IB化学IA实验报告 误差处理 结论写法
IB 化学的内部评估现在官方叫 scientific investigation，学生和老师嘴上还是说 IA。它没有标准答案，24 分全部落在四条评分标准上，写得清不清楚直接决定拿几分。下面按报告的实际写作顺序拆开：研究问题、变量、数据记录、不确定度、作图、结论、评估，每一节都说清评分标准要什么，以及学生实际写成什么样。
What IB examiners look for in a chemistry investigation is not a polished piece of writing but a reproducible argument: a question set in context, data whose precision you can defend, a conclusion that answers the question in numbers, and an evaluation that names real weaknesses rather than generic ones. The notes below follow the order in which the report is drafted, so they double as a checklist.
评分框架：四项标准各六分
按 IBO 官方 Sciences guide（2025 年首次评估）的口径，科学学科的科学探究占最终成绩的 20%，满分 24 分，四条标准各 6 分：Research design、Data analysis、Conclusion、Evaluation。正文上限 3000 词，官方建议投入 10 小时，最多 3 名学生可以共用同一次数据采集，但研究问题、分析和结论必须各自独立。旧大纲的 Personal engagement 与 Communication 两项已经取消。
这条改动的影响很直接：靠实验热情和排版漂亮拿分的空间没有了，24 分全部落在你做了什么、数字说不说得清、结论能不能站住上。很多学生以为 IA 是写作比赛，其实它更像一份别人能照着复现的实验档案。
Under the IBO Sciences guide for first assessment 2025, the investigation is worth 20 per cent of the final grade: 24 marks in four criteria of 6 marks each, a 3000-word limit and about 10 hours of recommended work. Up to three students may share one data collection, provided each writes a unique question, analysis and conclusion. Personal engagement and communication are no longer assessed, so marks now depend on what you measured, how precisely you processed it, and whether the conclusion follows from the numbers.
研究问题必须有上下文
Criterion A 要的不是一句短问题，而是一个带上下文的研究问题。官方对带上下文的解释相当具体：写明自变量与因变量（或两个相互关联的变量）、用一句话交代体系、再给一点直接相关的背景理论。对比两种写法，盐酸浓度对反应速率的影响只算把题目说出来；在 25 °C 下，改变盐酸浓度 0.50 至 2.50 mol dm-3，如何影响 2.0 cm 镁条在 30 秒内产生的气体体积，才把体系、变量范围和测量方式都放进去。
背景理论要直接相关，不需要把整章反应动力学抄一遍。写清楚为什么浓度会改变速率、你预期看到什么关系，两三句就够。常见的错误是搬来一大段课本内容，评分员找不到哪一句在支撑你的假设。
Criterion A rewards a question with context, not a short one. The guide asks for the independent and dependent variables, a concise description of the system, and background theory of direct relevance. Asking how acid concentration affects rate states a topic; naming the 0.50 to 2.50 mol dm-3 range, the 2.0 cm ribbon, the water bath temperature and the gas volume collected in 30 seconds turns it into something measurable. Keep the theory short and pointed, and let it say why concentration should change the rate and what relationship you expect to see.
变量与控制：方法要能复现
方法部分的判据是别人能不能照着做出来。仪器要写规格，比如 100 cm3 量筒、精度 ±0.05 cm3 的滴定管、到 0.1 °C 的温度计；取样要写体积和浓度；控制变量要落到具体动作上，同一水浴恒温 25 °C、同一批镁条切到 2.0 cm、每次读数前等 30 秒。三次平行实验不是形式，它决定你后面能不能算平均值和标准偏差。
安全与废弃物处理也占 Criterion A 的一句位置：浓酸操作、护目镜、废液回收。这句不决定档次高低，但缺了会被记成方法论考虑不完整。
The test for the method section is reproduction. Name the apparatus and its precision, the volumes and concentrations, and hold your variables still in terms a reader can copy: one water bath at 25 °C, ribbon from the same batch cut to 2.0 cm, thirty seconds before each reading. Three trials are not decoration, because they are what allows a mean and a standard deviation later. One line on safety and waste handling belongs here as well, since methodological considerations are part of the criterion.
