📚 IGCSE Cambridge Psychology: Experimental/Practical Assessment Essentials | IGCSE Cambridge 心理学：实验/实践考核要点
In Cambridge IGCSE Psychology, the experimental and practical assessment asks you to think like a researcher rather than a student reciting facts. You must plan a simple study, define the variables precisely, control what can be controlled, collect and present data sensibly, and then judge how much confidence the results deserve. This article walks through the whole practical chain: aims and hypotheses, operationalisation, experimental designs, sampling, controls, ethics, data presentation, descriptive and inferential reasoning, evaluation of validity and reliability, and the exam technique that turns that understanding into marks.
在剑桥 IGCSE 心理学中，实验与实践考核要求你像研究者一样思考，而不是像背知识点的学生。你必须设计一项简单研究、精确界定变量、控制可控因素、合理收集与呈现数据，并判断结果值得多少信任。本文完整梳理实践链条：目的与假设、操作化、实验设计、抽样、控制条件、伦理、数据呈现、描述与推断性推理、效度与信度的评价，以及把理解转化为分数的答题技巧。
1. What the practical component actually tests | 实践部分到底考什么
The practical component, whether it takes the form of a school-based investigation or an alternative-to-practical written paper, rewards exactly the same skill set. You must be able to plan, conduct, analyse and evaluate a simple psychological study. Questions often begin with a short description of a proposed investigation, then ask in sequence for hypotheses, variables, controls, data presentation and evaluation. The scenario changes every year; the skills do not.
实践部分无论是校内调查形式，还是实践替代笔试卷，考查的都是同一套技能：计划、实施、分析和评价一项简单的心理学研究。题目通常先给出一段简短的研究描述，然后依次要求写出假设、变量、控制条件、数据呈现和评价。情境每年不同，技能始终不变。
Do not simply define terms. Cambridge examiners expect you to apply them to the given context. For example, if a study tests memory in noisy and quiet rooms, you should state that the independent variable is the level of background noise, not merely write the phrase "independent variable". A definition with no context usually earns no credit.
不要只是给术语下定义。剑桥考官希望你结合给定情境应用术语。例如，如果研究比较嘈杂房间与安静房间中的记忆表现，你应写出自变量是背景噪音水平，而不是只写"自变量"三个字。没有情境的定义通常拿不到分。
Planning: writing an aim, an operationalised hypothesis, and a workable procedure. | 计划：写出研究目的、可操作化的假设和可执行的程序。
Analysing: choosing suitable tables and graphs, then calculating simple summary statistics. | 分析：选择合适的表格与图形，并计算简单的汇总统计量。
Evaluating: judging validity, reliability, generalisability and ethics in the specific context. | 评价：在具体情境中判断效度、信度、概化性和伦理性。
Applying: linking every answer back to the named study rather than to psychology in general. | 应用：把每个答案都扣回题目指定的研究，而不是泛谈心理学。
2. Aims, research questions and hypotheses | 研究目的、研究问题与假设
An aim states what the researcher wants to find out, in broad terms. A research question turns that aim into an answerable question. A hypothesis is a precise, testable prediction about the expected outcome, and it must contain both variables in operational form plus the word "significant" if a statistical test will be used. IGCSE answers usually require one clear hypothesis rather than a discussion of several possibilities.
研究目的用宽泛的语言说明研究者想弄清什么。研究问题把目的转化为可回答的问题。假设则是对预期结果的精确、可检验的预测，必须包含两个已操作化的变量；如果要用统计检验，还应出现"显著"一词。IGCSE 的答案通常只需要一个清晰假设，而不是讨论多种可能。
Distinguish directional (one-tailed) from non-directional (two-tailed) hypotheses. A directional hypothesis predicts the direction of the difference, while a non-directional hypothesis predicts only that a difference or relationship exists. If earlier research or a theory gives you a reason to expect a particular direction, the directional form is justified; if not, use the non-directional form. You should also be able to write a null hypothesis, which states that any difference found is due to chance.
要区分方向性（单尾）与非方向性（双尾）假设。方向性假设预测差异的方向；非方向性假设只预测存在差异或相关。如果先前研究或理论支持某个方向，方向性假设就有依据；否则用非方向性。你还要能写出零假设，即所发现的差异由随机因素造成。
Directional: participants in the noisy condition will recall significantly fewer words than participants in the quiet condition.
