📚 Use of Large Sample Results for an Unknown Population | 未知总体的大样本结果
When a population distribution is unknown, it may seem impossible to make reliable inferences about its mean or proportion. However, if the sample size is sufficiently large, the Central Limit Theorem guarantees that the distribution of the sample mean becomes approximately normal, regardless of the shape of the population. This allows statisticians to construct confidence intervals and carry out hypothesis tests without knowing the underlying distribution.
当总体分布未知时，似乎无法对其均值或比例做出可靠推断。然而，只要样本量足够大，中心极限定理保证样本均值的分布近似正态，无论总体形状如何。这使得统计学家能够在不知道总体分布的情况下构造置信区间并进行假设检验。
1. The Central Limit Theorem | 中心极限定理
The Central Limit Theorem states that if X₁, X₂, …, Xₙ are independent and identically distributed random variables with mean μ and variance σ², then for large n the sample mean x̄ is approximately normally distributed with mean μ and variance σ²/n.
中心极限定理指出，若 X₁, X₂, …, Xₙ 是独立同分布的随机变量，均值为 μ，方差为 σ²，则当 n 较大时，样本均值 x̄ 近似服从均值为 μ、方差为 σ²/n 的正态分布。
This result does not require the original population to be normal. Even if the population is skewed, uniform, bimodal, or discrete, the sampling distribution of the mean becomes more and more normal as n increases.
该结果并不要求原始总体服从正态分布。即使总体是偏态的、均匀的、双峰的或离散的，随着 n 增大，样本均值的抽样分布也会越来越接近正态。
The standardised form is approximately standard normal, which is the foundation for large-sample confidence intervals and hypothesis tests.
其标准化形式近似服从标准正态分布，这是大样本置信区间和假设检验的基础。
(x̄ – μ) / (σ/√n) ~ N(0, 1) approximately
2. Why the Population Can Be Unknown | 为什么总体分布可以未知
In many real-world problems, the exact shape of the population distribution is unknown. For example, the distribution of household incomes, waiting times, or exam scores may not be normal, and collecting full population data is often impossible.
在现实世界的许多问题中，总体分布的确切形状是未知的。例如，家庭收入、等待时间或考试分数的分布可能不是正态的，而收集完整的总体数据通常是不可能的。
The Central Limit Theorem provides a way around this difficulty: as long as the sample is random and independent, the mean of a large sample will behave as if it came from a normal distribution, provided the population variance is finite.
中心极限定理提供了解决这一困难的方法：只要样本是随机且独立的，且总体方差有限，大样本的均值就会表现得像来自正态分布一样。
This is why large-sample methods are often called robust or distribution-free with respect to the population shape, even though they still rely on parameters such as μ and σ.
这就是为什么大样本方法通常被称为对总体形状具有稳健性或免分布特性，尽管它们仍然依赖于 μ 和 σ 等参数。
3. When Is a Sample Large Enough? | 样本量多大才算大
A common rule of thumb is that n ≥ 30 is sufficient for the sample mean to be approximately normal, provided the population is not extremely skewed and there are no extreme outliers.
一个常用的经验法则是，只要总体不是极端偏态且没有极端异常值，n ≥ 30 就足以使样本均值近似正态。
If the population is roughly symmetric, the normal approximation can be good even for n as small as 15 or 20. If the population is heavily skewed, a larger sample such as n ≥ 50 or more may be required.
如果总体大致对称，即使 n 小到 15 或 20，正态近似也可能很好。如果总体严重偏斜，则可能需要更大的样本，例如 n ≥ 50 或更多。
For a population proportion, the large-sample condition is usually stated as np ≥ 5 and n(1-p) ≥ 5, or sometimes np ≥ 10 and n(1-p) ≥ 10. These conditions ensure that the binomial distribution is well approximated by a normal distribution.
对于总体比例，大样本条件通常表述为 np ≥ 5 且 n(1-p) ≥ 5，有时也使用 np ≥ 10 且 n(1-p) ≥ 10。这些条件确保二项分布能很好地用正态分布近似。
4. Distribution of the Sample Mean for Large n | 大样本下样本均值的分布
For large n, the sample mean x̄ has the approximate distribution N(μ, σ²/n), where μ is the unknown population mean and σ² is the population variance.
当 n 较大时，样本均值 x̄ 近似服从 N(μ, σ²/n) 分布，其中 μ 是未知的总体均值，σ² 是总体方差。
When the population standard deviation σ is unknown, we replace it with the sample standard deviation s. The estimated standard error of the mean is then s/√n.
当总体标准差 σ 未知时，我们用样本标准差 s 代替。此时均值的估计标准误差为 s/√n。
Although the exact small-sample distribution would be a t-distribution when σ is unknown, for large n the t-distribution is very close to the standard normal distribution, so z-based methods are acceptable.
