📚 IB Mathematics: Application and Interpretation of t-Tests | IB数学：t检验的应用与解释
The t-test is one of the most widely used statistical tools in IB Mathematics and scientific research. It helps us decide whether the difference between sample means is statistically significant or simply due to random variation. This article explores when and how to apply t-tests, how to interpret their results, and common pitfalls to avoid.
t检验是IB数学和科学研究中应用最广泛的统计工具之一。它帮助我们判断样本均值之间的差异是统计显著的，还是仅仅由随机波动导致。本文探讨何时以及如何应用t检验、如何解读其结果，以及需要避免的常见误区。
1. What Is a t-Test? | 什么是t检验？
A t-test is a hypothesis test that uses the t-distribution to determine whether the mean of a population differs from a specified value, or whether the means of two populations are equal. It is used when the population standard deviation is unknown and the sample size is small, usually less than 30.
t检验是一种利用t分布来判定总体均值是否与某个特定值不同，或两个总体均值是否相等的假设检验。它适用于总体标准差未知且样本量较小（通常小于30）的情形。
The t-distribution resembles the normal distribution but has heavier tails, which accounts for the extra uncertainty introduced by estimating the standard deviation from the sample. As the sample size increases, the t-distribution approaches the standard normal distribution.
t分布形似正态分布，但尾部更重，这反映了用样本估计标准差时所引入的额外不确定性。随着样本量增大，t分布趋近于标准正态分布。
2. Types of t-Tests | t检验的类型
There are three main types of t-tests. The one-sample t-test compares a single sample mean to a known population mean. The independent two-sample t-test compares the means of two unrelated groups. The paired t-test compares two related groups, such as before-and-after measurements on the same subjects.
t检验主要有三种类型。单样本t检验将单个样本均值与已知总体均值进行比较。独立双样本t检验比较两个无关联组的均值。配对t检验比较两个相关组，例如同一对象的前后测量值。
| Type | 类型 | Purpose | 用途 | Example | 示例 |
|---|---|---|
| One-sample | 单样本 | Compare sample mean to known value | Compare class test score to national average |
| Independent two-sample | 独立双样本 | Compare means of two separate groups | Compare test scores of two different teaching methods |
| Paired | 配对 | Compare means of related measurements | Compare blood pressure before and after treatment |
3. Assumptions of t-Tests | t检验的假设条件
Before applying a t-test, certain assumptions must be checked. The data should be continuous and approximately normally distributed. For two-sample tests, the observations within each group must be independent. The variances may be equal or unequal depending on which version of the t-test is used.
在应用t检验之前，必须先检查若干假设条件。数据应为连续型，且近似服从正态分布。对于双样本检验，各组内部的观测值必须相互独立。方差可以相等或不等，具体取决于使用哪种版本的t检验。
Independence means that one observation does not influence another. Normality can be checked using a histogram, a normal probability plot, or a Shapiro-Wilk test. The independent t-test also requires that the two groups are selected independently of each other.
独立性意味着一个观测值不影响另一个观测值。正态性可通过直方图、正态概率图或Shapiro-Wilk检验来检查。独立样本t检验还要求两个组相互独立地选取。
If the sample sizes are large, the t-test is fairly robust to moderate violations of normality due to the Central Limit Theorem. However, outliers and severe skewness can still distort results.
如果样本量较大，根据中心极限定理，t检验对正态性的中等程度偏离具有较好的稳健性。然而，异常值和严重偏斜仍可能使结果失真。
4. Hypothesis Testing Framework | 假设检验框架
Every t-test begins with two hypotheses. The null hypothesis, denoted H₀, states that there is no effect or no difference. The alternative hypothesis, denoted H₁ or Hₐ, states what we expect if there is an effect or difference.