方向性：嘈杂条件下的被试回忆出的单词数将显著少于安静条件下的被试。
Notice three features of that sentence: it names the independent variable (background noise), it names the dependent variable (number of words recalled), and it commits to a direction (fewer). A hypothesis that says "noise affects memory" is too vague and would not score well, because the DV is not operationalised and no direction is stated.
注意这句话的三个特征：点明自变量（背景噪音）、点明因变量（回忆出的单词数）、并确定了方向（更少）。若写成"噪音影响记忆"，就过于含糊，因变量没有操作化、也没给出方向，难以得高分。
3. Identifying and operationalising variables | 识别并操作化变量
The independent variable, or IV, is the factor that is deliberately changed or compared by the researcher. The dependent variable, or DV, is the factor that is measured. Operationalisation means defining the IV and DV precisely enough that they can be manipulated or measured in practice. Vague variables produce vague studies, and vague studies produce results that cannot be interpreted.
自变量（IV）是研究者有意改变或进行比较的因素；因变量（DV）是被测量的因素。操作化就是把自变量与因变量界定得足够精确，以便实际操纵或测量。模糊的变量产生模糊的研究，模糊的研究产生无法解释的结果。
| Term | 术语 | Meaning | 含义 | Example | 示例 |
| Independent variable | 自变量 | Changed or compared by the researcher | 由研究者改变或比较 | Background noise level: 70 dB vs 30 dB | 背景噪音：70 dB 与 30 dB |
| Dependent variable | 因变量 | Measured as the outcome | 作为结果被测量 | Number of words correctly recalled out of 20 | 20 个词中正确回忆的个数 |
| Extraneous variable | 额外变量 | Any unwanted factor that could affect the DV | 任何可能影响因变量的无关因素 | Time of day, age of participants | 测试时间、被试年龄 |
| Confounding variable | 混淆变量 | An extraneous variable that varies systematically with the IV | 与自变量系统性共同变化的额外变量 | Noisy group tested in the evening, quiet group in the morning | 噪音组晚上测、安静组早上测 |
| Control variable | 控制变量 | A factor deliberately held constant | 被有意保持恒定的因素 | Same word list, same room, same instructions | 同一词表、同一房间、同一指导语 |
You should also recognise levels of measurement, because they decide which statistics and graphs are suitable. Nominal data are categories (for example, whether a participant is in the noisy or quiet group). Ordinal data are ranked but with unequal intervals (for example, a 1-5 rating of how distracting the room felt). Interval data have equal intervals (for example, number of words recalled). Identify the level of measurement before choosing a graph or a summary statistic.
你还要能识别测量水平，因为它决定合适的统计方法与图形。称名数据是类别（例如被试属于噪音组还是安静组）；顺序数据有等级但间隔不等（例如对房间干扰程度打 1-5 分）；等距数据间隔相等（例如回忆出的单词数）。选图形或汇总统计量之前，先判断测量水平。
4. Experimental designs and participant allocation | 实验设计与被试分配
There are three basic experimental designs. In an independent measures design, different participants take part in each condition. In a repeated measures design, the same participants take part in every condition. In a matched pairs design, participants are paired on a relevant characteristic and one member of each pair is allocated to each condition. Each design has consequences for participant variables, order effects and the number of participants needed.
基本实验设计有三种。独立组设计中，每个条件由不同被试参与；重复测量设计中，同一批被试参与所有条件；配对组设计中，先按相关特征把被试配对，再把每对中的一人分到不同条件。每种设计都会影响被试变量、顺序效应以及所需被试数量。
| Design | 设计 | Strength | 优点 | Weakness | 缺点 |
| Independent measures | 独立组 | No order effects; no practice carry-over | 无顺序效应、无练习迁移 | Participant variables differ between groups; needs more people | 组间被试差异大、需要更多被试 |
| Repeated measures | 重复测量 | Participant variables controlled; fewer people needed | 控制被试变量、所需人数少 | Order effects, fatigue, practice, demand characteristics | 顺序效应、疲劳、练习、需求特征 |
| Matched pairs | 配对组 | Reduces participant variables without order effects | 既减少被试差异又避免顺序效应 | Matching is time-consuming and never perfect | 配对耗时且不可能完全匹配 |
When you describe allocation, be explicit: state whether participants were allocated randomly, for example by drawing names from a hat or using a random number generator. Random allocation is the strongest defence against participant variables becoming confounding variables in an independent measures design. If you use repeated measures, explain how you will counterbalance the order, such as half the participants doing condition A then B, and half doing B then A.
