Techniques for Reversing t-Scores from Probabilities in IB Maths | IB数学：由概率反求t分数的技巧

📚 Techniques for Reversing t-Scores from Probabilities in IB Maths | IB数学：由概率反求t分数的技巧

In IB Mathematics, especially in the Applications and Interpretation HL course, you often need to find a t-score (or t-critical value) when you are given a probability. This “reverse” skill is essential for confidence intervals, hypothesis testing, and constructing margins of error.

在IB数学课程中，尤其是数学应用与解释HL中，你经常需要在已知概率的情况下反求t分数（或t临界值）。这种”反向”技能对于置信区间、假设检验以及构建误差范围至关重要。

---

1. What Is a t-Score? | 什么是t分数

A t-score is a value from the Student’s t-distribution. It is often written as t, and it measures how many standard errors a sample mean is away from a population mean. When you reverse a probability, you are finding the t-value that marks a given area under the distribution curve.

t分数是学生t分布中的一个数值，通常记为t，它衡量样本均值与总体均值之间相差多少个标准误。当你由概率反求时，实际上是在t分布曲线下找到对应给定面积的t值。

t = (x̄ − μ) / (s / √n)

Here x̄ is the sample mean, μ is the population mean, s is the sample standard deviation, and n is the sample size.

其中x̄是样本均值，μ是总体均值，s是样本标准差，n是样本容量。

---

2. Degrees of Freedom | 自由度

The shape of the t-distribution depends on the degrees of freedom, abbreviated df. For a single sample, df = n − 1. As df increases, the t-distribution approaches the standard normal distribution.

t分布的形状取决于自由度，缩写为df。对于单个样本，df = n − 1。随着自由度增大，t分布逐渐接近标准正态分布。

df = n − 1

You must use the correct df when reversing a probability. Using the wrong df will give a completely wrong t-score.

在由概率反求时，必须使用正确的自由度。使用错误的自由度会得到完全错误的t分数。

---

3. One-Tailed and Two-Tailed Probabilities | 单尾与双尾概率

Before reversing, determine whether the probability refers to one tail or two tails. In a one-tailed problem, the total probability lies only on one side of the distribution. In a two-tailed problem, the total probability is split equally between both tails.

在反求之前，请先确定概率涉及单尾还是双尾。在单尾问题中，总概率只位于分布的一侧；在双尾问题中，总概率被等分为左右两个尾端。

Type | 类型	Probability given | 给定概率	Find t such that | 求t满足
One-tailed | 单尾	Area in right tail = α	P(T > t) = α
Two-tailed | 双尾	Total area in both tails = α	P(T > t) = α/2

For example, if α = 0.05 and the test is one-tailed, you solve P(T > t) = 0.05. If the test is two-tailed, you solve P(T > t) = 0.025.

例如，若α = 0.05且为单尾检验，则需解P(T > t) = 0.05；若为双尾检验，则需解P(T > t) = 0.025。

---

4. Reverse Lookup in the t-Distribution Table | 在t分布表中反向查表

A traditional t-table lists critical values for selected probabilities. To reverse, locate the row for your degrees of freedom, then move across to the column that matches the tail probability. The number at the intersection is the t-score.

传统t分布表列出了若干概率对应的临界值。要进行反向查表，先找到对应自由度的行，再横向移动至与尾概率匹配的列，交叉处的数值就是t分数。

In most IB tables, the top row shows “one-tail probability” and also “two-tail probability”. For a one-tailed 5% test with df = 10, the table gives t = 1.812.

在多数IB表中，表头行分别显示”单尾概率”与”双尾概率”。对于df = 10、单尾5%检验，查表得t = 1.812。

P(T > 1.812) = 0.05, for df = 10

---

5. Using a GDC to Reverse t-Scores | 使用图形计算器反求t分数

On a TI-Nspire, go to the Distributions menu, choose “Inverse t”, enter the area (cumulative probability) and df. On a Casio, use the “Inverse t” function in the statistics menu. Always check whether the calculator asks for the cumulative area from the left or the tail area.

在TI-Nspire上，进入分布菜单，选择”Inverse t”，输入面积（累积概率）和自由度。在Casio上，在统计菜单中使用”Inverse t”功能。务必确认计算器要求输入的是左侧累积面积还是尾端面积。

Most calculators define the inverse t with a left-tail cumulative probability. If you want a right-tail probability p, you must enter 1 − p.