每个t检验都始于两个假设。零假设记为H₀，表示没有效应或没有差异。备择假设记为H₁或Hₐ，表示若存在效应或差异时我们预期的结果。
For a one-sample test, the hypotheses are:
H₀: μ = μ₀
H₁: μ ≠ μ₀ (two-tailed) or μ > μ₀ or μ < μ₀ (one-tailed)
对于单样本检验，假设为：
H₀: μ = μ₀
H₁: μ ≠ μ₀（双尾）或 μ > μ₀ 或 μ < μ₀（单尾）
For an independent two-sample test, the hypotheses are:
H₀: μ₁ = μ₂
H₁: μ₁ ≠ μ₂ (two-tailed) or μ₁ > μ₂ or μ₁ < μ₂ (one-tailed)
对于独立双样本检验，假设为：
H₀: μ₁ = μ₂
H₁: μ₁ ≠ μ₂（双尾）或 μ₁ > μ₂ 或 μ₁ < μ₂（单尾）
5. Calculating the t-Statistic | 计算t统计量
The general formula for a one-sample t-statistic is:
t = (x̄ − μ₀) / (s / √n)
where x̄ is the sample mean, μ₀ is the claimed population mean, s is the sample standard deviation, and n is the sample size.
单样本t统计量的一般公式为：
t = (x̄ − μ₀) / (s / √n)
其中x̄是样本均值，μ₀是声称的总体均值，s是样本标准差，n是样本量。
For an independent two-sample t-test with unequal variances, known as Welch’s t-test, the formula is:
t = (x̄₁ − x̄₂) / √(s₁²/n₁ + s₂²/n₂)
The degrees of freedom for Welch’s test are approximated by a complex formula. IB Mathematics often uses pooled variance when the sample variances are assumed equal.
对于方差不等的独立双样本t检验，即Welch t检验，公式为：
t = (x̄₁ − x̄₂) / √(s₁²/n₁ + s₂²/n₂)
Welch检验的自由度通过一个复杂公式近似计算。IB数学在假设样本方差相等时通常使用合并方差。
The pooled version assumes equal population variances. First compute the pooled variance:
sₚ² = [(n₁ − 1)s₁² + (n₂ − 1)s₂²] / (n₁ + n₂ − 2)
Then the t-statistic is:
t = (x̄₁ − x̄₂) / √(sₚ²/n₁ + sₚ²/n₂)
with degrees of freedom df = n₁ + n₂ − 2.
合并方差版本假设总体方差相等。首先计算合并方差：
sₚ² = [(n₁ − 1)s₁² + (n₂ − 1)s₂²] / (n₁ + n₂ − 2)
然后计算t统计量：
t = (x̄₁ − x̄₂) / √(sₚ²/n₁ + sₚ²/n₂)
自由度为df = n₁ + n₂ − 2。
For a paired t-test, first compute the difference d for each pair, then apply the one-sample formula to the differences:
t = d̄ / (s_d / √n)
where d̄ is the mean difference and s_d is the standard deviation of the differences.
对于配对t检验，先计算每对数据之间的差异d，然后对差异应用单样本公式：
t = d̄ / (s_d / √n)
其中d̄是差异的均值，s_d是差异的标准差。
6. Degrees of Freedom | 自由度
Degrees of freedom represent the amount of independent information available in the data. For a one-sample t-test, df = n − 1. For a paired t-test, df = n − 1, where n is the number of pairs. For an independent pooled t-test, df = n₁ + n₂ − 2.
自由度表示数据中可独立变化的信息量。对于单样本t检验，df = n − 1。对于配对t检验，df = n − 1，其中n是配对数。对于独立样本合并方差t检验，df = n₁ + n₂ − 2。
The degrees of freedom determine the shape of the t-distribution. Smaller df produce heavier tails, meaning that more extreme t-values are needed to reject the null hypothesis. As df increases, the t-distribution approaches the standard normal distribution.
自由度决定t分布的形状。较小的自由度使尾部更厚，意味着需要更极端的t值才能拒绝零假设。随着自由度增大，t分布趋近于标准正态分布。
In formula booklets, the critical t-value is found by matching the desired significance level and the correct degrees of freedom. A larger df leads to a smaller critical value, making it easier to reject H₀ when a true effect exists.
在公式手册中，通过匹配所需的显著性水平和正确的自由度来查找临界t值。自由度越大，临界值越小，这使得在真实效应存在时更容易拒绝H₀。
7. Interpreting the p-Value | 解释p值
The p-value is the probability of obtaining a test statistic at least as extreme as the observed one, assuming the null hypothesis is true. A small p-value suggests that the observed result is unlikely under H₀, providing evidence against the null hypothesis.
p值是在零假设为真的前提下，获得与观测到的检验统计量至少同样极端的结果的概率。较小的p值表明在H₀下观测到的结果不太可能，从而提供了反对零假设的证据。
The significance level, often denoted α, is chosen before the test. Common choices are α = 0.05 and α = 0.01. If p ≤ α, we reject the null hypothesis and call the result statistically significant. If p > α, we fail to reject the null hypothesis.
