CAIE IGCSE统计实践考核要点与评分标准 | CAIE IGCSE Statistics: Practical Assessment Key Points and Mark Scheme

CAIE IGCSE统计实践考核要点与评分标准

一、CAIE IGCSE统计实践考核考什么：试卷结构与评分目标 / What the Practical Assessment Covers: Paper Structure and Assessment Objectives

CAIE 的 IGCSE 统计课程（代码 0479）自 2025 年起启用修订版大纲，全部考生须在同一次考试中完成两份笔试。Paper 1 时长 2 小时 15 分，满分 100 分，占总分 50%，题型为长短结合的问答，其中最后一道大题通常就是数据分析任务，提供原始数据集或现成图表，要求考生完成探索、清洗与解读。Paper 2 同样是 2 小时 15 分、100 分、占 50%，侧重综合问题解决，并包含一个调查场景，要求写出包含假设、方法、分析与结论的迷你报告。评分目标分为两块：AO1 统计技术的知识与理解占 80% 到 90%，AO2 解读、沟通与问题解决占 10% 到 20%，实践类题目重点考 AO2，会算不算本事，会解释才拿分。

The CAIE IGCSE Statistics course, syllabus code 0479, has followed a revised specification since 2025, and every candidate must sit two written papers in the same examination series. Paper 1 lasts 2 hours 15 minutes, carries 100 marks and is worth 50 percent, with questions of mixed lengths; the final question is usually a data analysis task that supplies a raw data set or ready-made charts and asks candidates to explore, clean and interpret the material. Paper 2 is also 2 hours 15 minutes with 100 marks and 50 percent weighting, focusing on extended problem solving and including an investigation scenario that requires a mini-report covering hypothesis, method, analysis and conclusion. Assessment objectives are split into AO1, knowledge and understanding of statistical techniques at 80 to 90 percent, and AO2, interpretation, communication and problem solving at 10 to 20 percent, which means practical questions target AO2: calculating alone is not enough, explaining earns the marks.

二、统计探究循环：实践考核的主线思维 / The Statistical Enquiry Cycle: The Core Thinking of Practical Assessment

统计探究循环（statistical enquiry cycle）是整套实践考核的主线，分为五个阶段：提出可统计的问题、设计调查方案、收集数据、分析与呈现数据、得出结论并沟通结果。考官把这五步当作给分框架，而不是写作文的装饰。提出问题时要确保问题能用数据回答，例如多少比例的七年级学生每天运动超过一小时；设计方案要写清调查对象、抽样方法、变量与收集工具；分析阶段要选对图表与统计量；结论必须回到原始问题，用证据说话。评估阶段发现问题之后，还可以带着新问题开始下一轮循环，这正是统计思维的真正面貌。

The statistical enquiry cycle is the backbone of the whole practical assessment, and it runs through five stages: posing a question that can be answered with data, planning the investigation, collecting data, analysing and presenting the data, and drawing conclusions that are then communicated. Examiners treat these stages as the marking framework rather than decorative structure. A good question must be answerable with data, for example what proportion of Year 7 students exercise for more than one hour each day; the plan must state the target group, sampling method, variables and collection tools; the analysis stage demands the right chart and the right summary measures; and the conclusion must return to the original question and speak from evidence. When evaluation reveals weaknesses, a new cycle begins with a refined question, and that is the real shape of statistical thinking.

三、Paper 1数据分析任务：从原始数据到合理解读 / The Paper 1 Data Analysis Task: From Raw Data to Reasoned Interpretation

Paper 1 最后的数据分析任务是全卷最具实践色彩的题目。题目会给出原始数据表、一组图表或现成的统计量，让你像真正的数据分析师一样工作。常见要求包括：发现数据中的错误或不一致、判断某个数据点是否属于异常值、选择合适的图表重新呈现、计算均值与四分位距等统计量，以及解释你的选择。得分的关键在最后一步，无论算出什么数字，都要回到情境里说清楚含义，例如这道题调查的是学生通勤时间，就要说明中位数 35 分钟意味着半数学生上学耗时不超过 35 分钟，而不是只写一个孤零零的数字。

