📚 Experimental Economics: A Step-by-Step Guide | 实验经济学操作指南
Experimental economics is a powerful method for testing economic theories by observing human behaviour in controlled settings. Unlike traditional field studies, laboratory experiments allow researchers to isolate variables, replicate conditions, and draw causal inferences. This guide covers the essential steps for designing, running, and analysing an economics experiment, with a focus on concepts relevant to IB and AQA syllabi, such as game theory, market mechanisms, and behavioural biases.
实验经济学是一种通过受控环境中观察人类行为来检验经济理论的有效方法。与传统的实地研究不同，实验室实验使研究者能够分离变量、复制条件并得出因果推断。本指南涵盖了设计、运行和分析经济学实验的基本步骤，重点关注与 IB 和 AQA 教学大纲相关的概念，如博弈论、市场机制和行为偏差。
1. Defining the Research Question | 定义研究问题
A well-defined research question is the foundation of any experiment. For example, ‘Do individuals cooperate more in a public goods game when they can communicate?’ or ‘Does a minimum wage reduce employment in a double auction market?’ The question should be specific, testable, and grounded in economic theory. It often stems from a gap between theoretical predictions and observed real-world behaviour.
清晰定义的研究问题是任何实验的基础。例如，“在公共物品博弈中，当个人能够交流时，他们会更多地合作吗？”或“在双向拍卖市场中，最低工资会减少就业吗？”问题应当具体、可检验，并立足于经济理论。它通常源于理论预测与观察到的现实世界行为之间的差距。
2. Reviewing the Literature and Theory | 回顾文献与理论
Before designing an experiment, review existing experimental studies and the theoretical model you plan to test. This helps you identify appropriate treatments, control variables, and expected outcomes. For instance, if you study the ultimatum game, familiarise yourself with the subgame perfect equilibrium prediction and known behavioural results, such as rejections of low offers.
在设计实验之前，回顾已有的实验研究和你要检验的理论模型。这有助于你确定合适的处理方式、控制变量和预期结果。例如，如果你研究最后通牒博弈，就要熟悉子博弈完美均衡的预测以及已知的行为结果，比如低出价被拒绝的现象。
3. Choosing the Experimental Design | 选择实验设计
Decide between a between-subjects design (different participants in each treatment) or a within-subjects design (the same participants experience all treatments). Between-subjects avoids order effects, but within-subjects requires fewer participants and controls for individual differences. Also determine the number of treatments and the control condition. For example, a baseline treatment without communication versus a treatment with chat communication.
决定使用被试间设计（每个处理组使用不同参与者）还是被试内设计（同样参与者经历所有处理）。被试间设计避免了顺序效应，而被试内设计所需参与者更少，并可以控制个体差异。还要确定处理数量和对照条件。例如，无沟通的基线处理与有聊天沟通的处理进行比较。
4. Participant Recruitment and Incentives | 参与者招募与激励
Recruit participants who are naïve to the experimental purpose. In student experiments, class peers can serve as subjects, but ensure they have not previously taken part in a similar study. Monetary incentives are crucial: participants should be paid based on their decisions to align their interests with the theoretical payoff structure. A typical payment includes a show-up fee plus earnings from the game, converted at a pre-announced rate.
招募对实验目的一无所知的参与者。在学生实验中，同班同学可以作为被试，但要确保他们之前未参加过类似研究。金钱激励至关重要：应根据参与者的决定支付报酬，使其利益与理论收益结构一致。典型的报酬包括出场费加上从博弈中获得的收益，并按预先公布的比例兑换。
5. Developing the Experimental Materials | 制作实验材料
Prepare clear instructions, consent forms, decision sheets (for paper-based experiments) or program the experiment using software like oTree or z-Tree. Instructions must be neutral in wording, avoiding demand effects. Use examples and control questions to check understanding. For a paper-based public goods game, design a table where participants can record their contributions and calculate payoffs.
准备清晰的指导语、知情同意书、决策表格（用于纸笔实验）或使用 oTree、z-Tree 等软件编写程序。指导语必须措辞中性，避免需求效应。使用示例和控制性问题检查理解情况。对于纸笔版的公共物品博弈，设计一个表格，让参与者记录他们的出资并计算收益。
6. Conducting a Pilot Test | 进行预实验
Run a pilot session with a small sample to identify flaws in the design, instructions, or technical setup. Observe how long the experiment takes, whether participants understand the tasks, and if any strategic confusion arises. Adjust the procedures accordingly. A pilot also provides a rough estimate of the variance for power analysis if statistical testing is planned.
