MYP科学 Criterion C 数据处理 误差分析 详解
在 IB MYP 科学课程中，Criterion C（Processing and evaluating，处理与评价）直接决定一份实验报告能否进入高分档。很多学生的实验设计做得漂亮，却在数据处理这一环丢分：有效数字前后不一致、不确定度完全缺失、图表没有标题和单位、结论只是把数据复述一遍。本文按 MYP 科学的官方评分细则，把从原始数据到有效结论的完整链条拆成八个步骤，配中英对照句式与真实计算示范，读完即可套用到下一次实验报告。
In the IB Middle Years Programme, Criterion C (Processing and evaluating) decides whether a laboratory report reaches the top band. Students often design a solid investigation, then lose marks in processing: inconsistent significant figures, no uncertainty at all, graphs without captions or units, and conclusions that merely repeat the numbers. This guide follows the official MYP Sciences assessment criteria and breaks the chain from raw data to a justified conclusion into eight practical steps, with bilingual sentence patterns and worked calculations you can reuse immediately.
一、Criterion C 的位置与评分梯度
MYP 科学共有四个评价标准：A 知识与理解、B 探究与设计、C 处理与评价、D 反思科学的影响。Criterion C 是唯一直接考察"你会不会算、会不会看数据"的标准，通常占实验报告总分的四分之一左右。评分细则分为四档：0 分表示未达成；1 至 2 分表示能在指导下完成基础的数据呈现；3 至 4 分表示能正确选择并应用数据处理方法，并识别主要误差来源；5 至 6 分要求系统地处理数据、量化不确定度、评价实验方法并给出基于证据的改进建议；7 至 8 分则要求处理过程严谨、结论与数据严格对应，并能批判性地讨论方法局限。理解这个梯度非常关键，它决定了你写报告时的用力方向：从"把数字抄下来"升级为"用数据说话"。
MYP Sciences is assessed against four criteria: A (Knowing and understanding), B (Inquiring and designing), C (Processing and evaluating), and D (Reflecting on the impacts of science). Criterion C is the only strand that directly tests whether you can handle numbers and interpret data, and it usually accounts for about a quarter of the total report marks. The descriptors run in bands: 0 shows no achievement; 1 to 2 marks indicate basic data presentation with guidance; 3 to 4 marks show that you can select and apply appropriate processing methods and identify the main sources of error; 5 to 6 marks require systematic processing, quantified uncertainty, evaluation of the method, and evidence-based improvements; 7 to 8 marks demand rigorous processing, conclusions that match the data exactly, and a critical discussion of the limitations of the method. Understanding this ladder tells you what to upgrade in your writing, from simply recording numbers to using them as evidence.
二、原始数据记录的三条硬规则
所有原始数据必须先以表格形式完整记录，然后才进入处理环节。三条硬规则不容妥协。第一，原始读数不做修饰：仪器显示 25.4 就写 25.4，绝不能提前四舍五入。第二，每个量都要带单位，而单位写在表头而不是每一格，例如"长度 l / cm"。第三，有效数字由仪器精度决定。量筒读数为 25.0 mL，表示精确到 0.1 mL；写成 25 mL 就丢失了精度信息。建议同时记录仪器的最小分度值，例如电子天平 0.01 g、米尺 1 mm，因为它直接决定后面不确定度的大小。此外，重复测量要逐次列出读数，不要只写平均值，平均值本身无法反映数据的离散程度。
Raw data must be recorded completely in a table before any processing begins. Three rules are non-negotiable. First, never polish raw readings: if the instrument shows 25.4, write 25.4 rather than rounding early. Second, every quantity carries a unit, and the unit belongs in the column heading, for example "length l / cm", not in each cell. Third, significant figures are set by the precision of the instrument: a measuring cylinder reading of 25.0 mL communicates precision to 0.1 mL, whereas writing 25 mL throws that information away. Record the smallest division of each instrument as well, such as 0.01 g for an electronic balance or 1 mm for a metre rule, because it fixes the size of the uncertainty you will quote later. When you repeat a measurement, list every individual reading instead of only the mean, since a mean alone hides how scattered the data are.
三、表格与图表的规范写法
表格要有编号与标题，标题描述变量而不是写成"数据表"。第一列放自变量，第二列放因变量，其后依次是导出量。图表方面，MYP 要求连续变量用散点图，分类变量用条形图，两者不能混用。坐标轴必须标注物理量与单位，例如"时间 t / s""温度 T / °C"。作图可用铅笔或软件，Excel、Google Sheets、Logger Pro 都可以，但不要用折线连点，除非数据本身就是连续记录的。趋势线用最小二乘法拟合，并写出方程与决定系数 R²。R² 接近 1 说明线性关系强；如果只有 0.6 左右，就要在评价部分讨论原因，例如读数区间太窄或存在系统误差。图表下方应有一句话说明趋势，让读者不必自己猜。
Tables need a number and a caption that describes the variables rather than the word "data". Put the independent variable in the first column, the dependent variable in the second, and derived quantities afterwards. For graphs, MYP expects scatter plots for continuous variables and bar charts for categorical ones; mixing them loses marks. Both axes must be labelled with the quantity and its unit, for example "time t / s" and "temperature T / deg C". Plot by hand with a sharp pencil or with software such as Excel, Google Sheets, or Logger Pro, and never join the points with a zigzag line unless the data were recorded continuously. Fit a line of best fit by least squares and quote the equation together with the coefficient of determination R squared. A value close to 1 indicates a strong linear relationship; if it is only around 0.6, discuss why in your evaluation, for instance too narrow a range of readings or a systematic error. A one-sentence trend statement under each graph saves the reader from guessing.