The data analysis task at the end of Paper 1 is the most practical question on the paper. It provides a raw data table, a set of diagrams or ready-made summary statistics and asks you to work like a real data analyst. Typical requirements include spotting errors or inconsistencies in the data, judging whether a value is an outlier, selecting a suitable chart to redraw the data, calculating statistics such as the mean and interquartile range, and justifying your choices. The marks hinge on the final step: whatever numbers you produce must be translated back into the context in plain language. If the survey is about student travel times, you should state that a median of 35 minutes means half the students spend no more than 35 minutes getting to school, rather than writing a bare number.

四、Paper 2调查场景：假设、方法与结论的小报告 / The Paper 2 Investigation Scenario: Hypothesis, Method and Mini-Report

Paper 2 的调查场景题要求你围绕一个现实问题写迷你报告，结构固定为四块。第一块假设（hypothesis），写一句可以检验的判断，例如女孩比男孩每周阅读时间更长，注意假设要具体到可以比较。第二块方法（method），交代样本是谁、用什么抽样、如何测量、如何控制无关变量，这是最容易丢分的地方，很多同学只写我们做了问卷，缺少样本与变量细节。第三块分析（analysis），用合适的图表和统计量呈现证据。第四块结论（conclusion），回到假设判断其是否成立，并指出证据的局限，例如样本太小不能推广。四块缺一不可，顺序本身就是得分框架。

The investigation scenario on Paper 2 asks you to write a mini-report about a real-world problem, and the structure is fixed in four blocks. The first is the hypothesis, a testable statement such as girls spend more time reading each week than boys, and it must be specific enough to compare. The second is the method, covering who is sampled, how the sample is chosen, how variables are measured and how irrelevant variables are controlled; this is where candidates drop the most marks, because many write only we used a questionnaire and omit sample and variable detail. The third block is the analysis, presenting evidence through suitable charts and summary measures. The fourth is the conclusion, which returns to the hypothesis, states whether it is supported and acknowledges limitations such as a small sample that cannot be generalised. All four blocks are compulsory and their order is itself the marking framework.

五、数据收集设计：普查、抽样与问卷 / Designing Data Collection: Censuses, Sampling and Questionnaires

修订版大纲把数据收集列为独立主题，普查（census）与抽样（sample）的取舍是常考点。普查调查全体对象，信息最全，但费时费钱，还可能难以实施；抽样只调查一部分，成本低、速度快，却要承担抽样偏差的风险。问卷设计同样高频，要避免引导性问题，例如是否同意学校应该延长课间休息这类暗示答案的提问；选项要互斥且覆盖全面；先做小规模试测（pilot）能发现歧义。还要分清数据类型：定性数据与定量数据，定量里又分离散与连续，因为数据类型直接决定后面选什么图表、算什么统计量。

The revised syllabus makes data collection a distinct topic, and the trade-off between a census and a sample is a frequent exam question. A census covers every member of the population and gives the fullest information, but it is slow, expensive and sometimes impractical; a sample covers only a part, which is cheap and quick but carries the risk of sampling bias. Questionnaire design is equally high-frequency: avoid leading questions such as do you agree that the school should extend break time, make options mutually exclusive and exhaustive, and run a small pilot test first to catch ambiguous wording. You also need to classify data types, qualitative versus quantitative, and quantitative into discrete and continuous, because the data type determines which chart and which statistic are appropriate later.