显著性水平通常记为α，在检验之前选定。常见选择为α = 0.05和α = 0.01。如果p ≤ α，我们拒绝零假设，并称结果在统计上显著。如果p > α，则无法拒绝零假设。
Importantly, failing to reject H₀ does not prove that the null hypothesis is true. It simply means that there is not enough evidence against it. Likewise, rejecting H₀ does not prove that the alternative hypothesis is true with certainty; it only indicates that the data are inconsistent with H₀ at the chosen significance level.
重要的是，未能拒绝H₀并不能证明零假设为真，只表示没有足够证据反对它。同样，拒绝H₀也不能确定地证明备择假设为真，仅表示在所选显著性水平下数据与H₀不一致。
8. One-Tailed vs Two-Tailed Tests | 单尾与双尾检验
A two-tailed test checks for any difference, whether greater or smaller. Its hypotheses are H₀: μ₁ = μ₂ and H₁: μ₁ ≠ μ₂. The p-value is calculated by finding the area in both tails of the t-distribution beyond the absolute value of the observed t-statistic.
双尾检验检查是否存在任何差异，无论更大还是更小。其假设为H₀: μ₁ = μ₂，H₁: μ₁ ≠ μ₂。p值通过计算t分布中超出观测t统计量绝对值的两个尾部的面积得出。
A one-tailed test checks for a difference in a specific direction. For example, H₁: μ₁ > μ₂ or H₁: μ₁ < μ₂. The p-value is the area in only one tail of the distribution. One-tailed tests have more power to detect an effect in the specified direction but are less flexible because they ignore the opposite direction.
单尾检验检查特定方向的差异。例如，H₁: μ₁ > μ₂ 或 H₁: μ₁ < μ₂。p值是分布中仅一个尾部的面积。单尾检验在检测指定方向上的效应时功效更强，但灵活性较差，因为它忽略了相反方向。
In IB Mathematics, two-tailed tests are the default unless the research question clearly specifies a direction. Choosing a one-tailed test after seeing the data is considered inappropriate because it inflates the chance of a false positive.
在IB数学中，除非研究问题明确指定了方向，否则默认使用双尾检验。在看到数据后再选择单尾检验被认为是不恰当的，因为这会增加假阳性的机会。
9. Worked Example | 实例分析
A researcher wants to test whether a new study method improves exam scores. She collects scores from 10 students using the new method. The population mean under the old method is known to be 70. The sample mean is 74.5 and the sample standard deviation is 5.2. Test at the 5% significance level whether the new method changes the score.
一位研究者想检验新的学习方法是否能提高考试成绩。她收集了10名使用新方法学生的成绩。已知旧方法下总体均值为70。样本均值为74.5，样本标准差为5.2。在5%显著性水平下检验新方法是否改变了成绩。
State the hypotheses:
H₀: μ = 70
H₁: μ ≠ 70
写出假设：
H₀: μ = 70
H₁: μ ≠ 70
Compute the test statistic:
t = (74.5 − 70) / (5.2 / √10) = 4.5 / 1.644 ≈ 2.737
计算检验统计量：
t = (74.5 − 70) / (5.2 / √10) = 4.5 / 1.644 ≈ 2.737
The degrees of freedom are df = 10 − 1 = 9. From t-table, the two-tailed critical value at α = 0.05 and df = 9 is 2.262. Since 2.737 > 2.262, we reject H₀. Alternatively, the p-value is approximately 0.023, which is less than 0.05.
自由度为df = 10 − 1 = 9。查t表可知，在α = 0.05和df = 9时双尾临界值为2.262。由于2.737 > 2.262，我们拒绝H₀。另一种方法是，p值约为0.023，小于0.05。
Conclusion: There is sufficient evidence at the 5% significance level that the new study method produces a different mean score from 70. The result is statistically significant.
