📚 SQA Advanced Higher Statistics: Key Points for Experimental / Practical Assessment | SQA 高级统计学实验/实践考核要点
The experimental and practical assessment is a defining component of the SQA Advanced Higher Statistics course, designed to evaluate your ability to conduct independent statistical investigation. It involves formulating a research hypothesis, planning data collection, applying rigorous analysis, interpreting outcomes and presenting a coherent report. Mastering these key points is essential for achieving a high grade.
实验与实践考核是 SQA 高级统计学课程的核心组成部分，旨在评估你开展独立统计调查的能力。它涵盖提出研究假设、规划数据收集、运用严格分析、解释结果并撰写出条理清晰的报告。掌握这些关键要点对获得高分至关重要。
1. Understanding the Assessment Structure | 理解评估结构
The practical assessment, often called the ‘investigation’, is internally marked and externally moderated. It gives you the freedom to design an experiment or survey on a topic of your choice, but within a structured framework. You must demonstrate the complete statistical cycle: problem identification, planning, data gathering, analysis, interpretation and critical evaluation. Your work is judged on the quality of statistical thinking, not just the final answer.
实践考核通常被称为“调查”，由内部评分、外部审核。它让你在限定的框架内自由选择主题设计实验或调查。你必须展示完整的统计周期：问题识别、计划、数据收集、分析、解释和批判性评估。评分的重点是统计思维的质量，而不仅仅是最终答案。
2. Formulating a Clear and Testable Hypothesis | 提出清晰可检验的假设
A well‑crafted hypothesis is the backbone of your investigation. State it in terms of population parameters (e.g. μ, p, ρ) and avoid ambiguous language. For a comparative study, write null and alternative hypotheses such as H₀: μ₁ = μ₂ vs. H₁: μ₁ ≠ μ₂. For an association study, specify the expected direction: ‘There is a positive linear relationship between hours of sleep and cognitive test score.’ Operationalise each variable so that it is measurable and record the measurement units.
精心构建的假设是调查的基石。用总体参数（如 μ、p、ρ）来表述，避免模糊语言。对于比较研究，写出零假设和备择假设，例如 H₀: μ₁ = μ₂ 与 H₁: μ₁ ≠ μ₂。对于关联研究，要指明预期方向：“睡眠时长与认知测试成绩之间存在正的线性关系”。将每个变量操作化，使其可测量，并注明测量单位。
3. Planning the Data Collection Strategy | 规划数据收集策略
Decide whether you will use primary data (collected by you) or reliable secondary data. Identify the target population and choose a suitable sampling method, such as simple random, stratified or systematic sampling. For experiments, describe the randomisation process and control of confounding variables. Calculate the minimum sample size needed by considering effect size, power (common standard 0.8) and significance level α = 0.05. Document every step so that your study can be replicated.
确定使用一手数据（自行收集）还是可靠的二手数据。明确目标总体，并选择合适的抽样方法，如简单随机、分层或系统抽样。对于实验，需描述随机化过程和对混杂变量的控制。通过考虑效应量、检验功效（常用标准 0.8）和显著性水平 α = 0.05，计算所需的最小样本量。记录每一步操作，使你的研究可重复。
4. Ensuring Validity, Reliability and Ethics | 确保效度、信度与伦理
Internal validity demands that any observed effect is truly due to the independent variable; achieve this through random allocation and blinding where possible. Reliability refers to the consistency of measurements – use calibrated instruments and clearly defined protocols. Ethically, you must obtain informed consent, guarantee anonymity and handle data securely under GDPR principles. If using human participants, a short briefing statement and the right to withdraw are compulsory.
内部效度要求观察到的效应确实由自变量引起；尽可能通过随机分配和盲法来实现。信度指测量的一致性——使用经过校准的工具和明确定义的规程。伦理方面，必须获得知情同意，保证匿名并按照 GDPR 原则安全处理数据。若有人类参与者，必须提供简短的情况说明和退出权利。
5. Cleaning and Exploring Your Data | 清洗与探索数据
Before jumping into hypothesis testing, examine the dataset for errors, missing values and outliers. Create visualisations: histograms to assess distribution shape, boxplots to spot anomalies, and scatterplots for bivariate relationships. Calculate descriptive statistics – mean, median, standard deviation, interquartile range – and present them in a well‑labelled table. Use parallel boxplots or density plots to compare groups visually.
在进行假设检验之前，先检查数据集中是否存在错误、缺失值和异常值。创建可视化图形：用直方图评估分布形态，用箱线图发现异常，用散点图探索双变量关系。计算描述性统计量——均值、中位数、标准差、四分位距——并在标注清晰的表格中呈现。使用平行箱线图或密度图进行分组视觉比较。
6. Selecting Appropriate Statistical Tests | 选择合适的统计检验
The choice of test depends on your research question, data type and whether parametric assumptions are met. The table below summarises common scenarios in Advanced Higher investigations.
