📚 IGCSE CIE Psychology: Experimental/Practical Assessment Essentials | IGCSE CIE 心理学：实验/实践考核要点
Research methods and experimental design form the backbone of IGCSE CIE Psychology, assessed primarily through Paper 2: Research Methods. Students must not only understand the core studies but also demonstrate the ability to design investigations, handle data and evaluate research critically. This revision guide covers the essential practical assessment points, from identifying variables to judging validity, all presented in clear English-Chinese pairs to strengthen both conceptual understanding and examination technique.
研究方法和实验设计是 IGCSE CIE 心理学的核心支柱，主要通过试卷二“研究方法”进行评估。学生不仅要理解核心研究，还必须展现设计调查、处理数据和批判性评价研究的能力。这份复习指南涵盖了实验/实践考核的核心要点，从识别变量到评判效度，均以清晰的英中对照呈现，以强化概念理解和考试技巧。
1. Understanding Variables | 理解变量
An experiment investigates the effect of an independent variable (IV) on a dependent variable (DV). The independent variable is the factor the researcher deliberately manipulates or changes, such as the presence of background noise versus silence. The dependent variable is the outcome measured to see if it was affected, for example the number of words correctly recalled. It must be operationalised – defined in a way that is precise, measurable and replicable, like 'number of words recalled from a list of 20 within 2 minutes'.
实验旨在考察自变量（IV）对因变量（DV）的影响。自变量是研究者有意操纵或改变的因素，例如有背景噪音与安静环境。因变量是被测量的结果，以观察是否受到影响，例如正确回忆的单词数量。它必须被操作性定义——即以精确、可测量、可复制的方式定义，例如“在 2 分钟内从 20 个单词列表中回忆出的单词数量”。
Extraneous variables are any variables other than the IV that could affect the DV. If an extraneous variable systematically changes with the IV, it becomes a confounding variable, making it impossible to tell whether the IV truly caused a change. Common extraneous variables include participant mood, time of day and noise levels. Researchers must control these through standardised procedures, random allocation or counterbalancing.
额外变量是指除自变量外任何可能影响因变量的因素。如果额外变量与自变量一起系统变化，它就成为混淆变量，导致无法判断自变量是否真正引起了变化。常见的额外变量有参与者情绪、时间和噪音水平。研究者必须通过标准化程序、随机分配或平衡法加以控制。
2. Experimental Designs | 实验设计
Independent groups design uses different participants in each condition of the IV. Its main advantage is that there are no order effects, as participants experience only one condition. However, participant variables such as age or intelligence can differ between groups, potentially confounding results. Random allocation helps reduce this bias but does not eliminate it completely.
独立组设计在每个自变量条件下使用不同的参与者。其主要优点是没有顺序效应，因为参与者只经历一种条件。然而，参与者变量（如年龄或智力）可能在组间存在差异，从而可能混淆结果。随机分配有助于减少这种偏差，但无法完全消除。
Repeated measures design uses the same participants in all conditions. This controls participant variables effectively because the same individuals act as their own baseline. Yet it introduces order effects: performance may improve due to practice or worsen due to fatigue or boredom. Counterbalancing – presenting conditions in different orders – helps minimise these effects. Demand characteristics may also increase as participants guess the aim after experiencing multiple conditions.
重复测量设计在所有条件下使用相同的参与者。这能有效控制参与者变量，因为同一个人作为自己的基线。但它会引入顺序效应：表现可能因练习而提高，或因疲劳、厌倦而下降。平衡法——以不同顺序呈现条件——有助于最小化这些影响。也可能出现需求特征，因为参与者在经历多个条件后可能猜到实验目的。
Matched pairs design involves pairing participants on key characteristics (e.g. memory span, age) and then splitting each pair into different conditions. It combines benefits of controlling participant variables and avoiding order effects, but matching is time-consuming and never perfect, as participants can only be matched on a few known variables.