四、不确定度与数学处理示范
数据处理的核心，是把重复测量转成"最佳估计值 ± 不确定度"。以测量金属块质量为例，三次读数为 24.98 g、25.02 g、25.00 g：平均值为 25.00 g；极差为 0.04 g，半极差为 0.02 g；仪器分度为 0.01 g。取三者中较大者作为绝对不确定度，写作 25.00 ± 0.02 g，相对不确定度为 0.02 除以 25.00，等于 0.08%。再测体积，三次读数平均为 8.0 cm³，不确定度 0.1 cm³，于是密度为质量除以体积，25.00 除以 8.0 等于 3.1 g/cm³。这里必须注意有效数字规则：乘除运算的结果，其有效数字应与参与运算中最不精确的那个量一致，因此计算器上的 3.125 要写成 3.1。百分误差用实验值与理论值之差的绝对值除以理论值再乘以 100%，例如理论密度 2.70 g/cm³，百分误差约为 15%。这个数字必须在评价部分得到解释，而不能只写一句"实验误差较大"。
The core of processing is turning repeat readings into a best estimate plus an uncertainty. Take the mass of a metal block measured three times as 24.98 g, 25.02 g, and 25.00 g. The mean is 25.00 g, the range is 0.04 g so half the range is 0.02 g, and the instrument division is 0.01 g. Take the largest of these as the absolute uncertainty and write 25.00 plus or minus 0.02 g. The relative uncertainty is 0.02 divided by 25.00, which is 0.08 per cent. Suppose the volume of the same block averages 8.0 cm cubed with an uncertainty of 0.1 cm cubed. Density is mass divided by volume, so 25.00 divided by 8.0 gives 3.1 g per cm cubed. Significant figures matter here: in multiplication and division the answer keeps the same number of significant figures as the least precise input, so 3.125 on your calculator must be rounded to 3.1. Percentage error is the absolute difference between your experimental and theoretical values divided by the theoretical value, multiplied by 100 per cent. If the accepted density is 2.70 g per cm cubed, the error is about 15 per cent, and that figure must be explained in your evaluation rather than dismissed as large experimental error.
五、从数据到结论：趋势描述句式
结论要同时做到三点：有数据、有方向、有力度的量化。比较两种写法。差例："加热之后温度上升了。"好例："在 0 至 60 ℃ 区间内，水温随加热时间呈线性上升，拟合直线斜率约为 4.2 乘以 10 的负 2 次方 ℃/s，R² 为 0.997，说明该区间内加热功率基本恒定。"描述趋势时使用固定句式：在某个范围内，随着自变量增加，因变量以某速率增大或减小；数据支持或不支持假设，因为某个具体证据。如果结果与理论预测不符，不要隐藏，反而要讨论可能原因，例如散热损失导致斜率偏低。最后一定要回应假设：MYP 明确要求说明结果是否支持你的假设，这一句话往往是 7 至 8 分档与 5 至 6 分档的分界线。
A strong conclusion has three qualities: it cites data, states a direction, and quantifies the strength of the evidence. Compare a weak version, "the temperature went up after heating", with a strong one: "between 0 and 60 deg C, the water temperature increased linearly with heating time, with a fitted gradient of about 4.2 times 10 to the power minus 2 deg C per second and an R squared value of 0.997, indicating a nearly constant heating power over this range". Use fixed patterns when describing trends: over the range from one value to another, as the independent variable increases, the dependent variable increases or decreases at a rate of something; the data support or do not support the hypothesis because of a specific piece of evidence. If the results contradict the prediction, do not hide them; discuss the likely reasons, such as heat loss reducing the gradient. Finally, answer the original hypothesis explicitly, because MYP rewards the sentence that states whether the evidence supports it, and that sentence frequently separates the top band from the middle band.
