📚 Investigation 1 - Standard Deviation | 探究一：标准差
In IB Mathematics, the internal assessment often requires you to explore a dataset and use statistical measures to draw meaningful conclusions. Standard deviation is one of the most powerful tools for understanding the spread of data. This investigation will guide you through the concept, calculation, interpretation, and reflection on standard deviation, using a realistic scenario that could form part of your own IA. By the end, you will see how a simple measure can reveal deeper patterns in datasets and why it is essential for comparing groups or testing hypotheses.
在IB数学内部评估中，你经常需要探究一个数据集，并使用统计量来得出有意义的结论。标准差是理解数据分散程度最有力的工具之一。这项探究将引导你掌握标准差的概念、计算、解读与反思，并通过一个真实的情景展示如何将其运用到你的IA中。最终你将发现，这个简单的指标如何揭示数据中更深层的模式，以及为什么它在比较不同组别或检验假设时至关重要。
1. Aim of the Investigation | 探究目标
This investigation aims to examine the variability in student performance on a mathematics test across two different classes. I will calculate and compare the standard deviation of test scores to determine which class demonstrates more consistent achievement. Additionally, I will analyse how sample size and potential outliers affect the reliability of the standard deviation as a measure of spread.
本探究的目标是检验两个不同班级在一次数学测试中成绩的变异性。我将计算并比较测试成绩的标准差，以确定哪个班级的表现更加稳定。此外，我将分析样本容量和潜在的异常值如何影响标准差作为离散程度指标的可靠性。
2. Understanding Standard Deviation | 理解标准差
Standard deviation quantifies how much individual data points deviate from the mean. A low standard deviation indicates that data points are clustered closely around the mean, while a high value signals greater dispersion. It is calculated as the square root of the variance. For a sample, the formula uses n–1 in the denominator to give an unbiased estimate of the population standard deviation. The sample standard deviation is given by:
标准差量化了个体数据点偏离平均值的程度。低标准差意味着数据点紧密聚集在均值周围，高值则表示数据分布更广。它是方差的平方根。对于样本，公式中分母使用n–1，从而得到总体标准差的无偏估计。样本标准差计算公式如下：
s = √[Σ(xᵢ – x̄)² / (n – 1)]
In this formula, xᵢ represents each individual data point, x̄ is the sample mean, and n is the number of observations. Understanding this formula is crucial before using technology, as it reveals how each deviation contributes to the overall spread.
公式中，xᵢ代表每个数据点，x̄是样本均值，n是观测值的个数。在借助计算工具之前理解这个公式至关重要，因为它揭示了每个偏差如何贡献于总体的离散程度。
3. Data Collection | 数据收集
For this investigation, I collected the test scores (out of 50) from two IB Mathematics classes, Class A and Class B, each containing 15 students. The data was gathered from an internal school assessment and anonymised. Ethical considerations were observed by obtaining consent from the teacher and ensuring that no individual could be identified. The scores are presented in the table below.
在本探究中，我收集了两个IB数学班级（A班和B班各15名学生）的测试成绩（满分50分）。数据来自校内评估，并进行了匿名处理。通过征得教师同意并确保无法识别个体身份，我遵守了研究的伦理要求。成绩如下表所示。
| Class A | 42, 38, 45, 40, 36, 44, 39, 41, 37, 43, 40, 38, 42, 44, 39 |
| Class B | 30, 48, 35, 50, 28, 45, 33, 40, 25, 47, 38, 29, 44, 31, 46 |
Both sets appear similar in range at first glance, but standard deviation will reveal the true difference in consistency.
乍看之下，两组数据的范围似乎相近，但标准差将揭示一致性方面的真正差异。
4. Calculating Standard Deviation by Hand | 手工计算标准差
p the mechanics, I first computed the standard deviation manually for Class A. The mean, x̄ 45+40+36+44+39+41+37+43+40+38+42+44+39) / 15 = 40.6. Then, I calculated each deviation (xᵢ – x̄ and summed the squared deviations. The sum of squared deviations was 144.4. Dividing by n–1 = 14 gave a variance of approximately 10.314, and the square root is √10.314 ≈ 3.21. Thus, the standard deviation for Class A is 3.21 marks.
为了完全掌握其运算机制，我首先手动计算了A班的标准差。均值x̄为 (42+38+45+40+36+44+39+41+37+43+40+38+42+44+39)/15 = 40.6。然后，我计算每个偏差(xᵢ – x̄)，平方，再求平方和。平方和等于144.4。除以n–1=14得到方差约为10.314，开平方根得√10.314 ≈ 3.21。因此，A班的标准差为3.21分。
Repeating the process for Class B yielded a mean of 37.6 and a sum of squared deviations of 1106.4, giving a variance of 79.03 and a standard deviation of √79 ghlights the heavy influence of extreme values on variance, as Class B includes scores as low as 25 and as high as 50.