虽然当 σ 未知时，小样本的精确分布是 t 分布，但当 n 很大时，t 分布非常接近标准正态分布，因此使用基于 z 的方法是可以接受的。
x̄ ~ N(μ, σ²/n) approximately, or x̄ ~ N(μ, s²/n) if σ is unknown
5. Confidence Interval for an Unknown Population Mean | 未知总体均值的置信区间
A confidence interval gives a range of plausible values for the unknown population mean μ. For a large sample, the interval is centred at the sample mean x̄.
置信区间给出了未知总体均值 μ 的合理取值范围。对于大样本，区间以样本均值 x̄ 为中心。
If σ is known, the confidence interval for μ is given by x̄ ± z* × σ/√n, where z* is the critical value from the standard normal distribution for the chosen confidence level.
如果 σ 已知，μ 的置信区间为 x̄ ± z* × σ/√n，其中 z* 是所选置信水平对应的标准正态分布临界值。
If σ is unknown, replace σ with s to obtain x̄ ± z* × s/√n. Common z* values are 1.645 for 90%, 1.96 for 95%, and 2.576 for 99% confidence.
如果 σ 未知，则用 s 代替 σ，得到 x̄ ± z* × s/√n。常用的 z* 值为：90% 置信水平对应 1.645，95% 对应 1.96，99% 对应 2.576。
Confidence interval for μ: x̄ ± z* × σ/√n, or x̄ ± z* × s/√n
6. Confidence Interval for a Population Proportion | 总体比例的置信区间
When estimating a population proportion p, the sample proportion is p̂ = X/n, where X is the number of successes in n trials.
当估计总体比例 p 时，样本比例为 p̂ = X/n，其中 X 是 n 次试验中成功的次数。
For large samples, p̂ is approximately normally distributed with mean p and variance p(1-p)/n, provided the success-failure condition is met.
对于大样本，只要满足成功-失败条件，p̂ 近似服从均值为 p、方差为 p(1-p)/n 的正态分布。
The confidence interval for p is therefore p̂ ± z* × √(p̂(1-p̂)/n). We use p̂ in the standard error because p is unknown.
因此，p 的置信区间为 p̂ ± z* × √(p̂(1-p̂)/n)。我们在标准误差中使用 p̂，因为 p 是未知的。
Confidence interval for p: p̂ ± z* × √(p̂(1-p̂)/n)
7. Hypothesis Test for a Population Mean | 总体均值的假设检验
A large-sample hypothesis test for an unknown population mean begins with a null hypothesis H₀: μ = μ₀ and an alternative H₁: μ ≠ μ₀, μ > μ₀, or μ < μ₀.
对未知总体均值的大样本假设检验从原假设 H₀: μ = μ₀ 开始，备择假设为 H₁: μ ≠ μ₀、μ > μ₀ 或 μ < μ₀。
The test statistic is z = (x̄ – μ₀) / (σ/√n) if σ is known, or z = (x̄ – μ₀) / (s/√n) if σ is unknown. Under H₀, this statistic is approximately standard normal for large n.
如果 σ 已知，检验统计量为 z = (x̄ – μ₀) / (σ/√n)；如果 σ 未知，则为 z = (x̄ – μ₀) / (s/√n)。在 H₀ 下，该统计量在大样本时近似服从标准正态分布。
Compare the calculated z value with the critical value from the normal distribution, or find the p-value and compare it with the significance level α.
将计算出的 z 值与正态分布的临界值进行比较，或求出 p 值并与显著性水平 α 进行比较。
z = (x̄ – μ₀) / (σ/√n) or z = (x̄ – μ₀) / (s/√n)
8. Hypothesis Test for a Population Proportion | 总体比例的假设检验
For a population proportion, the null hypothesis is H₀: p = p₀. The test statistic uses the hypothesised value p₀ in the standard error, not the sample proportion.
对于总体比例，原假设为 H₀: p = p₀。检验统计量在标准误差中使用假设值 p₀，而不是样本比例。
The large-sample test statistic is z = (p̂ – p₀) / √(p₀(1-p₀)/n). This is valid when np₀ ≥ 5 and n(1-p₀) ≥ 5.
大样本检验统计量为 z = (p̂ – p₀) / √(p₀(1-p₀)/n)。当 np₀ ≥ 5 且 n(1-p₀) ≥ 5 时，该统计量有效。
As with the mean test, compare z with the appropriate critical value or use the p-value to decide whether to reject H₀ at significance level α.
