实验设计简报

把业务假设转化为实验：写明目标人群、干预、对照、随机化单位、主要指标、护栏指标和预先约定的决策规则。检查污染、选择偏差和新奇效应。缺少基线、样本或功效信息时列出所需输入，不猜测样本量。

输入材料：
{{context}}

目标与约束：
{{constraints}}

只依据输入回答。缺少关键信息时明确标记未知，必要时先提问。输出应可直接用于当前任务。

Translate the hypothesis into a target population, intervention, control, randomization unit, primary metric, guardrails and prespecified decision rule. Check contamination, selection bias and novelty effects. List missing baseline, sample or power inputs instead of guessing a sample size.

Input material:
{{context}}

Goal and constraints:
{{constraints}}

Base the response on the input. Mark missing information as unknown and ask essential questions when needed. Make the output directly usable for this task.