心极限定理适用（通常 n ≥ 30）。如果总体严重偏态且样本量小，结果可能不可靠。务必检查异常值。

---

广告 / Sponsored

6. Finding the Critical t-value | 查找临界 t 值

Critical values are found using a t-table or calculator. For a 95% confidence interval, the two-tailed α = 0.05, so we need t_{0.025, df} n=15 (df=14), t* ≈ 2.145. As df grows, t* values get closer to z* (e.g., 1.96 for 95%). The t-table typically lists tail probabilities and degrees of freedom.

临界值可通过 t 分布表或计算器查找。对于 95% 置信区间，双尾 α = 0.05，因此需要 t_{0.025, df}。例如，样本量 n=15（df=14）时，t* 约等于 2.145。随着自由度增大，t* 趋近于 z*（如 95% 的 z*=1.96）。t 表通常列出尾部概率和自由度。

---

7. Worked Example | 计算实例

A random sample of 10 students' test scores: 78, 85, 92, 88, 76, 90, 84, 79, 93, 81. Sample mean x̄ = 84.6, sample standard deviation s = 6.09. Construct a 95% confidence interval for the population mean.

随机抽取 10 名学生的考试分数：78, 85, 92, 88, 76, 90, 84, 79, 93, 81。样本均值 x̄ = 84.6，样本标准差 s = 6.09。构造总体均值的 95% 置信区间。

df = 9, t* for 95% confidence with df=9 is 2.262 (t_{0.025, 9} × 1.926 ≈ 4.36. Confidence interval: 84.6 ± 4.36, i.e. (80.24, 88.96).

df = 9，95% 置信度下自由度为 9 的 t* = 2.262 (t_{0.025, 9})。标准误 = 6.09/√10 ≈ 1.926。误差边际 = 2.262 × 1.926 ≈ 4.36。置信区间：84.6 ± 4.36，即 (80.24, 88.96)。

We are 95% confident that the true mean test score lies between 80.24 and 88.96.

我们有 95% 的把握认为真实的平均考试分数介于 80.24 和 88.96 之间。

---

8. Interpretation of Confidence Level | 置信水平的解释

A 95% confidence level does not mean there is a 95% probability that the specific computed interval contains μ. Instead, if we repeatedly sample and construct intervals in the same way, 95% of those intervals will capture the true population mean. The interval either captures μ or it doesn't—we express confidence in the process.

95% 的置信水平并不意味着计算出的这个特定区间有 95% 的概率包含 μ。正确的解释是：如果我们用同样方法反复抽样并构造区间，其中 95% 的区间将会包含真实的总体均值。对于某一个区间，μ 要么在其中要么不在，我们是对这一方法有信心。

---

9. Factors Affecting Interval Width | 影响区间宽度的因素

The width of the t-interval depends on three main factors: (1) Sample size n: larger n reduces standard error and t* (through df), narrowing the interval. (2) Confidence level: higher confidence (e.g., 99%) produces a larger t*, widening the interval. (3) Sample variability s: greater s increases standard error, making the interval wider.

t 区间的宽度取决于三个主要因素：(1) 样本量 n：较大的 n 减小标准误，并通过自由度减小 t*，使区间变窄。(2) 置信水平：更高的置信度（如 99%）会使 t* 变大，区间变宽。(3) 样本变异性 s：s 增大会使标准误变大，区间变宽。

---

10. Comparison with z-Interval | 与 z 区间的比较

When σ is known and n is large, the z-interval is slightly narrower than the corresponding t-interval because z* < t* for the same confidence level. However, as n increases, t* approaches z*, and the two intervals become nearly identical. In practice, when σ is unknown, always use the t-interval regardless of sample size to stay safe.

当 σ 已知且 n 较大时，z 区间会比相应的 t 区间略窄，因为同一置信水平下 z* < t*。然而，随着 n 增大，t* 趋近于 z*，两种区间几乎相同。在实际应用中，只要 σ 未知，无论样本量大小，都应该使用 t 区间以保证推断的准确性。

---

Published by TutorHao | Mathematics Revision Series | aleveler.com

更多咨询请联系16621398022（同微信）