数据记录：有效数字与仪器精度
Criterion B 里 clear 和 precise 是两个独立要求：clear 指处理过程看得懂，precise 指单位、小数位、有效数字和图表标注都合规。原始数据按仪器精度记录，不要在记录阶段就四舍五入，滴定管读到 24.30 cm3 就写 24.30，分析天平称 0.5120 g 就保留四位。平均值的小数位跟原始数据一致，不要因为计算器多显示了两位就抄上去。
表格要有标题，单位写在表头，量纲前后统一。原始数据超过一页就放附录，正文只留汇总表。这一步做干净，后面算不确定度和作图都省事。
Clear and precise are separate requirements. Clear means the processing can be followed; precise means units, decimal places, significant figures and graph annotation follow convention. Record raw data at the precision of the instrument and do not round while recording: a burette reading of 24.30 cm3 stays 24.30, an analytical balance reading gives 0.5120 g. Means keep the decimal places of the raw values rather than whatever the display shows. Tables need titles and unit rows, and bulky raw data belongs in an appendix with a summary table left in the body.
不确定度：从读数到百分误差
不确定度是最容易丢分也最容易做对的一段。单次读数的不确定度取仪器精度，滴定管要读初读和末读两次，所以一次滴定体积的绝对不确定度是 ±0.10 cm3；百分不确定度等于绝对不确定度除以测量值再乘 100，24.30 ± 0.10 cm3 就是 0.41%。乘除法把百分不确定度相加，加减法把绝对不确定度相加，取对数或平方时按倍数放大。
常见的错误是只在表格里写一个正负号，正文却从没算过百分不确定度，也没在图上画误差棒。评分员看的是你有没有把不确定度用到结论里，最终速率写成 (2.4 ± 0.2) × 10-3 mol dm-3 s-1，而不是光秃秃的 0.0024。
Uncertainties are where marks are most often lost and most easily kept. Take the instrument precision as the reading uncertainty; a burette is read twice, so the titre carries ±0.10 cm3. Percentage uncertainty is the absolute value divided by the measurement, times 100, which makes 24.30 ± 0.10 cm3 a 0.41 per cent uncertainty. Add percentage uncertainties for multiplication and division, absolute ones for addition and subtraction. A signed value sitting in a table, with no percentage uncertainty, no error bars and no uncertainty on the final result, is the pattern that costs marks.
随机误差与系统误差：评估前先分清
评估写不下去，多半是没分清误差的类型。随机误差让重复测量的数据在平均值两侧散开，可以靠多次测量取平均削弱；系统误差让所有数据朝同一个方向偏，取多少次平均都去不掉，只能靠校准仪器、做空白实验或者换方法来发现。高锰酸钾滴定里终点颜色判断偏早属于系统误差，因为每次都会早半滴，很多学生却把它写成随机波动。
另一种常见情形是仪器本身的偏差，例如未校准的移液管每次都多吸 0.05 cm3。判断办法是拿标准样品做一次，看结果是不是稳定偏离文献值，稳定偏离就不是偶然。
Many evaluations stall because random and systematic error are treated as one category. Random error scatters repeated readings around a mean and shrinks as you average more trials; systematic error pushes every reading the same way and never averages out, so it has to be found through calibration, a blank run or a change of method. Judging a permanganate end point half a drop early is systematic, since it happens every time, yet it usually gets written up as random scatter.
图表：散点、趋势线与误差棒
作图不是把数字丢进 Excel 就算完。化学 IA 多用散点加趋势线，变量之间的关系用散点，随时间的变化用折线；每个点配误差棒，误差棒来自上一步的百分不确定度换算回的绝对量；坐标轴写物理量和单位，图的标题交代体系和条件。线性拟合的斜率要带单位，那个斜率往往就是你要求的反应速率常数，或者它的一部分。
R2 接近 1 只说明点贴着直线，不说明模型对。有学生拿 R2 = 0.99 当结论，却没注意低浓度端的点已经弯了，这时候该看残差，或者换一种拟合方式。
Plotting is not dropping numbers into a spreadsheet. Scatter plots suit two variables and line graphs suit change over time; each point should carry an error bar converted from the percentage uncertainty into absolute units, and both axes need a quantity and a unit. Report the gradient of a linear fit with its unit, because that gradient is usually the quantity your question is about. A value of R2 near 1 says the points sit close to the line, not that the model is right: if the low-concentration points curve away, residuals or a different fit are the honest response.