对B班重复此过程，得到均值为37.6，平方和为1106.4，方差为79.03，标准差为√79.03 ≈ 8.89。手工计算突显了极端值对方差的显著影响，因为B班包含低至25分、高至50分的成绩。
5. Using Technology for Calculation | 利用技术计算
hnology ensures accuracy and saves time in an IA context. I used a graphing calculator (TI‑ ed results with Excel. For each dataset, I entered the scores into a list, then used the 1‑ B. Technology also provides additional statistics such as quartiles and the interquartile range, which can complement the investigation.
手工计算有助于理解，但在IA中利用技术可以确保准确性并节省时间。我使用了图形计算器（TI‑84），并通过Excel进行了验证。对于每一组数据，我将成绩输入列表，然后使用1‑Var Stats功能。输出确认了样本标准差：A班sₓ=3.21，B班为8.89。技术工具还提供了四分位数和四分位距等附加统计量，可以补充本探究。
Using the built‑ dentical results. The ease of technology allows you to focus on interpretation rather than arithmetic, but you must always reference the tools used in your IA write‑up.
在Excel中使用内置公式=STDEV.S()也得到了相同的结果。技术带来的便利让你可以专注于解读而非算术，但你必须在IA报告中注明所使用的工具。
6. Interpreting the Results | 解读结果
The standard deviation of 3.21 for Class A suggests that most students scored within about 3.2 marks of the mean of 40.6, indicating a tightly clustered performance. In contrast, Class B's 8.89 standard deviation implies a far wider spread; typical scores deviate from the mean by nearly 9 marks. This means Class A is more homogeneous in achievement, while Class B shows significant disparity between high and low performers.
A班的标准差3.21表明，大多数学生的成绩在均值40.6上下约3.2分之内，显示出高度集中的表现。相反，B班8.89的标准差意味着分布范围宽得多；典型成绩偏离均值近9分。这说明A班的成绩更加整齐，而B班高分段与低分段之间存在巨大差距。
As a rule of thumb, if the data are approximately symmetric, roughly 68% of observations lie within one standard deviation of the mean. Applying this, we expect 68% of Class A scores between 37.4 and 43.8, which is indeed plausible. For Class B, the interval is 28.7 to 46.5, covering almost the entire range, confirming the high variability.
根据经验法则，若数据近似对称，约68%的观测值会落在均值上下一个标准差范围内。据此，我们预期A班68%的成绩在37.4到43.8之间，这确实合理。对于B班，该区间为28.7至46.5，几乎覆盖了整个取值范围，进一步确认了其高变异性。
7. Comparing Data Sets Using Standard Deviation | 使用标准差比较数据集
s that Class B is nearly three times as variable. This difference could reflect teaching methods, student motivation, or the presence of outliers. When reporting in an IA, always contextualise such comparisons to avoid misinterpretation.
只有在均值相近或比较同类单位的变异性时，直接比较标准差才有意义。本探究中两个班级参加的是同一场考试，因此比较有效。标准差比值（8.89/3.21 ≈ 2.77）表明B班的变异性几乎是A班的三倍。这种差异可能反映了教学方法、学生积极性的不同，或异常值的存在。在IA报告中撰写此类比较时，务必结合具体情境，以免误读。
Another approach is to use the coefficient of variation (CV = s/x̄ × 100%). For Class A, CV = 3.21/40.6 × 100% ≈ 7.9%, while for Class B it is 8.89/37.6 × 100% ≈ 23.6%. This reinforces the conclusion that Class B has far greater relative variability.
另一种方法是使用变异系数（CV = s/x̄ × 100%）。A班CV为3.21/40.6×100%≈7.9%，而B班为8.89/37.6×100%≈23.6%。这进一步强化了B班相对变异性大得多的结论。
8. Standard Deviation and the Normal Distribution | 标准差与正态分布
When a dataset follows a bell‑ standard deviation becomes even more powerful. In a perfect normal curve, approximately 68% of data lies within μ ± σ, 95% within μ ± 2σ, and 99.7% my Class A data, a quick check of the histogram shows a roughly symmetric shape, so the 68% ss B, the distribution is more irregular, limiting the practical use of these percentages.