六、四种抽样方法怎么选：随机、系统、分层与便利 / Choosing Sampling Methods: Random, Systematic, Stratified and Convenience

抽样方法常考四种，关键是会解释为什么选它。简单随机抽样（simple random sampling）用抽签或随机数表，每个成员被选中的机会相等，没有人为偏见，但前提是有一份完整的名单。系统抽样每隔固定间隔抽取，例如每 20 人取 1 人，操作简单，但名单若存在周期规律就会出问题。分层抽样（stratified sampling）先把总体按性别、年级等特征分层，再按比例从每层抽取，能保证各子群体都被代表。便利抽样（convenience sampling）最容易做，但偏差最大，通常只在初步探索时使用。答题时先描述方法，再说优缺点，最后结合题目情境给出选择理由，三句话一个都不能少。

Four sampling methods dominate the exams, and the key skill is explaining why you choose one. Simple random sampling uses a lottery or random number table so that every member has an equal chance of selection, with no human bias, but it requires a complete list of the population. Systematic sampling selects every nth item, such as one person in every twenty, which is easy to operate but fails if the list hides a periodic pattern. Stratified sampling divides the population into strata by gender, year group or another characteristic and takes proportional numbers from each layer, guaranteeing that every subgroup is represented. Convenience sampling is the easiest to carry out but the most biased, and it is usually reserved for initial exploration. In your answer, describe the method, state its strengths and weaknesses, then justify your choice in the given context; all three sentences are needed.

七、图表选择原则：数据类型决定呈现方式 / Choosing Diagrams: How Data Type Determines Presentation

图表题的第一步永远是判断数据类型。定性数据用条形图或饼图；离散定量数据也常用条形图；连续数据用直方图、累积频率图与箱线图；时间序列数据用折线图，并可能要求计算移动平均、识别季节波动、作出简单预测；双变量数据用散点图，配合相关性与回归直线。绘图规范是固定的得分点：坐标轴必须标名标刻度，描点要精确到半格以内，线性趋势用直尺画直线，非线性趋势画平滑曲线。考官特别提醒，无论数据多漂亮，图画得不规范就拿不到满分，读图读数也要求精确到半小格。

The first step in any chart question is judging the data type. Categorical data suits bar charts or pie charts; discrete quantitative data is also commonly shown with bar charts; continuous data uses histograms, cumulative frequency graphs and box plots; time series data uses line graphs, with possible demands to calculate moving averages, identify seasonal variation and make simple forecasts; and bivariate data uses scatter diagrams together with correlation and a regression line. Plotting conventions are fixed marks: axes must be labelled with names and scales, points must be plotted within half a square, linear trends are drawn with a ruled straight line and non-linear trends with a smooth curve. Examiners stress that no matter how fine the data, an untidy diagram cannot earn full marks, and reading values from graphs also requires accuracy within half a small division.

八、数据清洗与精确度：异常值、缺失值与有效数字 / Data Cleaning and Accuracy: Outliers, Missing Values and Significant Figures

实践考核里数据往往并不干净。遇到异常值（outlier）时，先判断它是否可能来自记录错误，再决定剔除还是保留，无论哪种都要写明理由，例如该值明显偏离其余数据，疑似输入错误，故剔除。缺失数据要如实说明，不能假装不存在，同时指出它对结果的影响。精确度同样是得分点：数值答案要按题目要求或合理精度给出，但中间计算通常需要保留更多位数，避免累积误差；答案不要写成分数与小数混用的形式。图表和读数要求半格精度，单位不能漏，这些细节看似琐碎，却常常决定 A 与 B 的差别。

In practical assessment the data is rarely clean. When you meet an outlier, judge first whether it could come from a recording error, then decide whether to remove it or keep it, and always state your reason, for example the value lies far from the rest and looks like a typing mistake, so it was removed. Missing values must be acknowledged honestly rather than ignored, and you should note their effect on the results. Accuracy is also a mark earner: numerical answers should be given to a suitable degree of accuracy, but working usually needs extra digits to avoid accumulating errors, and a numerical answer should not mix fractions and decimals. Charts and readings require half-square precision and units must never be dropped; these details look trivial but often separate grade A from grade B.