检验的选择取决于你的研究问题、数据类型以及参数假设是否满足。下表总结了高级统计调查中的常见情形。
| Research Question | Recommended Test | Key Statistic / Equation |
|---|---|---|
| Compare two independent means (equal variance) | Two‑sample t‑test (pooled) | t = (x̄₁ − x̄₂) / (sₚ√(1/n₁ + 1/n₂)) |
| Compare two independent means (unequal variance) | Welch’s t‑test | t = (x̄₁ − x̄₂) / √(s₁²/n₁ + s₂²/n₂) |
| Compare paired means | Paired t‑test | t = d̄ / (s_d / √n) |
| Association between two categorical variables | Chi‑squared test for independence | χ² = ∑ (O − E)² / E |
| Linear relationship between two continuous variables | Pearson’s r / simple linear regression | ŷ = b₀ + b₁x; t = b₁ / SE(b₁) |
| Compare more than two independent groups | One‑way ANOVA | F = MS_between / MS_within |
Always justify your choice with reference to the data structure, level of measurement and preliminary graphical analysis. If assumptions are violated, consider a non‑parametric alternative such as the Mann–Whitney U test or Spearman’s rank correlation.
始终根据数据结构、测量层次以及初步图形分析来证明你的选择。若假设被违反，可考虑非参数替代方法，如 Mann–Whitney U 检验或 Spearman 秩相关。
7. Checking Assumptions and Transforming Data | 检验假设与数据转换
Parametric tests rest on assumptions that must be verified. For a t‑test or ANOVA, check normality (Shapiro‑Wilk test, Q‑Q plots) and homogeneity of variance (Levene’s test). If your data are skewed, apply a transformation: logarithmic (y’ = ln y) for right‑skewed data, square root (y’ = √y) for count data, or Box‑Cox for a general power transform. Document the transformation and re‑evaluate the assumptions afterwards.
参数检验依赖于必须验证的假设。对于 t 检验或方差分析，需检验正态性（Shapiro‑Wilk 检验、Q‑Q 图）和方差齐性（Levene 检验）。如果数据偏态，可进行转换：右偏数据用对数转换（y’ = ln y），计数数据用平方根转换（y’ = √y），或使用一般的幂转换 Box‑Cox。记录转换过程，并随后重新检查假设。
8. Interpreting p‑values, Effect Sizes and Confidence Intervals | 解释 p 值、效应量与置信区间
A small p‑value (p < α) leads you to reject H₀, but it does not measure the magnitude of an effect. Complement the p‑value with an effect size statistic, such as Cohen’s d for mean differences or Cramér’s V for chi‑squared. Report confidence intervals (e.g. 95% CI for μ₁ − μ₂) to indicate the precision of your estimate. Always distinguish statistical significance from practical importance.
较小的 p 值（p < α）使你可以拒绝 H₀，但它并不衡量效应的大小。用效应量统计量补充 p 值，例如均值差异的 Cohen’s d 或卡方检验的 Cramér’s V。报告置信区间（如 μ₁ − μ₂ 的 95% 置信区间）以表明估计的精确度。务必区分统计显著性与实际重要性。
9. Critical Evaluation and Identifying Limitations | 批判性评估与识别局限性
No investigation is perfect. Discuss potential biases (selection, measurement, confounding), limitations of your sampling method, and the impact of small sample size on generalisability. Reflect on whether the data truly satisfy the test assumptions, and consider how outliers or missing data may have influenced results. Propose realistic improvements you would make if replicating the study.
没有哪项调查是完美的。讨论潜在的偏倚（选择偏倚、测量偏倚、混杂）、抽样方法的局限性，以及小样本量对可推广性的影响。反思数据是否真正满足检验假设，并考量异常值或缺失数据对结果可能产生的影响。提出若重复研究你将会做出的切实改进。
10. Structuring an Outstanding Report | 构建出色的报告
Follow a logical structure: title, abstract, introduction (background and hypothesis), methodology (design, participants, apparatus), results (descriptive and inferential statistics with tables and fully labelled graphs), discussion (interpretation, comparison to theory, limitations) and conclusion. Use clear headings and number all tables and figures. Write in a precise, objective style, avoiding personal pronouns where possible.
遵循逻辑结构：标题、摘要、引言（背景与假设）、方法（设计、参与者、器材）、结果（描述性与推断性统计，配以表格和完整标注的图表）、讨论（解释、与理论比较、局限性）和结论。使用清晰的标题，并为所有表格和图形编号。用精确、客观的文体撰写，尽可能避免使用人称代词。
11. Time Management and Project Milestones | 时间管理与项目里程碑
Break the project into manageable stages with deadlines: topic approval, planning sheet, data collection, analysis draft, final write‑up. Allocate time for unforeseen delays, especially when collecting human participant data. Regularly consult your teacher for formative feedback, but remember that the final submission must be your own independent work.
将项目分解为带截止日期的可管理阶段：主题审批、计划表、数据收集、分析初稿、最终成文。为意外延误预留时间，特别是在收集人类参与者数据时。定期咨询老师以获取形成性反馈，但要记住终稿必须是你自己的独立成果。
12. Final Checks and Common Mistakes to Avoid | 最终检查与常见错误
Before submission, proofread for calculation errors and check that all graphs are appropriately scaled and labelled. Ensure your hypotheses are restated in the conclusion and that the discussion does not overclaim beyond what the data support. Typical pitfalls include using the wrong test (e.g., t‑test for ordinal data), ignoring assumptions, presenting raw computer output without interpretation, and forgetting to reference sources or the data origin.
提交前，仔细校对计算错误，并确保所有图形的刻度与标注恰当。确保在结论中再次陈述假设，讨论部分不要做出数据不支持的主张。常见误区包括：使用了错误的检验（例如对定序数据使用 t 检验）、忽略假设、直接呈现未经解释的原始计算机输出，以及忘记标注参考来源或数据出处。
Published by TutorHao | Statistics 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