当数据集服从钟形正态分布时，标准差的作用更加显著。在理想正态曲线下，大约68%的数据落在μ±σ内，95%落在μ±2σ内，99.7%落在μ±3σ内。这些经验法则使我们能够确定可能范围并识别异常观测值。在我的A班数据中，快速检查直方图后发现分布大致对称，因此68%规则适用。对于B班，分布更加不规则，限制了这些百分数的实际应用。
If an investigation involves hypothesis testing, such as a t‑test for the difference of two means, the standard deviation is essential for calculating the test statistic. Although not performed here, noting this extension in your IA shows deeper statistical reasoning.
如果探究涉及假设检验，比如双样本均值t检验，标准差对于计算检验统计量至关重要。虽然本探究未进行此操作，但在IA中提及这一延伸可以展现更深的统计推理能力。
9. Impact of Outliers | 异常值的影响
xtreme values because squaring deviations amplifies their effect. In Class B, the score of 25 contributes (25– , while a typical deviation of 3 contributes only 9. Removing this outlier reduces Class B' roximately 7.2, a noticeable drop. This sensitivity means that when outliers are present, it might be wise to also report the interquartile range or to use a trimmed standard deviation. However, outliers should not be removed without justification; they may contain important information about the population.
标准差对极端值敏感，因为偏差平方会放大其影响。在B班，25分这个值造成的平方偏差为(25–37.6)²=158.76，而一个典型偏差3只贡献9。去除该异常值后，B班的标准差从8.89降至约7.2，降幅明显。这种敏感性意味着当存在异常值时，同时报告四分位距或使用截尾标准差可能是明智的。但不应在没有正当理由的情况下删除异常值，因为它们可能包含关于总体的重要信息。
Within an IA, you can perform a sensitivity analysis: calculate standard deviation with and without the outlier, then discuss the implications. This demonstrates critical thinking about the robustness of your conclusions.
在IA中，你可以进行敏感性分析：分别计算包含和不含异常值的标准差，然后讨论其影响。这展示了你对结论稳健性的批判性思维。
10. Reflection and Evaluation | 反思与评估
of the estimates; a larger dataset would yield more stable results. The data were collected from a single school, limiting generalisability. Moreover, the test scores depend on the difficulty of the assessment, so standard deviation might change with different instruments. For a more robust IA, I might consider using stratified sampling or repeated measurements across several tests.
本探究表明，标准差是衡量一致性的有效工具，但也存在局限性。小样本容量（n=15）降低了估计的可靠性；更大的数据集会产生更稳定的结果。数据仅来自一所学校，限制了结论的普遍性。此外，测试成绩取决于评估工具本身的难度，因此标准差可能会因不同试卷而变化。为得到更可靠的IA，我可能会考虑采用分层抽样或在多次测试中重复测量。
Another point is that standard deviation assumes the mean is a good measure of centre, but for skewed data, the median and interquartile range might be more appropriate. Combining several measures avoids overdependence on a single statistic.
另一点是，标准差假设均值是一个好的中心度量，但对于偏态数据，中位数和四分位距可能更合适。结合多种统计量可以避免过度依赖单一指标。
11. Conclusion | 结论
9). The investigation confirmed that standard deviation is a straightforward yet deeply informative statistic, particularly when paired with graphical displays and other measures. Through manual calculation and technology use, I gained a thorough understanding of its computation and meaning. The process also highlighted the importance of considering sample size, outliers, and data distribution when drawing inferences. Standard deviation will remain a foundational tool in any data‑driven IA exploration.
总之，标准差清晰地表明，A班测试成绩的一致性（s=3.21）远高于B班（s=8.89）。这项探究证实，标准差是一个直观又极具信息量的统计量，尤其在与图形展示和其他度量结合使用时。通过手工计算和应用技术工具，我深刻理解了它的运算方式和意义。探究过程也凸显出在推论时考虑样本大小、异常值和数据分布的重要性。在任何数据驱动的IA探索中，标准差都将是一个基础性工具。
12. Suggestions for Further Investigation | 进一步探究建议
To extend this work, one could examine whether the observed difference in standard deviations is statistically significant using an F‑test for equality of variances. Alternatively, investigating the relationship between study hours and test scores within each class might explain the variability. A longitudinal study tracking the same students over multiple assessments could reveal whether the standard deviation stabilises or changes over time. These extensions would deepen the statistical analysis and strengthen the overall quality of the IA.
要扩展本探究，可以使用F检验（方差齐性检验）来检验所观察到的标准差差异是否统计显著。另一种方向是研究每个班级内学习时长与测试成绩的关系，或许能解释变异性的来源。一项跟踪同一批学生多次评估的纵向研究则可以揭示标准差是趋于稳定还是随时间变化。这些延伸将深化统计分析，并提升IA的整体质量。
Published by TutorHao | Mathematics Revision Series | aleveler.com
更多咨询请联系16621398022（同微信）
或截图后用「扫一扫 → 相册」选择本图