📚 A-Level CAIE Statistics: Key Points for Statistical Experiments / Practical Assessments | A-Level CAIE 统计：实验/实践考核要点
Practical statistical work in A-Level CAIE Statistics goes beyond calculation. It asks you to plan an investigation, collect or simulate data, apply appropriate models, and interpret findings in context. This guide summarises the key points for experimental and practical assessment tasks under the CAIE Probability & Statistics syllabus.
A-Level CAIE 统计中的实践工作不仅是计算，更要求你规划一项探究、收集或模拟数据、选用合适的模型并在具体情境中解释结果。本文梳理 CAIE 概率与统计大纲下实验与实践考核的关键要点。
1. The Statistical Enquiry Cycle | 统计探究循环
Every practical statistical task follows the enquiry cycle: formulate a question, plan data collection, collect or simulate data, analyse using appropriate techniques, and interpret conclusions in context. Examiners look for evidence that you understand the whole cycle, not just isolated calculations.
每一项统计实践任务都遵循探究循环：提出问题、规划数据收集、收集或模拟数据、使用合适方法分析，并在情境中解释结论。考官看的是你是否理解完整循环，而不是孤立的计算。
Common marks are awarded for clearly stating the problem, describing variables, and linking conclusions back to the original question. In CAIE papers, a practical-style question often asks you to comment on whether a conclusion is reliable or whether a method should be improved.
常见得分点包括清晰陈述问题、描述变量，以及把结论与原始问题联系起来。在 CAIE 试卷中，实践类题目常要求你评论结论是否可靠，或方法应如何改进。
2. Formulating Clear Questions and Hypotheses | 明确问题与假设
A good statistical investigation begins with a precise question, such as ‘Do two brands of batteries last the same length of time?’ From this, you can write null and alternative hypotheses: H₀: μ₁ = μ₂ and H₁: μ₁ ≠ μ₂ for a two-tailed test.
好的统计探究始于精确的问题，例如 “两个品牌的电池使用时间是否相同？” 据此可以写出原假设和备择假设：H₀: μ₁ = μ₂，H₁: μ₁ ≠ μ₂（双侧检验）。
The alternative hypothesis may be one-tailed if the context suggests a direction, for example ‘Brand A lasts longer than Brand B’. Using a one-tailed or two-tailed test depends on the wording of the investigation question.
如果情境暗示方向，备择假设可以是单侧，例如 “品牌 A 的使用时间长于品牌 B”。使用单侧还是双侧检验取决于探究问题的表述。
3. Designing Experiments: Control, Randomisation and Replication | 设计实验：控制、随机化与重复
In an experiment, you change one variable (the independent variable) and measure another (the dependent variable). Control means keeping all other variables constant so they do not affect the response. Randomisation allocates subjects or items to groups by chance, reducing bias.
在实验中，你改变一个变量（自变量）并测量另一个变量（因变量）。控制是指保持所有其他变量不变，使其不影响响应。随机化是按机会将对象或项目分配至各组，以减少偏差。
Replication means repeating measurements or using enough participants so results are not due to one unusual observation. Blocking can be used when there is a known source of variation, such as age or location, by grouping similar units before randomising.
重复意味着反复测量或使用足够多的参与者，使结果不取决于单个异常观测。当存在已知变异来源（如年龄或地点）时，可以使用区组设计，在随机化前将相似单元分组。
4. Sampling Methods and Sources of Bias | 抽样方法与偏差来源
Practical work often uses a sample to represent a population. Simple random sampling gives every member an equal chance of selection. Stratified sampling divides the population into groups and samples from each, ensuring important subgroups are represented.
实践工作常用样本来代表总体。简单随机抽样使每个成员被选中的机会相等。分层抽样将总体分组并从每组抽样，确保重要子群被代表。
Systematic sampling selects every kth item, but it can be biased if there is a hidden pattern. Quota sampling is convenient but is not random. Bias arises from undercoverage, non-response, leading questions, or convenience samples.
系统抽样每隔 k 个项目抽取一个，但如果存在隐藏模式，可能产生偏差。配额抽样方便但不是随机。偏差来自覆盖不足、无回应、诱导性问题或便利抽样。
5. Data Collection and Questionnaire Design | 数据收集与问卷设计
Data can be primary (collected by you) or secondary (from existing sources). Design questionnaires with clear, unambiguous questions. Avoid leading questions such as ‘Do you agree that the new app is excellent?’ because they push respondents toward one answer.
数据可以是一手（由你收集）或二手（来自现有来源）。设计问卷时问题要清晰、无歧义。避免诱导性问题，如 “你是否同意新应用很出色？” 因为这会引导受访者选择某一答案。
Response scales should be consistent, and categories should not overlap. For example, age bands 15–20, 21–25, 26–30 are better than 15–20, 20–25, 25–30, which create ambiguity about where 20 belongs.
