📚 Anomalies in Experimental Physics | 物理实验中的异常值
In A-level Physics, an anomaly is a data point or reading that does not fit the overall pattern of repeated measurements. Identifying and handling anomalies correctly is a core practical skill assessed across AQA papers, especially in required practicals and data-analysis questions. This article explains what anomalies are, why they appear, and how to deal with them scientifically.
在A-level物理中，异常值是指与重复测量整体规律不符的数据点或读数。正确识别和处理异常值是AQA试卷中考查的核心实验技能，尤其在必修实验和数据分析题中。本文将解释什么是异常值、它们为何出现，以及如何科学地处理它们。
1. What Is an Anomaly? | 什么是异常值
An anomaly is an experimental result that lies noticeably outside the range of other repeated readings taken under the same conditions. It is often called an outlier or anomalous result. An anomaly is not simply a result you dislike; it must be judged against the spread of the other data.
异常值是在相同条件下重复测量时，明显落在其他重复读数范围之外的实验结果，通常也称为离群值。异常值并不是你不喜欢的结果，而必须根据其他数据的分散程度来判断。
In a set of repeat measurements of the time period of a pendulum, for example, readings such as 1.21 s, 1.24 s, 1.22 s and 1.97 s contain a clear anomaly at 1.97 s. The anomaly is much further from the cluster than the normal random scatter.
例如，在单摆周期的重复测量中，如果读数为 1.21 s、1.24 s、1.22 s 和 1.97 s，那么 1.97 s 就是一个明显的异常值。它与其他数据簇的偏离程度远远超过正常的随机波动。
2. Random vs Systematic Errors | 随机误差与系统误差
To understand anomalies, you need to distinguish between random errors and systematic errors. Random errors cause readings to scatter above and below the true value with no fixed pattern; they are reduced by taking repeats and averaging. Systematic errors shift all readings in the same direction, often due to an instrument zero error or a flawed method.
要理解异常值，需要先区分随机误差和系统误差。随机误差使读数在真值上下无固定规律地波动，可通过多次重复取平均来减小。系统误差使所有读数朝同一方向偏移，通常由仪器零点误差或方法缺陷引起。
An anomaly is usually the result of a one-off random mistake, such as misreading a scale or a sudden disturbance. It is different from a systematic error because a systematic error affects every result consistently and does not produce a single isolated outlier.
异常值通常是一次性随机失误造成的，例如读错刻度或突然受到干扰。它不同于系统误差，因为系统误差始终影响每一个结果，不会只产生一个孤立的离群值。
3. Identifying Anomalous Results | 识别异常结果
The first step in data analysis is to scan repeated readings for values that are inconsistent with the rest. A simple visual check is often enough when one reading is several instrument divisions away from the cluster. However, you should always use the resolution and uncertainty of the instrument before calling a reading anomalous.
数据分析的第一步是检查重复读数中与其他数据不一致的值。当一个读数与其他数据簇相差好几个仪器分度时，简单的目视检查通常就足够了。但在判定读数异常之前，应始终考虑仪器的分辨率和不确定度。
For example, if a digital voltmeter reads 2.01 V, 2.02 V, 2.00 V and 2.35 V, the 2.35 V reading is likely anomalous because the difference of about 0.33 V is far larger than the instrument resolution of 0.01 V.
例如，如果数字电压表的读数为 2.01 V、2.02 V、2.00 V 和 2.35 V，那么 2.35 V 很可能是异常值，因为约 0.33 V 的差值远大于仪器 0.01 V 的分辨率。
4. Sources of Anomalies | 异常值的来源
Anomalies can come from human mistakes, equipment faults or environmental changes. Common human mistakes include recording the wrong digit, reading the scale from the wrong angle (parallax error), starting and stopping a stopwatch at the wrong moment, or transposing numbers when copying results.
异常值可能来自人为失误、设备故障或环境变化。常见的人为失误包括记录错数字、从错误角度读取刻度（视差误差）、按停秒表的时机不对，或抄写结果时数字颠倒。
Equipment faults may be intermittent, such as a loose connection, a sticking ammeter needle, or a data logger that occasionally fails to trigger. Environmental disturbances, such as a sudden draught, a vibration, or a change in temperature, can also create an isolated anomalous reading.
设备故障可能是间歇性的，例如接触不良、电流表指针卡住，或数据记录器偶尔未能触发。环境干扰（如突然的气流、振动或温度变化）也可能产生孤立的异常读数。
5. Repeating Measurements to Confirm Anomalies | 重复测量确认异常值
If you suspect that a reading is anomalous, the best practical response is to repeat the measurement several more times under the same conditions. Additional repeats reveal whether the suspect value was a random one-off or whether there is a genuine new effect.
如果怀疑某个读数异常，最好的实际操作是再在相同条件下重复测量多次。额外重复可以揭示该可疑值是一次性随机现象，还是存在真正的新效应。
In many AQA required practicals, repeating each measurement three times and taking a mean is standard practice. If one of the three repeats is anomalous and you have time, you should repeat it again rather than automatically discarding it. This gives evidence for excluding the point.
在许多 AQA 必修实验中，每个测量重复三次并取平均是标准做法。如果三次重复中有一个异常值，并且时间允许，应再次重复该测量，而不是自动将其剔除。这可以为排除该点提供依据。
6. Using Statistics to Judge Outliers | 用统计方法判断离群值
A more rigorous way to identify an outlier is to compare the suspect value with the mean and standard deviation of the remaining data. A common rule of thumb is that a reading more than about 2 to 3 standard deviations from the mean is likely anomalous. The standard deviation σ is calculated using the spread of the repeat readings.
