📚 Designing and Analyzing Animal Behavior Experiments | 动物行为学实验设计与分析
Animal behaviour experiments allow psychologists to investigate the biological and environmental bases of action, learning, memory and emotion under controlled or naturalistic conditions. A well-designed ethological study requires a clear research question, careful choice of observation methods, precise operational definitions and appropriate statistical analysis. This article provides a structured guide to designing, conducting and analysing animal behaviour experiments, with a focus on the principles commonly tested in A-level and IB psychology courses.
动物行为实验使心理学家能够在受控或自然条件下研究行为、学习、记忆和情绪的生物学及环境基础。一项设计良好的行为学研究需要明确的研究问题、谨慎选择观察方法、精确的操作性定义以及恰当的统计分析。本文提供了一份关于设计、开展和分析动物行为实验的结构化指南，重点涵盖 A-level 和 IB 心理学课程中常考的原理。
1. Ethological Context and Research Questions | 行为学背景与研究问题
Ethology is the biological study of behaviour in natural environments, while comparative psychology often emphasises laboratory studies of learning and cognition. Before designing an experiment, researchers must formulate a precise research question that can be tested empirically. For example, ‘Do rats show a preference for a familiar versus a novel object?’ is more suitable than ‘Do rats have memory?’ because the former specifies variables and a measurable outcome.
行为学是在自然环境中对行为进行生物学研究，而比较心理学则更强调在学习与认知方面的实验室研究。在设计实验之前，研究者必须提出一个可以用实证检验的精确研究问题。例如，“大鼠是否表现出对熟悉物体而非新奇物体的偏好？”比“大鼠有记忆吗？”更合适，因为前者明确了变量和可测量的结果。
A good research question emerges from existing theory, previous observational data or an applied issue such as the effects of enrichment on captive animals. It should identify the species, the behaviour of interest, the conditions to be compared and the expected direction of effect, even if the hypothesis is non-directional.
好的研究问题来源于现有理论、先前的观察数据或应用性问题（如丰容对圈养动物的影响）。它应当指明物种、感兴趣的行为、待比较的条件以及预期的效应方向，即使假设是非方向性的。
2. Naturalistic Observation versus Controlled Experiments | 自然观察与受控实验
Naturalistic observation provides high external validity because behaviour occurs in the animal’s normal environment. However, the researcher has little control over extraneous variables, and causal conclusions are difficult to establish. In contrast, controlled laboratory experiments allow manipulation of an independent variable and random allocation of subjects, increasing internal validity but potentially reducing ecological realism.
自然观察具有较高的外部效度，因为行为发生在动物正常的环境中。然而，研究者对额外变量的控制很少，因果关系难以确定。相比之下，受控实验室实验可以操纵自变量并随机分配被试，从而提高了内部效度，但可能降低生态真实性。
The choice between these approaches depends on the aim of the study. If the goal is to describe behaviour patterns, observation is appropriate. If the goal is to test whether a specific stimulus causes a change in behaviour, an experiment is needed. Many modern studies use a hybrid design: manipulating a variable while maintaining a semi-natural enclosure.
两种方法的选择取决于研究目的。如果目标是描述行为模式，采用观察法比较合适。如果目标是检验特定刺激是否引起行为变化，则需要进行实验。许多现代研究采用混合设计：在维持半自然环境的同时操纵变量。
3. Measuring Behaviour: Ethograms and Sampling Methods | 行为测量：行为谱与取样方法
An ethogram is an exhaustive catalogue of species-specific behaviours, each defined objectively. For example, ‘rearing’ may be defined as ‘standing on hind legs with both front paws off the floor’. Operational definitions reduce ambiguity and enable reliable scoring by multiple observers.
行为谱是物种特异性行为的详尽目录，其中每个行为都有客观定义。例如，“直立”可以定义为“用后腿站立，两个前爪离开地面”。操作性定义减少了模糊性，并使得多个观察者能够可靠评分。
Common sampling methods include ad libitum (recording everything visible), focal sampling (observing one individual for a set period), scan sampling (recording the behaviour of all individuals at regular intervals) and all-occurrence sampling (recording every instance of a particular behaviour). Each method has advantages and disadvantages depending on whether the target behaviour is frequent, rare, brief or prolonged.