大多数计算器定义的t逆函数使用左侧累积概率。如果你需要右侧尾概率p，则必须输入1 − p。

If P(T > t) = 0.05, enter area = 0.95

---

6. Worked Example 1: One-Tailed Reverse | 例题1：单尾反求

A random sample has 15 observations. Find t such that P(T > t) = 0.01.

一个随机样本有15个观测值。求t，使得P(T > t) = 0.01。

Here df = 15 − 1 = 14. We need the right-tail probability 0.01. Using GDC, enter left cumulative area 1 − 0.01 = 0.99, with df = 14.

这里df = 15 − 1 = 14。需要右尾概率0.01。使用计算器时，输入左侧累积面积1 − 0.01 = 0.99，自由度df = 14。

t = 2.624

Interpretation: 1% of the area under the t-curve with df = 14 lies to the right of 2.624.

解释：在df = 14的t曲线下，有1%的面积位于2.624的右侧。

---

7. Worked Example 2: Two-Tailed Reverse | 例题2：双尾反求

For a sample size of 9, find the critical t-value for a 95% confidence interval.

对于样本容量为9的情况，求95%置信区间对应的临界t值。

For a 95% confidence interval, the total tail probability is α = 0.05. In a two-tailed setup, each tail contains 0.025. So we need t such that P(T > t) = 0.025, with df = 9 − 1 = 8.

对于95%置信区间，总尾概率为α = 0.05。在双尾设定下，每个尾端包含0.025。因此需要求满足P(T > t) = 0.025且df = 9 − 1 = 8的t值。

t = 2.306

This value is used when calculating the confidence interval: x̄ ± t × (s/√n).

该值用于计算置信区间：x̄ ± t × (s/√n)。

---

8. Worked Example 3: Combining Probability and Mean | 例题3：结合概率与均值

Suppose a t-distribution has df = 20, and P(X ≤ c) = 0.90. What is c? Here X follows a t-distribution with 20 degrees of freedom.

假设t分布的自由度df = 20，且P(X ≤ c) = 0.90，求c？这里X服从自由度为20的t分布。

Because the probability is already a left-tail cumulative value, we enter 0.90 directly into the inverse t function with df = 20.

因为该概率已经是左侧累积值，所以直接将0.90输入t逆函数，自由度df = 20。

c = 1.325

Check: the t-table with df = 20 and one-tail 0.10 gives 1.325, confirming the result.

验证：t分布表中df = 20、单尾0.10对应1.325，结果一致。

---

9. Common Mistakes | 常见错误

• Using n instead of n − 1 as the degrees of freedom.
• Forgetting to halve the probability in a two-tailed test.
• Entering the tail probability instead of the cumulative probability on a calculator.
• Using the z-score for small samples when the population variance is unknown.

• 使用n而不是n − 1作为自由度。
• 在双尾检验中忘记将概率除以2。
• 在计算器中输入尾概率而不是累积概率。
• 在总体方差未知的小样本中使用z分数。

Always identify the degrees of freedom first, then the direction of the tail, and finally the exact input your calculator expects.

务必先确定自由度，再确定尾的方向，最后确认计算器期望的输入类型。

---

10. t-Score and z-Score: When to Use Which | t分数与z分数：何时使用

Use the z-score when the population standard deviation σ is known and n is large. Use the t-score when σ is unknown and you rely on the sample standard deviation s, especially for small n.

当总体标准差σ已知且n较大时使用z分数；当σ未知、必须依靠样本标准差s时使用t分数，尤其是n较小时。

Situation | 情境	Distribution | 分布
σ known, n large | σ已知，n大	z
σ unknown, n small | σ未知，n小	t
σ unknown, n large | σ未知，n大	t ≈ z

As df increases, the t-score approaches the z-score for the same probability.

随着自由度增大，相同概率下t分数会趋近于z分数。

---

11. Summary of the Reverse Process | 反求过程总结

Step 1: Determine n and calculate df = n − 1. Step 2: Determine whether the probability is one-tailed or two-tailed. Step 3: Adjust the probability to the tail area needed. Step 4: Use the t-table inverse function or GDC to get the t-score. Step 5: Check your answer for reasonableness against a table of common values.

第1步：确定n并计算df = n − 1。第2步：判断概率是单尾还是双尾。第3步：将概率调整为所需的尾面积。第4步：使用t分布表逆函数或计算器得到t分数。第5步：与常见值表对照，检查答案是否合理。

With practice, reversing t-scores becomes a fast and reliable skill for IB exams.

经过练习，由概率反求t分数会成为IB考试中快速而可靠的技能。

Published by TutorHao | Mathematics Revision Series | aleveler.com

更多咨询请联系16621398022（同微信）