结论：回答研究问题并对比文献
Criterion C 最高档用的词是 justified，IBO 的评分描述要求结论既由你的分析支撑，又和公认的科学背景做比较。实际写法是把结果写成带不确定度的数值，再和文献值对：算百分差，说明差值是否落在你的不确定度范围内，然后解释差在哪。测得的活化能比文献值低，是水浴温度漂了，还是反应已经开始才按的秒表。
很多学生以为结论要写得漂亮，其实一句在 0.50 至 2.50 mol dm-3 范围内速率与盐酸浓度成正比，测得速率常数 k = (2.4 ± 0.2) × 10-3，与文献值 2.6 × 10-3 相差 8%，超出实验不确定度，比三段抒情更有用。
The top band of Criterion C uses the word justified: the conclusion has to follow from your own analysis and be compared with accepted science. Report the result as a number with its uncertainty, work out the difference from the literature value, say whether that difference sits inside your uncertainty, then explain it. If the activation energy comes out below the accepted value, was the bath drifting, or did timing start after the reaction had already begun? A conclusion that answers the question in numbers and names one chemical reason beats three paragraphs of summary.
评估：具体弱点与相对影响
Criterion D 刻意区分 generic 和 specific。写存在人为误差、样本量不够只能进低档；写中和反应进行 4 分钟后体系温度开始下降，说明向环境散热，这会把 ΔT 压低约 0.3 °C，对应的焓变被低估约 6%，才够 describe 甚至 explain 的档次。每个弱点后面要跟一条对得上的改进，加保温层、换发泡杯、缩短读数间隔。
阅卷时最常看到的就是一长串互不相关的缺点，没有一条说明影响有多大。评估的对象是方法，不是结果本身，结果偏了不需要道歉，需要解释它为什么被限制在这个范围内。
Criterion D separates generic from specific. Human error and small sample size sit in the low band; a sentence such as the system began cooling four minutes into the neutralisation, which depressed the temperature rise by about 0.3 °C and the enthalpy change by roughly 6 per cent, reaches the upper bands. Each weakness needs an improvement that addresses it directly: insulation, a foam cup, shorter intervals between readings. What examiners see most often is a list of unrelated flaws with no indication of which one mattered.
时间安排与常见扣分点
官方给的 10 小时里，大部分学生把八小时花在实验室，最后两小时赶报告，结果数据分析和评估同时塌掉。更稳的分法是实验与数据采集 4 小时、处理与作图 3 小时、写作与引用 3 小时。写作顺序从数据往外推：先做表和不确定度，再写结论，最后回头补研究设计，因为漏掉的变量通常在那时候才被发现。
字数也一样，3000 词上限不是目标。设计 500 词、数据 700 词、结论 500 词、评估 700 词，剩下的给引言和文献，通常够用。原始大表放附录，附录不计入字数，没有人会因为附录长而扣分。
Of the recommended ten hours, most students spend eight at the bench and two rushing the report, which is exactly when data analysis and evaluation collapse together. A steadier split is four hours for the practical and data collection, three for processing and graphs, three for writing. Draft outwards from the data: build the table, the uncertainties and the graph first, write the conclusion from them, then repair the research design section, where the missing controlled variable usually surfaces. Treat 3000 words as a ceiling rather than a target, and keep long raw tables in an appendix.
海外直订AQA GCSE Chemistry Student Book AQA GCSE化学学生手册¥588.0📱 去京东购买京东精选
需要一对一辅导？Need One-on-One Tutoring?
IB / A-Level / GCSE 化学与全科辅导，含 IA 报告逐段批改
16621398022 同微信
Visit aleveler.com
或截图后用「扫一扫 → 相册」选择本图