常见的取样方法包括：全面取样（记录所有可观察到的行为）、焦点取样（在设定时间内观察一个个体）、扫描取样（按固定时间间隔记录所有个体的行为）以及全事件取样（记录某一特定行为的每一次发生）。每种方法各有优缺点，具体取决于目标行为是频繁、罕见、短暂还是持久。
Table 1. Sampling method selection
| Method | Best for | Main limitation |
| Ad libitum | Pilot observation | Bias towards conspicuous behaviour |
| Focal | Individual time budgets | Time-consuming |
| Scan | Group activity patterns | Misses brief behaviours |
| All-occurrence | Rare or important events | Difficult in large groups |
4. Variables and Experimental Control | 变量与实验控制
The independent variable (IV) is the factor manipulated by the researcher, for example the dose of a drug or the presence of a predator cue. The dependent variable (DV) is the measured behaviour, such as time spent freezing or frequency of exploration. Extraneous variables are other factors that could influence the DV, including temperature, light cycle, handling, cage size and noise.
自变量（IV）是研究者操纵的因素，例如药物剂量或捕食者气味的存在与否。因变量（DV）是测量的行为，如冻结时间或探索频率。额外变量是可能影响因变量的其他因素，包括温度、光周期、抓取方式、笼子大小和噪音。
Controls include keeping environmental conditions constant, using a control group that receives no treatment or a placebo, and counterbalancing the order of conditions in repeated-measures designs. Random allocation to groups helps distribute unknown individual differences such as age, weight or prior experience evenly across conditions.
控制方法包括保持环境条件恒定、设置不接受处理或接受安慰剂的对照组，以及在重复测量设计中平衡条件顺序。随机分配被试到各组有助于使年龄、体重或先前经验等未知个体差异在各条件间均匀分布。
5. Equipment and Test Procedures | 设备与测试程序
Standard laboratory apparatus include the open field, the elevated plus maze, the T-maze, the Morris water maze and operant conditioning chambers. Each test is designed to measure a specific behavioural construct. For example, the open field measures locomotion and anxiety-related thigmotaxis, while the T-maze measures spatial memory and alternation.
标准实验室设备包括旷场实验箱、高架十字迷宫、T 迷宫、莫里斯水迷宫和操作条件反射箱。每项测试旨在测量特定行为构念。例如，旷场实验测量运动能力和与焦虑相关的趋触性，而 T 迷宫测量空间记忆和交替行为。
In a typical T-maze experiment, the animal is placed at the start arm and allowed to choose between the left and right arms after a delay. The DV may be the number of correct choices, latency to reach the reward, or percentage of alternation on consecutive trials. The apparatus must be cleaned between trials to remove olfactory cues that could confound the results.
在典型的 T 迷宫实验中，动物被放置在起始臂，并允许在延迟后选择左臂或右臂。因变量可以是正确选择的次数、到达奖励的潜伏期，或在连续试验中的交替百分比。设备必须在每次试验之间清洗，以去除可能干扰结果的气味线索。
Choice accuracy = (Number of correct trials ÷ Total trials) × 100%
6. Habituation and Pilot Studies | 习惯化与预实验
Habituation is the process of exposing the animal to the apparatus, handler and experimental room before testing begins. This reduces the novelty-related stress that can distort baseline behaviour. For example, rats placed in an open field for 30 minutes on three consecutive days become less active as they habituate; this baseline should be stable before drug treatment or behavioural testing.
习惯化是指在正式测试开始前让动物熟悉设备、实验操作者和实验房间的过程。这能减少可能扭曲基线行为的新奇相关应激。例如，大鼠连续三天在旷场中放置 30 分钟后，随着习惯化其活动量逐渐下降；基线应在药物处理或行为测试前保持稳定。
Pilot studies are small-scale trials used to identify practical problems, such as ambiguous ethogram definitions, inappropriate test duration, or insufficient reward motivation. Pilot data can also be used to estimate effect size and sample size. A good pilot study improves the efficiency of the main experiment and prevents waste of animals and resources.