与均值检验一样，将 z 与适当的临界值进行比较，或使用 p 值来决定是否在显著性水平 α 下拒绝 H₀。
z = (p̂ – p₀) / √(p₀(1-p₀)/n)
9. P-values and Significance | P值与显著性
The p-value is the probability, assuming H₀ is true, of obtaining a test statistic at least as extreme as the one observed. For large-sample tests, this probability is calculated from the standard normal distribution.
p 值是在 H₀ 为真的前提下，得到至少与观测值一样极端的检验统计量的概率。对于大样本检验，该概率由标准正态分布计算得出。
For a two-tailed test, the p-value is 2 × P(Z > |z|). For a one-tailed test, it is P(Z > z) or P(Z < z) depending on the direction of H₁.
对于双尾检验，p 值为 2 × P(Z > |z|)。对于单尾检验，根据 H₁ 的方向，p 值为 P(Z > z) 或 P(Z < z)。
If the p-value is less than α, the result is statistically significant and we reject H₀. Otherwise, we do not have enough evidence to reject H₀.
如果 p 值小于 α，则结果具有统计显著性，我们拒绝 H₀。否则，我们没有足够的证据拒绝 H₀。
10. Choosing the Sample Size | 样本量的选择
Before collecting data, researchers often want to determine the sample size needed to estimate μ or p with a desired margin of error E and confidence level.
在收集数据之前，研究人员通常希望确定在给定误差幅度 E 和置信水平下所需的样本量，以估计 μ 或 p。
For a mean, the required sample size is n = (z* × σ / E)². If σ is unknown, a pilot study or previous data may be used to estimate it.
对于均值，所需样本量为 n = (z* × σ / E)²。如果 σ 未知，可使用试点研究或以往数据来估计它。
For a proportion, n = z*² × p(1-p) / E². If no prior estimate of p is available, use p = 0.5 because it maximises p(1-p) and gives the most conservative sample size.
对于比例，n = z*² × p(1-p) / E²。如果没有 p 的先验估计，则使用 p = 0.5，因为它使 p(1-p) 最大，并给出最保守的样本量。
11. Conditions and Limitations | 条件与局限性
Large-sample methods require the sample to be random and independent. If the data are biased or clustered, the normal approximation may not be valid even with a very large n.
大样本方法要求样本是随机且独立的。如果数据存在偏差或聚类，那么即使 n 非常大，正态近似也可能不成立。
The Central Limit Theorem also assumes the population variance σ² is finite. Distributions with extremely heavy tails may require a much larger sample before the normal approximation is reliable.
中心极限定理还假设总体方差 σ² 是有限的。尾部极重的分布可能需要更大的样本，正态近似才可靠。
Large-sample results give approximate, not exact, confidence levels and significance levels. When n is only moderately large, the actual coverage probability may differ slightly from the nominal value.
大样本结果给出的是近似的置信水平和显著性水平，而非精确值。当 n 仅为中等大时，实际覆盖率可能与名义值略有差异。
Outliers can distort x̄ and s, so data should be checked for anomalies before applying large-sample formulas.
异常值会扭曲 x̄ 和 s，因此在应用大样本公式之前，应检查数据是否存在异常。
12. Exam Technique and Summary | 考试技巧与总结
In an Edexcel exam question, first identify whether the unknown parameter is a mean μ or a proportion p. Then state the appropriate null and alternative hypotheses clearly.
在 Edexcel 考试题中，首先要确定未知参数是均值 μ 还是比例 p。然后清楚地写出适当的原假设和备择假设。
Check the large-sample conditions before using z-based methods. For a mean, confirm n is large enough or the population is roughly normal. For a proportion, confirm np ≥ 5 and n(1-p) ≥ 5.
在使用基于 z 的方法之前，先检查大样本条件。对于均值，确认 n 足够大或总体大致正态。对于比例，确认 np ≥ 5 且 n(1-p) ≥ 5。
Write down the formula, substitute the values carefully, and give a conclusion in the context of the question: reject H₀ or do not reject H₀.
写出公式，仔细代入数值，并在题目情境中给出结论：拒绝 H₀ 或不拒绝 H₀。
Remember that confidence intervals are used for estimation, while hypothesis tests are used for making decisions. Both rely on the same large-sample normal approximation.
请记住，置信区间用于估计，而假设检验用于做出决策。两者都依赖于相同的大样本正态近似。
Finally, always relate statistical conclusions back to the original problem, using non-technical language where possible.
最后，始终将统计结论与原始问题联系起来，并尽可能使用非技术性语言。
Published by TutorHao | Mathematics Revision Series | aleveler.com
更多咨询请联系16621398022（同微信）
屏轩国际教育cambridge primary/secondary checkpoint, cat4, ukiset,ukcat,igcse,alevel,PAT,STEP,MAT, ibdp,ap,ssat,sat,sat2课程辅导，国外大学本科硕士研究生博士课程论文辅导