六、六个高频失分点
第一，处理了数据却没有计算任何不确定度，直接掉一个档。第二，图表没有标题、坐标轴没有单位。第三，有效数字前后不一致，表格里写 25.00，结论里写 25。第四，只描述趋势不评价方法，丢掉 Criterion C 后半段"evaluating"的分数。第五，改进建议无法执行，例如"下次更小心"，应改成"把读数区间从 0 至 20 ℃ 扩大到 0 至 80 ℃，并使用数据记录器每秒采样，以降低随机误差"。第六，随手删除异常点却不给理由，正确做法是写明剔除依据，例如该点偏离趋势线超过三个标准差。以上六点几乎覆盖了中等分数段学生的全部丢分原因，逐条对照就能把分数稳定提上来。
Six mistakes appear again and again. First, processing the data without calculating any uncertainty, which drops you a band immediately. Second, graphs with no caption and axes with no units. Third, inconsistent significant figures, for instance 25.00 in the table but 25 in the conclusion. Fourth, describing the trend without evaluating the method, which forfeits the second half of Criterion C. Fifth, improvements that cannot be acted on, such as being more careful next time, instead of extending the range from 0 to 20 deg C to 0 to 80 deg C and logging with a data logger sampling every second to reduce random error. Sixth, deleting an anomalous point without justification; state the criterion for exclusion, for example a deviation of more than three standard deviations from the line of best fit. These six points account for almost every mark lost by students in the middle bands, and working through them one by one lifts the grade steadily.
七、完整范例：弹簧拉伸实验的数据处理
以探究弹簧伸长量与拉力关系为例，演示完整流程。自变量为悬挂砝码质量 m / g，因变量为弹簧长度 L / cm，导出量为伸长量 x 等于 L 减去原长。原长 L0 测得 12.0 cm，挂 50 g 时读数 14.6 cm，100 g 时 17.1 cm，150 g 时 19.7 cm，200 g 时 22.2 cm。每格读数不确定度为 0.1 cm，半极差同为 0.1 cm，故绝对不确定度取 0.1 cm。计算出的伸长量依次为 2.6、5.1、7.7、10.2 cm。用最小二乘法拟合伸长量对质量的直线，斜率约为 0.0508 cm/g，R² 为 0.9997，说明在弹性限度内伸长量与拉力成正比，符合胡克定律。把斜率换算成劲度系数时要注意单位统一，重力加速度取 9.81 N/kg，算得 k 约为 193 N/m。评价部分应指明最大误差来源：砝码标称质量的制造公差，以及弹簧超过弹性限度后读数偏离线性。改进建议包括改用分度更小的游标卡尺测量长度，并把最大悬挂质量控制在弹性限度以内。
Take the classic investigation of how a spring stretches under load. The independent variable is the hanging mass m in grams, the dependent variable is the spring length L in centimetres, and the derived quantity is the extension, equal to the measured length minus the natural length. With an unloaded length of 12.0 cm, the readings are 14.6 cm at 50 g, 17.1 cm at 100 g, 19.7 cm at 150 g, and 22.2 cm at 200 g. Each reading carries an uncertainty of 0.1 cm and half the range is also 0.1 cm, so the absolute uncertainty is 0.1 cm. The calculated extensions are 2.6, 5.1, 7.7, and 10.2 cm. A least squares fit of extension against mass gives a gradient of about 0.0508 cm per gram with an R squared value of 0.9997, confirming that within the elastic limit the extension is proportional to the load, exactly as Hooke's law predicts. Converting that gradient into a spring constant requires consistent units: taking g as 9.81 newtons per kilogram gives a stiffness of about 193 newtons per metre. The evaluation should name the largest sources of error, namely manufacturing tolerance in the nominal masses and deviation from linearity once the spring passes its elastic limit. Reasonable improvements are to measure length with a vernier caliper of smaller division and to keep the maximum load below the elastic limit.
八、把 Criterion C 变成习惯
提升 Criterion C 最有效的方法是建立固定模板。每次实验后按同一顺序完成：原始数据表、导出量表、图表、不确定度计算、趋势描述、与理论值比较、方法评价、改进建议。把这份清单打印出来贴在实验本首页，写作时逐项打勾。此外建议精读一到两份 MYP 官方样本作业，例如 MYP Sciences teacher support material 中的 assessed sample，对比自己与 7 至 8 分样本在评价深度上的差距。多数学生缺的不是计算能力，而是用证据说话的习惯，而习惯是可以通过模板反复训练出来的。下一次实验课，先设计表格再动手，你会发现报告写作时间反而缩短了。
The most efficient way to raise your Criterion C mark is to build a fixed processing template. After every practical, work through the same sequence: raw data table, derived quantity table, graph, uncertainty calculation, trend description, comparison with the theoretical value, evaluation of the method, and suggested improvements. Print that list and tape it inside the front cover of your lab book so you can tick each item as you write. It also helps to read one or two official assessed samples from the MYP Sciences teacher support material and compare the depth of their evaluation with your own. Most students do not lack calculating skills; what they lack is the habit of letting evidence do the talking, and habits can be trained by repetition. On your next practical, design the table before you touch the apparatus, and you will find that writing the report actually takes less time.
DK汽车大百科 升级修订版 儿童科普百科¥372.1📱 去京东购买京东精选需要一对一辅导？Need One-on-One Tutoring?
A-Level / GCSE / IB / MYP 理科与数学全科辅导
或截图后用「扫一扫 → 相册」选择本图