预实验是小规模试验，用于发现实际问题，例如行为谱定义模糊、测试持续时间不当或奖赏动机不足。预实验数据也可用于估算效应量和样本量。良好的预实验能提高主实验的效率，避免浪费动物和资源。
7. Reliability in Behavioural Measurement | 行为测量的信度
Reliability refers to the consistency of measurement. Inter-observer reliability is the degree to which two independent observers agree in their ratings of the same behaviour. It is usually quantified using percentage agreement or Cohen’s kappa. A kappa value above 0.75 is generally considered excellent, while values below 0.40 suggest poor agreement.
信度指测量的一致性。观察者间信度是两位独立观察者对同一行为评分的一致性程度，通常用百分比一致性或 Cohen 的 κ 系数量化。κ 值大于 0.75 通常被认为极佳，低于 0.40 则表明一致性较差。
Test-retest reliability involves administering the same test to the same animals on separate occasions and correlating the scores. However, repeated exposure may cause learning or habituation, so researchers must distinguish genuine stability from practice effects. Clear operational definitions, structured training sessions and video recording all help improve reliability.
重测信度是在不同时间对同一动物施测相同测试并计算分数相关。然而重复暴露可能引起学习或习惯化，所以研究者必须区分真实稳定性与练习效应。清晰的操作性定义、结构化训练课程和视频录制都有助于提高信度。
8. Validity in Animal Studies | 动物研究的效度
Internal validity is the extent to which the observed effect on the DV is caused by the IV rather than by confounds. Threats include order effects, experimenter bias, environmental changes and differential attrition. Blinding the observer to the treatment condition can reduce confirmation bias, and automated tracking software can provide objective measurements.
内部效度是指观察到的因变量效应确实由自变量而非混杂因素引起的程度。其威胁包括顺序效应、实验者偏差、环境变化和不同组别的流失。对观察者隐藏处理条件可减少确认偏差，而自动追踪软件可提供客观测量。
External validity is the generalisability of findings to other species, settings and individuals. Construct validity concerns whether the test truly measures the intended psychological concept. For example, immobility in the forced swim test is interpreted as ‘behavioural despair’, but some researchers argue it may instead reflect adaptation to an inescapable situation, which challenges construct validity.
外部效度是研究结果向其他物种、情境和个体推广的程度。构念效度涉及测试是否真正测量了预期的心理概念。例如，强迫游泳测试中的不动状态被解释为“行为绝望”，但一些研究者认为它可能反映了对不可逃脱情境的适应，这挑战了构念效度。
9. Ethical and Legal Considerations | 伦理与法律考量
All animal research must comply with national and institutional regulations, such as the Animals (Scientific Procedures) Act 1986 in the UK or the NIH Guide in the US. The principles of the 3Rs — Replacement, Reduction and Refinement — are central. Researchers should consider whether a non-animal model could be used, whether the minimum number of animals can achieve statistical power, and whether procedures can be made less painful or stressful.
所有动物研究必须遵守国家和机构法规，如英国《1986 年动物（科学程序）法》或美国 NIH 指南。3R 原则——替代、减少、优化——是核心。研究者应考虑是否可以使用非动物模型、是否能使用最少的动物数量以达到统计效力，以及是否能使程序减轻疼痛或应激。
Ethical review boards require detailed protocols describing housing, enrichment, food and water schedules, painful procedures, humane endpoints and euthanasia methods. Investigators must be trained, and animals must be monitored regularly. Even when an experiment is technically legal, psychologists must justify the scientific value against the potential suffering of the animals.
伦理审查委员会要求提供详细方案，描述饲养、丰容、食物和水的时间表、疼痛性操作、人道终点和安乐死方法。研究者必须接受培训，并且动物必须定期监测。即使在技术上合法，心理学家也必须权衡科学价值与动物可能遭受的痛苦。
10. Statistical Analysis of Behavioural Data | 行为数据的统计分析
The choice of statistical test depends on the number of groups, the level of measurement and whether the data meet the assumptions of normality and homogeneity of variance. Parametric tests such as the independent t-test and one-way ANOVA are used for normally distributed, interval or ratio data. Non-parametric alternatives include the Mann-Whitney U test and the Kruskal-Wallis H test for ordinal or skewed data.