结论：在5%显著性水平下，有充分证据表明新学习方法产生的平均分与70不同。结果具有统计学显著性。
10. Worked Example: Two-Sample t-Test | 实例分析：双样本t检验
Twenty students are randomly assigned to two teaching methods, A and B. Method A has 10 students with sample mean x̄₁ = 68 and sample standard deviation s₁ = 4. Method B has 10 students with sample mean x̄₂ = 72 and sample standard deviation s₂ = 5. Using pooled variance, test at α = 0.05 whether the two methods differ.
20名学生被随机分配到两种教学方法A和B。方法A有10名学生，样本均值x̄₁ = 68，样本标准差s₁ = 4。方法B有10名学生，样本均值x̄₂ = 72，样本标准差s₂ = 5。使用合并方差，在α = 0.05下检验两种方法是否有差异。
Hypotheses:
H₀: μ₁ = μ₂
H₁: μ₁ ≠ μ₂
假设：
H₀: μ₁ = μ₂
H₁: μ₁ ≠ μ₂
First compute the pooled variance:
sₚ² = [(9)(16) + (9)(25)] / 18 = (144 + 225) / 18 = 369 / 18 = 20.5
首先计算合并方差：
sₚ² = [(9)(16) + (9)(25)] / 18 = (144 + 225) / 18 = 369 / 18 = 20.5
Then compute the t-statistic:
t = (68 − 72) / √(20.5/10 + 20.5/10) = −4 / √4.1 ≈ −4 / 2.025 ≈ −1.975
然后计算t统计量：
t = (68 − 72) / √(20.5/10 + 20.5/10) = −4 / √4.1 ≈ −4 / 2.025 ≈ −1.975
The degrees of freedom are df = 10 + 10 − 2 = 18. The two-tailed critical value at α = 0.05 is 2.101. Since |t| = 1.975 < 2.101, we fail to reject H₀.
自由度为df = 10 + 10 − 2 = 18。α = 0.05时的双尾临界值为2.101。由于|t| = 1.975 < 2.101，我们无法拒绝H₀。
Conclusion: The evidence is not strong enough to conclude that the two teaching methods produce different mean scores at the 5% significance level.
结论：没有足够强的证据表明两种教学方法在5%显著性水平下产生不同的平均成绩。
11. Common Mistakes | 常见错误
One common mistake is using a t-test when the data are not independent. For example, comparing two groups that contain the same individuals at different times should use a paired t-test, not an independent two-sample t-test. Another mistake is ignoring the normality assumption with very small samples.
一个常见错误是在数据不独立时使用t检验。例如，比较不同时间包含相同个体的两组数据应该使用配对t检验，而不是独立双样本t检验。另一个错误是在样本非常小的情况下忽略正态性假设。
Misinterpreting the p-value is also frequent. A p-value is not the probability that the null hypothesis is true. It is the probability of observing data as extreme as the actual data, given that H₀ is true. Thus, p = 0.03 does not mean there is a 3% chance that H₀ is correct.
误解p值也很常见。p值不是零假设为真的概率。它是在H₀为真的情况下，观测到与实际数据一样极端的数据的概率。因此，p = 0.03并不意味着H₀正确的概率为3%。
Another common error is using a one-tailed test without a strong directional hypothesis stated before data collection. Also, selecting a significance level after computing the p-value allows data-driven decisions that weaken the validity of the conclusion.
另一个常见错误是事先没有明确的单尾方向假设就使用单尾检验。此外，在计算出p值之后再选择显著性水平，会允许基于数据的决策，从而削弱结论的有效性。
Students often forget to check whether to use equal or unequal variances. The pooled test is only appropriate when the sample variances are reasonably close. If one variance is much larger than the other, Welch’s t-test is safer.
学生们常常忘记检查应使用等方差还是异方差。合并方差检验仅在样本方差比较接近时适用。如果一个方差远大于另一个，使用Welch t检验更为安全。
12. t-Test vs z-Test vs ANOVA | t检验、z检验与方差分析的比较
A z-test is used when the population standard deviation is known and the sample size is large, typically n ≥ 30. The t-test is more appropriate for small samples with unknown population standard deviation. In practice, the t-test is often used for both small and large samples because it is more conservative.
z检验用于总体标准差已知且样本量较大（通常n ≥ 30）的情形。t检验更适合总体标准差未知的小样本。在实践中，t检验对大样本和小样本都常被使用，因为它更为保守。
ANOVA is used when comparing the means of three or more groups. Using multiple t-tests to compare all pairs increases the risk of committing a Type I error, meaning finding a significant difference by chance. ANOVA controls this overall error rate by testing all group means simultaneously.