用小样本进行一次预实验，找出设计、指导语或技术设置中的缺陷。观察实验所需的时间、参与者是否理解任务，以及是否出现任何策略混淆。根据情况调整步骤。如果计划进行统计检验，预实验还可以为功效分析提供方差的粗略估计。
7. Implementing Randomisation and Anonymity | 实施随机化与匿名性
Randomly assign participants to treatments and roles (e.g., buyer/seller, proposer/responder). Anonymity is vital to minimize social desirability bias; use identification numbers rather than names, and ensure decisions cannot be traced back to individuals. In computerised experiments, network isolation prevents participants from knowing who they are interacting with.
随机分配参与者到不同处理组和角色（如买方/卖方、提议者/响应者）。匿名性对于减少社会期望偏差至关重要；使用身份编号而非姓名，并确保决策无法追溯到个人。在计算机化的实验中，网络隔离可防止参与者知道他们在与谁互动。
8. Running the Experiment Session | 运行实验环节
On the day of the experiment, read the instructions aloud to ensure consistency. Allow time for questions, and then let participants make their decisions independently. Monitor the room to prevent communication unless it is part of the treatment. Record all decisions accurately. For a double auction market, the experimenter may need to facilitate trading rounds orally or via software.
在实验当天，大声朗读指导语以确保一致性。留出提问时间，然后让参与者独立做出决策。监控房间，防止未被允许的交流。准确记录所有决策。对于双向拍卖市场，实验者可能需要口头主持交易回合或通过软件进行。
9. Debriefing and Payment | 事后说明与支付
After the experiment, explain the true purpose of the study, reveal any deception (which should be avoided whenever possible in economics), and answer participants’ questions. Pay participants privately according to their accumulated earnings. This debriefing stage is also an opportunity to gather qualitative feedback about their thought process during the experiment.
实验结束后，解释研究的真实目的，披露任何欺骗手段（经济学中应尽量避免使用欺骗），并回答参与者的问题。根据参与者累计的收益私下支付报酬。这个事后说明阶段也是收集参与者实验过程中思维过程的定性反馈的机会。
10. Data Analysis and Statistical Testing | 数据分析与统计检验
Enter the data into a spreadsheet or statistical software. Compute average behaviours per treatment and compare them using appropriate tests. For parametric data, an independent-samples t‑test (between subjects) or paired t‑test (within subjects) is common. For non-parametric data, use the Mann‑Whitney U‑test or Wilcoxon signed‑rank test. Report effect sizes and confidence intervals alongside p‑values. For example, comparing the mean contribution in a public goods game across treatments: meantreatment = 12.5, meancontrol = 8.3, t(40) = 3.21, p < 0.01.
将数据输入电子表格或统计软件。计算每个处理组的平均行为，并使用适当的检验进行比较。对于参数数据，通常使用独立样本 t 检验（被试间）或配对 t 检验（被试内）。对于非参数数据，使用曼‑惠特尼 U 检验或威尔科克森符号秩检验。在报告 p 值的同时报告效应量和置信区间。例如，比较公共物品博弈中各处理组的平均出资额：处理组均值 = 12.5，对照组均值 = 8.3，t(40) = 3.21，p < 0.01。
11. Interpreting Results and Drawing Conclusions | 解释结果并得出结论
Link your findings back to the economic theory or policy question. If behaviour deviates from rational choice predictions, discuss potential behavioural explanations such as bounded rationality, social preferences, or loss aversion. Acknowledge limitations like sample size, student subjects, or artificiality of the lab setting. Suggest how future research could extend your design, for example by varying the framing or payoff structure.
将你的发现与经济理论或政策问题联系起来。如果行为偏离了理性选择预测，讨论可能的行为解释，如有限理性、社会偏好或损失厌恶。承认诸如样本量、学生被试或实验室环境的人为性等局限性。建议未来的研究如何扩展你的设计，例如通过改变框架或收益结构。
12. Ethical Considerations in Economic Experiments | 经济学实验中的伦理考量
Economic experiments must adhere to ethical guidelines: informed consent, voluntary participation, confidentiality, and the right to withdraw at any time. Unlike psychology, economics usually avoids deception because it can undermine participant trust and contaminate future experiments involving the same subject pool. Ensure that the payments offered are fair for the time spent, and that no psychological distress is caused by the tasks.
经济学实验必须遵循伦理准则：知情同意、自愿参与、保密以及随时退出的权利。与心理学不同，经济学通常避免欺骗，因为它会破坏参与者信任并污染未来涉及同一被试库的实验。确保提供的报酬与花费的时间相称，并且任务不会造成心理困扰。
Published by TutorHao | Economics 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