九、高分解读四步法：描述、比较、联系与推断 / Four Steps to High-Scoring Interpretation: Describe, Compare, Relate and Infer

解读题按四步走，步骤越全分越高。第一步描述：说出分布的形状、中心与离散程度，例如数据右偏，中位数 42，四分位距 18。第二步比较：用具体统计量对比两组数据，例如女生组均值高出 12 分，且极差更小，说明成绩更整齐。第三步联系：把数字翻译回现实情境，说明它对原始问题的意义。第四步推断：从样本小心推广到总体，注意加限定词，例如样本量有限，结论仅供参考。还要看清指令词，state 只需一句话，calculate 给出过程与结果，compare 必须同时说相同与不同，interpret 必须结合情境，答错类型等于白写。

Interpretation questions reward a four-step approach, and the more steps you complete, the higher the mark. First describe: state the shape, centre and spread of the distribution, for example the data are skewed to the right with a median of 42 and an interquartile range of 18. Second compare: contrast two sets using concrete statistics, for example the girls group has a mean 12 points higher and a smaller range, so its results are more consistent. Third relate: translate the numbers back into the real-world context and say what they mean for the original question. Fourth infer: generalise cautiously from the sample to the population, adding qualifiers such as the conclusion is only indicative given the limited sample. Also read the command words carefully: state needs one sentence, calculate needs working and an answer, compare must mention similarities and differences, and interpret must use the context; answering the wrong type wastes your effort.

十、评估与局限性：批判性分析的得分点 / Evaluation and Limitations: Scoring with Critical Analysis

评估题是 AO2 的主战场，也是拉开差距的地方。常见局限包括：样本量太小，结果无法推广到总体；抽样方法带偏差，例如只在学校门口拦住路人；测量工具不准确或问题有歧义，导致答案失真；应答者隐瞒真实情况，造成社会期望偏差。写评估不能只说有局限，要具体指出局限是什么、影响哪一步结论，再给出改进建议，例如改用分层抽样、扩大样本量、重写问卷问题、加入控制变量。还要识别假设条件，例如假设样本能代表总体、两次测量相互独立。每写一条局限加一条改进，就是标准的满分结构。

Evaluation questions are the main battlefield of AO2 and the place where grades separate. Common limitations include a sample too small to generalise to the population, a biased sampling method such as stopping people outside the school gate, inaccurate measurement tools or ambiguous question wording that distorts answers, and respondents hiding the truth, which creates social desirability bias. When evaluating, do not just say there is a limitation; state exactly what it is, which conclusion it affects, and offer an improvement, such as switching to stratified sampling, enlarging the sample, rewriting questionnaire items or adding control variables. You should also identify assumptions, for example that the sample represents the population and that separate measurements are independent. One limitation paired with one improvement is the standard full-mark structure.

十一、常见失误与考场时间分配 / Common Mistakes and Exam Time Management

高频失误第一类是结论脱离证据，只写我认为趋势会上升，却没有引用图表或统计量；第二类是不展示过程，只有答案没有 working，步骤分全丢；第三类是图表类型与数据不匹配，例如连续数据画成饼图；第四类是忘记单位、漏标坐标轴、读数不精确；第五类是把相关性当因果，看见正相关就断言一方导致另一方。时间分配建议：Paper 1 和 Paper 2 都是 2 小时 15 分，做最后的数据分析任务或报告前预留至少 25 到 30 分钟；报告题先花两分钟列四块提纲，再动笔。考前用近五年真题计时模拟，把每类题型的用时记下来，考场就不慌。

The first class of frequent mistakes is conclusions detached from evidence, writing I think the trend will rise without citing any chart or statistic; the second is omitting working, so method marks are lost even when the final answer is right; the third is mismatching charts and data, for example drawing a pie chart for continuous data; the fourth is dropping units, leaving axes unlabelled or reading values inaccurately; and the fifth is confusing correlation with causation, claiming one variable causes another simply because they are positively correlated. For timing, both Paper 1 and Paper 2 last 2 hours 15 minutes, so reserve at least 25 to 30 minutes for the final data analysis task or report; for the report, spend two minutes sketching the four blocks before writing. In revision, do timed mock papers from the last five years and record how long each question type takes, so the real exam feels calm.

需要IGCSE统计辅导？

16621398022 同微信

同微信 | 免费咨询课程规划

访问 aleveler.com 了解更多

更多咨询请联系16621398022（同微信）