方差分析用于比较三个或更多组的均值。使用多次t检验比较所有配对会增加犯第一类错误的风险，即偶然发现显著差异。方差分析通过同时检验所有组均值来控制总体错误率。
| Feature | 特征 | z-Test | t-Test | ANOVA |
|---|---|---|---|
| Number of groups | 组数 | 1 or 2 | 1 or 2 | 3 or more |
| Population σ known | 总体σ已知 | Yes | No | Not required |
| Sample size | 样本量 | Large | Small or large | Per group small or large |
| Typical exam use | 典型考试用途 | A2 statistics | IB AI SL/HL, A-Level | IB AI HL |
13. Using Technology | 使用计算工具
IB Mathematics exams and internal assessments often allow the use of a graphing calculator. Calculators such as the TI-84, Casio fx-CG50, or software such as Desmos and GeoGebra can compute t-statistics and p-values directly from raw data or summary statistics.
IB数学考试和内部评估通常允许使用图形计算器。TI-84、Casio fx-CG50等计算器，以及Desmos和GeoGebra等软件，都可以直接从原始数据或汇总统计量计算t统计量和p值。
When using a calculator, you still need to state the hypotheses, justify your choice of test, and interpret the p-value in the context of the problem. The calculator only provides the numerical result; the reasoning and conclusion must come from you.
使用计算器时，你仍然需要写出假设、证明选择该检验的合理性，并结合问题情境解释p值。计算器只提供数值结果；推理和结论必须由你自己完成。
For IB internal assessments, it is recommended to include a screenshot of the calculator output or a table of key values. However, the write-up should focus on explanation, not merely on pasting outputs. Examiners look for evidence of statistical understanding, not just calculation.
对于IB内部评估，建议包含计算器输出的截图或关键数值的表格。然而，书面报告应注重解释，而不仅仅是粘贴输出。考官看重的是统计理解，而不仅仅是计算。
14. Reporting Results Correctly | 正确报告结果
In IB Mathematics internal assessments, results should be reported in a clear and consistent format. A typical report of a t-test includes the t-statistic, degrees of freedom, the p-value, and a conclusion in context. For example: t(18) = 2.45, p = 0.025, therefore there is a significant difference between the means.
在IB数学内部评估中，结果应以清晰一致的格式呈现。t检验的典型报告包括t统计量、自由度、p值和结合情境的结论。例如：t(18) = 2.45，p = 0.025，因此均值之间存在显著差异。
When reporting confidence intervals, use the sample mean plus or minus the margin of error. The margin of error is t* × s / √n, where t* is the critical t-value for the desired confidence level. Confidence intervals provide a range of plausible values for the population mean.
报告置信区间时，使用样本均值加减误差范围。误差范围为t* × s / √n，其中t*是所需置信水平下的临界t值。置信区间提供了总体均值的合理范围。
Remember that statistical significance does not necessarily imply practical importance. A very large sample can make a small, unimportant difference appear statistically significant. Always evaluate the effect size in the context of the research question.
请记住，统计显著性并不一定意味着实际重要性。非常大的样本可以使微小的、不重要的差异显得统计显著。始终在研究问题的背景下评估效应量。
15. Conclusion | 结论
The t-test is a powerful and versatile tool for comparing means when the population standard deviation is unknown. Understanding its assumptions, types, calculations, and interpretation is essential for IB Mathematics students, particularly for the Applications and Interpretation course where statistics plays a central role.
t检验是总体标准差未知时比较均值的有力而通用的工具。理解其假设、类型、计算和解释，对于IB数学学生至关重要，尤其是在统计占据核心地位的“应用与解释”课程中。
By mastering t-tests, you will be able to draw meaningful conclusions from data, avoid common statistical traps, and communicate results in a scientifically rigorous way. Practice with both raw data and summary statistics, and always interpret results in context.
通过掌握t检验，你将能够从数据中得出有意义的结论，避免常见的统计陷阱，并以科学严谨的方式传达结果。练习使用原始数据和汇总统计量，并始终结合情境解释结果。
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