匹配对设计先将参与者在关键特征（如记忆广度、年龄）上进行配对，再把每对分入不同条件。它结合了控制参与者变量和避免顺序效应的优点，但匹配过程耗时且永远无法完美，因为参与者只能就少数已知变量进行匹配。
3. Hypotheses | 假设
A hypothesis is a clear, testable prediction about the relationship between the IV and DV. The alternative (experimental) hypothesis states that there will be a significant effect or difference. A two-tailed (non-directional) hypothesis simply predicts 'there will be a difference' without specifying direction, used when previous evidence is mixed. A one-tailed (directional) hypothesis specifies the expected direction, e.g. 'participants in silence will recall more words than those in noise'. The null hypothesis always states there will be no significant effect or difference, and any observed result is due to chance.
假设是对自变量与因变量之间关系的清晰、可检验的预测。备择（实验）假设陈述将会出现显著效应或差异。双尾（非定向）假设仅预测“会有差异”，不指定方向，适用于先前证据不一致的情况。单尾（定向）假设指定预期方向，例如“安静环境下的参与者比噪音下的参与者回忆更多单词”。零假设总是陈述不会有显著效应或差异，任何观察结果都源于偶然。
A well-written hypothesis must include operationalised variables. For example: 'There will be a significant difference in the number of words recalled from a list of 20 (DV) between participants tested in silence and those tested with loud traffic noise (IV).' This allows replication and clear interpretation.
一个书写良好的假设必须包含操作性定义的变量。例如：“在安静环境下测试的参与者与在嘈杂交通噪音下测试的参与者，从 20 个单词列表中回忆出的单词数量（因变量）将存在显著差异（自变量）。”这允许复制和清晰的解释。
4. Sampling Methods | 抽样方法
A sample is the group of participants drawn from a target population. Random sampling gives every member of the target population an equal chance of selection, reducing researcher bias, but it requires a complete list of the population and may still produce a non-representative sample by chance. Opportunity sampling selects participants who are readily available, making it quick and cheap; however, it is likely to be biased because the sample may consist of similar types of people.
样本是从目标总体中抽选的参与者群体。随机抽样让目标总体的每个成员有同等被选中的机会，减少研究者偏差，但它需要完整的总体名单，且偶然仍可能产生不具代表性的样本。机会抽样选择容易接触到的参与者，快速且成本低；但很可能有偏差，因为样本可能由相似类型的人组成。
Volunteer (self-selected) sampling involves participants offering to take part, often through advertisements. While ethical and straightforward, it often yields a biased sample of highly motivated or curious individuals. Stratified sampling divides the population into relevant subgroups and then selects randomly from each in proportion to their occurrence. This produces a highly representative sample but is complex and time-consuming.
志愿者（自选）抽样涉及参与者主动提出参加，常通过广告进行。虽然符合伦理且操作直接，但通常产生有高度动机或好奇心的个体的偏差样本。分层抽样将总体划分为相关子群，再按比例从每个子群随机抽取。这能产生高度代表性的样本，但过程复杂且耗时。
5. Ethical Considerations | 伦理考量
All psychological research must adhere to ethical guidelines to protect participants. Informed consent means participants must be given enough information about the study to make a voluntary decision. If full disclosure would bias results, presumptive consent or prior general consent may be used, and participants must be fully debriefed afterwards. The right to withdraw allows participants to leave the study at any time without penalty, and their data must be removed if requested.
所有心理学研究都必须遵守伦理准则以保护参与者。知情同意意味着必须向参与者提供足够的研究信息，使其做出自愿决定。如果完全披露会偏导结果，可使用推定同意或事先一般同意，且事后必须充分解释（脱敏）。退出权利允许参与者在任何时间无惩罚地离开研究，若要求删除其数据则必须删除。
Researchers must protect participants from psychological and physical harm. Deception should be avoided unless absolutely necessary and justified; if unavoidable, a thorough debrief is required to restore well-being. Confidentiality and privacy must be maintained – data should be anonymised and stored securely, and observations in private settings generally require consent.