更严格的识别离群值方法是将可疑值与其余数据的平均值和标准差进行比较。常用的经验法则是，如果读数偏离平均值超过约 2 到 3 个标准差，就可能是异常值。标准差 σ 根据重复读数的分散程度计算。
The mean x̄ and standard deviation σ can be written as:
平均值 x̄ 和标准差 σ 可表示为：
mean x̄ = (Σxᵢ) / n
σ = √[Σ(xᵢ - x̄)² / (n - 1)]
ere xᵢ are the individual readings and n is the number of readings. A value lying beyond x̄ ± 3σ is generally treated as an outlier.
其中 xᵢ 是各次读数，n 是读数次数。超出 x̄ ± 3σ 的值通常被视为离群值。
ove the upper quartile. This method is less affected by extreme values than the standard deviation.
对于小数据集，四分位距（IQR）也很有用。如果读数低于下四分位数超过 1.5 × IQR，或高于上四分位数超过 1.5 × IQR，则该读数可疑。与标准差相比，该方法受极端值的影响较小。
7. Anomalies on Graphs | 图中的异常点
When plotting a graph, an anomalous point appears as a point that lies well away from the smooth trend line or curve formed by the other points. In A-level Physics, you should plot all data points first, then draw the best-fit line, and only then judge whether a point is anomalous.
绘制图形时，异常点表现为远离其他点所形成的光滑趋势线或曲线的点。在 A-level 物理中，应首先绘制所有数据点，然后画出最佳拟合线，最后再判断某个点是否为异常点。
Do not erase the anomalous point. Instead, circle it or label it 'anomalous' and exclude it from the line of best fit if you have a valid reason. The examiner wants to see that you have recognised the anomaly and handled it transparently.
不要擦除异常点。相反，应将它圈出或标注为 'anomalous'，在有合理理由的情况下将其排除在最佳拟合线之外。考官希望看到你识别了异常值并透明地处理了它。
If an anomaly is excluded, the best-fit line should ignore it, but the point must still be visible on the graph. When calculating a gradient or intercept, use only the points or the line that represent the reliable data.
如果排除了异常点，最佳拟合线应忽略它，但该点仍必须保留在图上。计算斜率或截距时，只使用代表可靠数据的点或拟合线。
8. Handling Anomalies in Calculations | 计算中处理异常值
In data tables and calculations, you should decide whether to include an anomalous reading in the mean. If the anomaly was caused by a clear error, such as a misrecorded value, you can exclude it from the average and state that you have done so. If the cause is unclear, you should keep it and discuss its effect.
在数据表和计算中，需要决定是否将异常读数纳入平均值。如果异常值由明显错误引起，例如记录错误，可从平均值中剔除，并说明已这样做。如果原因不明，则应保留该值并讨论其影响。
Excluding a point without justification is bad scientific practice and can be penalised in exams. You should always write a short explanation, for example: 'The reading 1.97 s was excluded because it was more than three standard deviations from the mean of the other repeats and was likely due to a timing error.'
没有理由就剔除数据点是不良的科学实践，考试中可能被扣分。应始终写一个简短解释，例如：'读数 1.97 s 被排除，因为它与其他重复测量平均值的偏差超过三个标准差，很可能是计时错误造成的。'
When an anomalous value is excluded, the mean and the uncertainty in the mean may change. The mean becomes more representative of the central cluster, and the spread of the remaining readings is smaller. This can reduce the random uncertainty, but you must not hide data to make results look better.
排除异常值后，平均值和平均值的不确定度可能会改变。平均值更能代表中心数据簇，其余读数的分散程度更小。这可以减小随机不确定度，但不能为了美化结果而隐藏数据。
9. Designing Experiments to Minimise Anomalies | 设计实验减少异常值
Good experimental design reduces the chance of anomalies. You can minimise human errors by using digital instruments that record data automatically, by viewing analogue scales at eye level to avoid parallax, and by using a consistent method for starting and stopping timers.
良好的实验设计可以降低异常值出现的概率。可通过使用自动记录数据的数字仪器、眼睛平视模拟刻度以避免视差，以及使用一致的启动和停止计时方法，来减少人为误差。
You can also reduce the impact of environmental disturbances by shielding the apparatus from draughts, vibrations and temperature changes. For example, when measuring the resistivity of a wire, keep the wire straight and avoid moving connections during a set of readings.
还可以通过屏蔽装置免受气流、振动和温度变化的影响，来减少环境干扰。例如，在测量导线电阻率时，保持导线笔直，并在一组读数期间避免移动连接。
Taking more repeats and recording them clearly in a table with units and correct significant figures also helps you spot and justify anomalies quickly. A well-organised results table makes anomalous values stand out rather than hiding them.
进行更多重复测量，并在带单位和正确有效数字的表格中清晰记录，也有助于快速发现和判断异常值。组织良好的结果表能让异常值一目了然，而不是被隐藏起来。
10. Exam Technique: Discussing Anomalies | 考试技巧：讨论异常值
In AQA Physics exam questions, you may be asked to identify an anomaly in a table or on a graph, suggest a cause, or explain how to deal with it. Always refer to the specific data and the instrument resolution in your answer, rather than giving a vague statement.
在 AQA 物理考试题中，可能要求你在表格或图中识别异常值、提出可能原因，或解释如何处理它。答题时应结合具体数据和仪器分辨率，而不是给出模糊的表述。
A strong answer might say: 'The point at 1.97 s is anomalous because it deviates from the trend by about 0.75 s, which is much larger than the 0.01 s resolution of the stopwatch. I would repeat that measurement to check whether it was a timing error before deciding to exclude it.'
一个有力的答案可以这样说：'1.97 s 处的点是异常值，因为它偏离趋势约 0.75 s，远大于秒表 0.01 s 的分辨率。在决定剔除之前，我会重复该测量，以检查
Published by TutorHao | A-Level Physics 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