统计检验的选择取决于组数、测量水平以及数据是否满足正态性和方差齐性假设。参数检验（如独立样本 t 检验和单因素方差分析）用于正态分布的等距或等比数据。非参数替代检验包括 Mann-Whitney U 检验和 Kruskal-Wallis H 检验，用于顺序数据或偏态数据。
χ² = Σ [(Observed − Expected)² ÷ Expected]
The chi-square test is appropriate for frequency data, such as the number of rats choosing the left versus the right arm in a T-maze. A significant result indicates that the observed distribution differs significantly from chance expectation. Effect sizes, such as Cohen’s d or eta-squared, should be reported alongside p-values to show the magnitude of the effect.
卡方检验适用于频率数据，例如在 T 迷宫中选择左臂与右臂的大鼠数量。显著性结果表示观察到的分布与随机预期显著不同。应同时报告效应量（如 Cohen 的 d 或 η²）与 p 值，以显示效应的大小。
Table 2. Example test selection
| Design | Data type | Recommended test |
| Two independent groups | Interval/ratio, normal | Independent t-test |
| Two independent groups | Ordinal/skewed | Mann-Whitney U |
| More than two groups | Interval/ratio, normal | One-way ANOVA |
| Frequency counts | Categorical | Chi-square goodness of fit |
11. Interpreting Findings and Avoiding Common Pitfalls | 结果解释与常见误区
A significant statistical result does not automatically imply a large or practical effect. Researchers must examine the pattern of group means, confidence intervals and effect sizes before forming conclusions. It is also important to differentiate between statistical significance and biological importance; a subtle change in locomotion may be statistically reliable but not relevant to the animal’s welfare or survival.
统计结果显著并不自动意味着效应很大或具有实用价值。研究者必须先检查组均值的模式、置信区间和效应量，再得出结论。同时必须区分统计显著性与生物学重要性；运动量的微小变化可能在统计上可靠，但与动物的福利或生存并不相关。
Common pitfalls include pseudoreplication (treating repeated measurements from the same animal as independent), observer bias, equipment drift, order effects and overgeneralising across species. For instance, results in laboratory rats should not be freely applied to all rodents without evidence. Critical evaluation of these threats is an essential skill in the psychology examination.
常见误区包括伪重复（将来自同一动物的重复测量视为独立数据）、观察者偏差、设备漂移、顺序效应以及跨物种过度概括。例如，实验室大鼠的结果不应在没有证据的情况下任意推广到所有啮齿动物。批判性评估这些威胁是心理学考试中的重要技能。
12. Conclusion: Rigour, Replication and Communication | 结论：严谨、重复与沟通
Animal behaviour experiments are powerful tools for testing biological and psychological mechanisms, but their success depends on careful design, reliable measurement, ethical practice and appropriate statistics. A rigorous study should clearly state its hypotheses, define behaviours, control extraneous variables, use appropriate sampling methods and report effect sizes. Replication with different populations, environments and laboratories strengthens the generalisability of findings.
动物行为实验是检验生物与心理机制的强大工具，但其成功取决于精心设计、可靠测量、伦理实践和适当统计。一项严谨的研究应当清楚说明假设、定义行为、控制额外变量、使用适当的取样方法并报告效应量。在不同群体、环境和实验室中进行重复研究可加强结果的可推广性。
Finally, results must be communicated transparently, including limitations and the precise conditions under which the experiment was conducted. By following both scientific and ethical standards, psychologists can produce meaningful knowledge that informs the treatment of animals and the understanding of human behaviour.
最后，结果必须透明地传达，包括局限性和实验进行的精确条件。通过遵循科学与伦理标准，心理学家可以产生有意义的知识，既为动物福利提供指导，也加深对人类行为的理解。
Published by TutorHao | Psychology 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Cancel reply