研究者必须保护参与者免受心理和身体伤害。除非绝对必要且有正当理由，否则应避免欺骗；若不可避免，需要彻底的事后解释以恢复福祉。必须维护保密性和隐私——数据应匿名化并安全存储，在私密场所的观察通常需要同意。
6. Data Collection Techniques | 数据收集技术
Experiments are the strongest method for establishing cause–effect relationships. Laboratory experiments offer high control over extraneous variables and easy replication, but may lack ecological validity. Field experiments take place in real-world settings, boosting mundane realism but reducing control. Natural and quasi-experiments exploit naturally occurring changes in the IV, allowing research where manipulation would be unethical, though participants cannot be randomly allocated, weakening causal claims.
实验是建立因果关系的最强方法。实验室实验对额外变量控制度高，易于复制，但可能缺乏生态效度。现场实验在真实世界环境中进行，增强了生活真实性，但控制度降低。自然实验和准实验利用自变量的自然变化，可在操纵不道德的情况下进行研究，但参与者无法随机分配，削弱了因果推论。
Non-experimental methods include self-reports (questionnaires and interviews) and observations. Questionnaires can collect large amounts of data quickly, but responses may be affected by social desirability bias. Interviews allow deeper exploration but are time-consuming and interviewer effects may arise. Observations capture natural behaviour; structured observations use behavioural categories for quantitative data, while unstructured observations produce qualitative detail. A reliable observational system requires clearly defined, mutually exclusive categories.
非实验方法包括自我报告（问卷和访谈）和观察。问卷能快速收集大量数据，但回答可能受社会赞许性偏差影响。访谈允许深入探索，但耗时且可能出现访谈者效应。观察捕捉自然行为；结构式观察使用行为类别获得定量数据，而非结构式观察产生质性细节。一个可信的观察系统需要清晰界定、互斥的行为类别。
7. Descriptive Statistics | 描述统计
Descriptive statistics summarise data and allow comparisons. Measures of central tendency indicate a typical value in a data set. The mean (average) uses all values and is sensitive to extreme scores; the median is the middle value when ordered, unaffected by outliers, ideal for skewed data; the mode is the most frequent value, useful for categorical data. In an exam, you may need to calculate the mean of a condition and comment on which measure is most appropriate.
描述统计概括数据并允许比较。集中趋势测量值指示数据集中的典型值。均值（平均数）使用所有值且对极端分数敏感；中位数是按顺序排列的中间值，不受异常值影响，适合偏态分布数据；众数是出现频率最高的值，对分类数据有用。考试中您可能需要计算一个条件的均值，并评论哪种测量最合适。
adly affected by a single extreme value. Standard deviation is a more precise measure reflecting the average distance of each score from the mean. A larger standard deviation means scores are more spread out. The formula for standard deviation is: √(Σ(x - x̄)² / n), where x is each score, x̄ is the mean and n is the number of scores.
离散程度测量描述分数的分布。极差是最高分与最低分之差；计算简单，但极易受单个极端值影响。标准差是更精确的测量，反映各分数与均值的平均距离。标准差越大，分数越分散。标准差的公式是：√(Σ(x - x̄)² / n)，其中 x 是每个分数，x̄ 是均值，n 是分数个数。
8. Graphs and Charts | 图表
Visual representation of data must be appropriate. Bar charts are used for discrete (categorical) independent variables, such as comparing two conditions (silence vs. noise). The bars do not touch, emphasising the separateness of categories. The y-axis shows the mean score of the DV, and the x-axis lists conditions. A histogram, in contrast, displays continuous interval data; bars touch to show the continuity of the scale. Scattergrams plot the relationship between two co-variables, with each point representing one participant’s paired scores. A line of best fit can indicate the direction and strength of a correlation.