回答量表应保持一致，分类不应重叠。例如年龄段 15–20、21–25、26–30 优于 15–20、20–25、25–30，后者会让人不清楚 20 岁属于哪一组。
6. Organising Data: Tables, Charts and Summary Statistics | 整理数据：表格、图表与汇总统计
After collecting data, organise it using frequency tables, histograms, cumulative frequency curves, or box plots. Histograms are used for continuous data and require frequency density = frequency ÷ class width. Box plots compare distributions and highlight outliers.
收集数据后，使用频数表、直方图、累积频率曲线或箱线图进行整理。直方图用于连续数据，需要频率密度 = 频数 ÷ 组距。箱线图用于比较分布并突出异常值。
Summary statistics include measures of central tendency (mean, median, mode) and spread (range, interquartile range, standard deviation). For skewed data, the median and interquartile range are more robust than the mean and standard deviation.
汇总统计包括集中趋势指标（平均数、中位数、众数）和离散程度指标（极差、四分位距、标准差）。对于偏斜数据，中位数和四分位距比平均数与标准差更稳健。
7. Probability Models in Practical Investigations | 实践探究中的概率模型
Practical questions often ask whether a situation can be modelled by a binomial distribution: fixed number of trials, two outcomes, constant probability, and independent trials. For example, counting defective items in a batch of 20 with a 5% defect rate.
实践题常问某一情境是否可以用二项分布建模：固定试验次数、两种结果、概率不变、试验独立。例如，在次品率为 5% 的 20 个产品中计算缺陷品数量。
The Poisson distribution models rare events occurring independently in a fixed interval, such as calls per hour at a call centre. Choosing the correct model is part of practical assessment; if conditions are not met, explain why the model may be inappropriate.
泊松分布用于对固定区间内独立发生的稀有事件建模，例如呼叫中心每小时接到的电话数。选择正确模型是实践考核的一部分；如果条件不满足，要说明模型为何可能不合适。
8. The Normal Distribution and Real Data | 正态分布与实际数据
Many continuous variables such as heights or test scores are approximately normal. Standardise using z = (x – μ) / σ to find probabilities. Use continuity correction when approximating a binomial or Poisson distribution by a normal distribution.
许多连续变量，如身高或考试分数，近似正态分布。标准化使用 z = (x – μ) / σ 来求概率。当用正态分布近似二项或泊松分布时，需要使用连续校正。
In practical work, check that data are roughly symmetric before using normal methods. If the data are heavily skewed, the normal model can give misleading probabilities, so a different approach may be needed.
在实际工作中，使用正态方法前应检查数据是否大致对称。如果数据严重偏斜，正态模型可能给出误导性概率，因此可能需要不同方法。
9. Estimation and Confidence Intervals | 估计与置信区间
An estimator is a rule for estimating a population parameter, such as using sample mean x̄ to estimate μ. A confidence interval gives a range of plausible values. For a large sample mean, a 95% interval is x̄ ± 1.96 × σ/√n when σ is known.
估计量是用来估计总体参数的规则，例如用样本平均数 x̄ 估计 μ。置信区间给出一个可能取值的范围。对于大样本平均数，当 σ 已知时，95% 置信区间为 x̄ ± 1.96 × σ/√n。
If the population standard deviation is unknown and the sample is small, use the t-distribution rather than the normal distribution. In CAIE practical-style questions, you should state assumptions and interpret the interval in context.
如果总体标准差未知且样本较小，应使用 t 分布而非正态分布。在 CAIE 实践类题目中，应陈述假设并在情境中解释区间。
10. Hypothesis Testing in Practical Work | 实践工作中的假设检验
A hypothesis test compares sample evidence with a null hypothesis. Calculate a test statistic, compare with a critical value, or find the p-value. State the significance level, usually 5% or 1%, before testing.
假设检验用样本证据与原假设进行比较。计算检验统计量，与临界值比较，或求出 p 值。在检验前声明显著性水平，通常为 5% 或 1%。
For binomial tests, use the exact probabilities. For normal tests, use z-scores. Decisions must be written in context: ‘Reject H₀. There is sufficient evidence at the 5% level that the mean time has increased.’
对于二项检验，使用精确概率。对于正态检验，使用 z 分数。结论必须写在情境中：”拒绝 H₀。在 5% 显著性水平下，有充分证据表明平均时间增加了。”
11. Using Technology and Statistical Software | 使用技术与统计软件
In practical assessments, you may use calculators, spreadsheets, or software to enter data, draw charts, and compute statistics. Show your method clearly: an examiner cannot award method marks if only a final number is given without working.
在实践考核中，你可能使用计算器、电子表格或软件录入数据、绘图和计算统计量。要清晰展示方法：如果只给出最终数字而没有过程，考官无法给予方法分。
Use technology to check hand calculations, but do not rely on it blindly. Rounding errors can accumulate, so keep several more decimal places during intermediate steps and round only the final answer.
使用技术检查手算结果，
Published by TutorHao | A-Level 统计 Revision Series | aleveler.com
更多咨询请联系16621398022（同微信）
屏轩国际教育cambridge primary/secondary checkpoint, cat4, ukiset,ukcat,igcse,alevel,PAT,STEP,MAT, ibdp,ap,ssat,sat,sat2课程辅导，国外大学本科硕士研究生博士课程论文辅导