描述分配方式时要明确：说明被试是否被随机分配，例如抽签或使用随机数字生成器。在独立组设计中，随机分配是防止被试变量变成混淆变量的最有力手段。若使用重复测量，要说明如何平衡顺序，例如一半被试先做 A 再做 B，另一半先做 B 再做 A。
5. Sampling and participant selection | 抽样与被试选择
The target population is the whole group the researcher wants to draw conclusions about, while the sample is the smaller group actually tested. The sampling method determines how representative the sample is, and therefore how far the results can be generalised. In IGCSE answers you should name the sampling method, justify it briefly, and identify one limitation of it in the given context.
目标总体是研究者想下结论的整群体，样本是实际接受测试的较小群体。抽样方法决定样本的代表性，进而决定结果能被概化到什么程度。在 IGCSE 答案中，你应点明抽样方法、简要说明理由，并在给定情境中指出它的一个局限。
Random sampling: every member of the target population has an equal chance of selection; strongest for representativeness but hard to organise in a school. | 随机抽样：目标总体中每个成员被选中的机会相同；代表性最强，但在学校中难以组织。
Opportunity sampling: whoever is available is used; quick and convenient but biased towards that particular group. | 机会抽样：谁方便就用谁；快速便捷，但会偏向该特定群体。
Volunteer sampling: participants put themselves forward; easy to recruit but volunteers tend to differ from non-volunteers. | 志愿者抽样：被试自愿报名；招募容易，但志愿者往往与非志愿者不同。
Systematic sampling: every nth person on a list is chosen; appears objective but the list order may create a pattern. | 系统抽样：按名单每隔 n 人选取；看似客观，但名单顺序可能形成规律。
Stratified sampling: the population is divided into subgroups and sampled proportionally; very representative but complex to carry out. | 分层抽样：把总体分成子群并按比例抽取；代表性很好，但实施复杂。
Sample size matters too. A sample of five participants will produce unstable results, while an unnecessarily huge sample wastes time and raises ethical questions about burdening participants. A sensible answer identifies a reasonable number, explains that it improves reliability, and acknowledges that a school-based study is usually limited by practical constraints.
样本量也很重要。五名被试的结果极不稳定；而无必要地扩大样本既浪费时间，也带来加重被试负担的伦理问题。合理的答案会给出一个恰当数量，说明它能提高信度，同时承认校内研究通常受现实条件限制。
6. Controls, standardisation and extraneous variables | 控制条件、标准化与额外变量
Good control is what separates an experiment from a casual observation. Standardisation means keeping everything except the IV as constant and identical as possible: the same instructions, the same materials, the same room, the same time of day, the same time limit, and the same way of recording responses. Every extraneous variable you leave uncontrolled is a potential alternative explanation for your results.
良好的控制是实验与随意观察的分界线。标准化意味着除自变量之外的一切尽可能保持恒定一致：同样的指导语、同样的材料、同样的房间、同样的时段、同样的时限、同样的记录方式。任何未加控制的额外变量，都可能成为结果的另一种解释。
Standardised instructions read aloud word for word, so that no group receives extra hints. | 指导语逐字朗读，保证任何一组都得不到额外提示。
Counterbalancing in repeated measures designs to neutralise order effects. | 重复测量设计中使用平衡法，以抵消顺序效应。
Random allocation in independent measures designs to spread participant variables evenly. | 独立组设计中使用随机分配，使被试变量均匀分布。
Single-blind and double-blind procedures to reduce demand characteristics and investigator effects. | 单盲与双盲程序，用来减少需求特征和研究者效应。
A control group that receives no treatment or a neutral condition, providing a baseline for comparison. | 控制组不接受处理或接受中性条件，作为比较基线。
Consistent measurement rules, such as always scoring a response correct only if the exact word is produced. | 一致的计分规则，例如只有准确说出该词才记为正确。
Some variables cannot be eliminated, only acknowledged. Individual differences in memory ability, motivation, hearing acuity or prior knowledge will always exist. In an evaluation question, the examiner wants you to say what was not controlled, explain how it could have affected the DV, and state the direction of that effect. For instance, if the noisy group was tested after lunch, drowsiness may have lowered their scores, making the noise effect look larger than it is.
有些变量无法消除，只能承认。记忆能力、动机、听力敏锐度或先前知识等个体差异永远存在。在评价题中，考官希望你说出什么没有被控制、说明它可能如何影响因变量，并指出影响方向。例如，若噪音组在午饭后测试，困倦可能降低其得分，使噪音效应看起来比实际更大。
7. Ethical considerations in school-based research | 校内研究中的伦理考量
Ethics are not decoration added at the end of an answer; they shape the design from the start. Cambridge expects you to know the core principles and to apply them to a school context, where many participants are under sixteen and therefore require parental or guardian consent as well as their own agreement.