数据的可视化呈现必须恰当。条形图用于离散（分类）自变量，例如比较两个条件（安静 vs. 噪音）。条形不接触，强调类别的独立性。y 轴显示因变量的均值，x 轴列出条件。与之相反，直方图显示连续区间数据，条形接触以显示量尺的连续性。散点图绘制两个共变量之间的关系，每个点代表一个参与者的配对分数。最佳拟合线可指示相关的方向和强度。
When describing graphs, always comment on the general trend, any obvious differences between conditions, and the variability shown by error bars if present. Do not simply restate numbers; interpret what the data pattern means with respect to the hypothesis. A clear, labelled graph can be worth several marks in Paper 2, especially when linked to a conclusion.
描述图表时，务必评论总体趋势、条件间明显差异以及误差条（如果存在）显示的变化性。不要只是重述数字；要结合假设解释数据模式的含义。一个清晰标注的图表在试卷二中可能值好几分，尤其是当它与结论相关联时。
9. Controlling Extraneous Variables | 控制额外变量
To increase internal validity, researchers must minimise the impact of extraneous variables. Standardised procedures ensure every participant experiences the same environment, instructions and timing, except for the manipulated IV. Random allocation to conditions distributes participant characteristics evenly in independent groups designs. Counterbalancing in repeated measures designs controls for order effects by varying the order of conditions across participants.
为提高内部效度，研究者必须最小化额外变量的影响。标准化程序确保除操纵的自变量外，每个参与者经历相同的环境、指导和计时。随机分配至各条件能在独立组设计中均匀分配参与者特征。重复测量设计中的平衡法通过改变参与者经历条件的顺序来控制顺序效应。
Single-blind procedures keep participants unaware of the condition to which they are assigned, reducing demand characteristics – where participants change behaviour to fit perceived expectations. Double-blind procedures also keep the researcher unaware, preventing investigator effects where a researcher unconsciously influences results. Use of a placebo condition matches the treatment setting without the active component, controlling for expectancy. All these controls strengthen the conclusion that the IV alone caused the change in DV.
单盲程序让参与者不知道自己属于哪个条件，从而减少需求特征——即参与者为迎合感知期望而改变行为。双盲程序也使研究者不知晓分配情况，防止研究者效应，即研究者无意识地影响结果。使用安慰剂条件匹配干预设置但不含活性成分，控制期望效应。所有这些控制方法都加强了结论：唯有自变量引起了因变量的变化。
10. Evaluating Studies | 评价研究
Evaluating research involves judging validity, reliability, and relevance. Internal validity asks whether the study truly measured what it intended to test – were confounding variables properly controlled? External validity, including ecological validity (realism) and population validity (generalisability), considers whether findings apply beyond the study context. Reliability means consistency: would the same procedure yield similar results if replicated? Standardised procedures and inter-rater reliability checks support this.
评价研究涉及判断效度、信度和相关性。内部效度追问研究是否真正测量了想要测试的内容——混淆变量是否得到恰当控制？外部效度，包括生态效度（生活真实性）和总体效度（可推广性），考虑结果能否应用于研究情境之外。信度意味着一致性：同样的程序如果重复是否会得到相似结果？标准化程序和评估者间信度检验支持这一点。
Application to real life and ethical strengths and weaknesses also matter. A method might be highly controlled but lack mundane realism, limiting practical use. Conversely, a naturalistic observation may be ecologically valid but have ethical concerns over consent. The GRAVE acronym (Generalisability, Reliability, Applicability, Validity, Ethics) is a useful evaluation framework for IGCSE essays. Always support evaluations with specific evidence from the study or method being discussed.
现实生活的应用以及伦理优劣也很重要。某种方法可能控制度很高，但缺乏生活稳定性，限制了实际应用。相反，自然观察可能生态效度高，但在同意方面有伦理疑虑。GRAVE 缩写（可推广性、信度、应用性、效度、伦理）是 IGCSE 论述题中有用的评价框架。始终用所讨论研究或方法的具体证据来支撑评价。
Published by TutorHao | Psychology Revision Series | aleveler.com
更多咨询请联系16621398022（同微信）
或截图后用「扫一扫 → 相册」选择本图