📚 Experiment: Design, Evidence and the Limits of Scientific Testing | 实验：设计、证据与科学检验的边界
Experiments are among the most powerful tools human beings have for finding out how the world works. By deliberately changing something and observing what follows, researchers can test hypotheses, uncover new phenomena, and gather evidence that supports or corrects existing theories. Whether the setting is academic research, engineering design or product development, the experiment remains a cornerstone of reliable knowledge.
实验是人类认识世界最有力的工具之一。通过有意识地改变某个条件并观察随之而来的结果，研究者能够检验假说、发现新现象，并获取支持或修正既有理论的证据。无论是在学术研究、工程设计还是产品开发中，实验都是可靠知识的重要基石。
1. What an Experiment Really Is | 实验究竟是什么
An experiment is a controlled procedure in which the researcher deliberately manipulates one or more factors and observes the effect on an outcome. This deliberate intervention is what separates an experiment from a purely observational study, in which the researcher records what happens without changing any condition.
实验是一种受控的研究程序，研究者有意识地操纵一个或多个因素，并观察它们对结果产生的影响。正是这种主动干预，把实验与纯粹的观察性研究区分开来：后者只记录自然发生的过程，不改变任何条件。
The logic behind every experiment is counterfactual. To know what a cause does, we need to know what would have happened without it. A comparison group or control group supplies that missing picture, standing in for the world in which the treatment was never applied.
每个实验背后的逻辑都是反事实的。要判断某个原因起到什么作用，就必须知道如果没有它会怎样。比较组或对照组正是提供这幅"缺失画面"的角色，它代表着从未施加处理的那个世界。
Experiments are not confined to laboratories. Engineers test prototypes, product teams run A/B tests, teachers compare two teaching methods. In every case the same underlying logic applies: intervene, compare, and reach a conclusion that data can support or refute.
实验并不局限于实验室。工程师测试样机，产品团队做 A/B 测试，教师比较两种教学方法。在这些场景中，底层逻辑完全相同：干预、比较，然后得出能被数据支持或推翻的结论。
2. The Core Stages of an Experiment | 实验的核心环节
A complete experiment begins with a clear objective and a testable hypothesis. A good hypothesis is falsifiable: it makes a specific prediction that observation could prove wrong. Vague aims such as "to see what happens" tend to produce results that cannot be interpreted.
一个完整的实验始于明确的目标与可检验的假说。好的假说应当是可证伪的：它给出具体预测，而这种预测有可能被观察结果推翻。诸如"看看会发生什么"这样含糊的目标，往往只能得到无法解读的结果。
The second stage is the experimental plan, or protocol: what will be measured, how many units will be tested, in what order, and under which conditions. The protocol should be written before data collection begins, so that decisions are not quietly adjusted to fit the results afterwards.
第二个环节是实验方案，也就是 protocol：测量什么、测试多少个单位、按什么顺序、在哪些条件下进行。方案应在收集数据之前写好，以免决策在事后被悄悄调整，去迎合已经得到的结果。
Data collection, analysis and interpretation follow. Analysis turns raw measurements into evidence, while interpretation asks what that evidence means for the original hypothesis and for the wider theory to which it belongs.
随后是数据收集、分析与解读。分析把原始测量转化为证据，而解读则回答：这些证据对最初的假说意味着什么，对它所属的更大理论又意味着什么。
| Stage | 环节 | Key question / 关键问题 |
| Objective and hypothesis | 目标与假说 | What exactly are we testing? 我们究竟要检验什么？ |
| Protocol design | 方案设计 | How will we control and measure? 如何控制与测量？ |
| Data collection | 数据收集 | What did we actually observe? 我们实际观察到了什么？ |
| Analysis | 数据分析 | Is the pattern real or chance? 这个规律是真实的还是偶然的？ |
| Interpretation and reporting | 解读与报告 | What does it mean? 它意味着什么？ |
3. Variables: Independent, Dependent, Controlled and Confounding | 变量：自变量、因变量、控制变量与混杂变量
The independent variable is the factor the researcher deliberately changes
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