伦理不是答案末尾的装饰，而是从设计之初就影响研究。剑桥希望你掌握核心原则，并能应用于学校情境：许多被试未满十六岁，因此除本人同意外，还需要父母或监护人同意。
Informed consent: participants must know the purpose, procedure, duration and their rights before agreeing. | 知情同意：被试在同意前必须了解目的、程序、时长和自己的权利。
Right to withdraw: participants may leave at any point, and may withdraw their data afterwards. | 退出权：被试可在任何时点退出，并可在事后撤回数据。
Confidentiality and anonymity: data are stored securely and never linked to identifiable names in the write-up. | 保密与匿名：数据安全保存，报告中不与可识别的姓名关联。
Protection from harm: no physical or psychological harm, including embarrassment or stress during memory tests. | 免受伤害：不造成身体或心理伤害，包括记忆测试中的窘迫或压力。
Debriefing: after the study, participants are told the true aim and given a chance to ask questions. | 事后说明：研究结束后告知真实目的，并给被试提问机会。
Deception: should be avoided; if used, it must be justified, minimal and always followed by a full debrief. | 欺骗：应避免；若使用，必须有正当理由、程度最小，并始终配合完整的事后说明。
A strong answer links ethics to validity as well as to morality. Participants who feel pressured may comply without effort, and participants who feel watched may behave unnaturally, producing demand characteristics that distort the DV. Ethical treatment therefore protects both the participant and the quality of the data.
优秀的答案会把伦理与效度联系起来，而不只是道德层面。感到被施压的被试可能敷衍应付；感到被注视的被试可能表现不自然，产生需求特征并扭曲因变量。因此，合乎伦理的做法既保护被试，也保护数据质量。
8. Data collection, presentation and description | 数据收集、呈现与描述
Raw data must be organised before it can be interpreted. Start with a clearly labelled table showing every participant or every condition, then choose a graph that fits the level of measurement. Bar charts are used for categorical or discrete comparisons between conditions, histograms for continuous frequency distributions, scatter graphs for relationships between two co-variables, and line graphs for trends across time or ordered conditions.
原始数据必须先整理才能解释。先用标注清楚的表格列出每位被试或每个条件，再选择与测量水平相符的图形。条形图用于类别或离散数据的条件比较，直方图用于连续变量的频数分布，散点图用于两个共变量之间的关系，折线图用于随时间或有序条件变化的趋势。
Every graph needs a title, labelled axes with units, and evenly spaced scales that start at zero where appropriate. Omitting units or starting the y-axis at an arbitrary value are common mark-losing errors because they exaggerate differences visually. If the data are ranks rather than measurements, say so, because that changes which summary statistic is appropriate.
每幅图都需要标题、带单位的坐标轴标签，以及适当的、通常从零开始的等距刻度。漏写单位，或把纵轴起点设为任意值，都会因视觉上夸大差异而常见失分。如果数据是等级而非测量值，要明确说明，因为这决定哪种汇总统计量合适。
Mean = Σx ÷ n | Range = highest value − lowest value
The mean uses all the scores but is distorted by extreme values. The median is more robust when there are outliers or when data are ordinal. The mode is the only measure of central tendency suitable for nominal data. Dispersion matters as well: two conditions may have the same mean but very different ranges, so report both a measure of central tendency and a measure of spread.
平均数使用全部数据，但受极端值影响。存在离群值或数据为顺序数据时，中位数更稳健。众数是唯一适用于称名数据的集中量数。离散程度同样重要：两个条件可能平均数相同但全距差异很大，因此应同时报告集中量数与离散量数。
9. Interpreting patterns and spotting problems | 解读数据形态与发现问题
Once the numbers are on the page, describe what you can actually see before you explain it. Say whether one condition scored higher, whether the spread overlaps, whether one participant was an outlier, and whether the difference looks large or trivial. Interpretation means linking that pattern back to the hypothesis, not just repeating the numbers in words.
数据摆出来之后，先描述你实际看到的内容，再解释。说明哪个条件得分更高、分布是否重叠、是否有被试属于离群值、差异看起来是大还是小。解读就是把该形态与假设联系起来，而不是把数字换成文字重复一遍。
Watch for ceiling and floor effects. If almost everyone scores the maximum on a memory task, the test is too easy and the IV cannot show any effect. If almost everyone scores
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