基础专项：总体平均与群体差异

唯一新困难：群体差异

总体平均值可能掩盖不同群体的相反反应。题目若限定新手、专家、低收入者或某年龄组，不能拿全体数据替代。本课只训练“总体—分组”两层证据。

自拟短文

A company tested optional quiet hours in its open office. Overall productivity changed little. Employees doing routine administrative work reported no clear benefit, while programmers working on complex tasks completed fewer corrections and rated their concentration higher. New employees, however, said the quiet period made it harder to ask quick questions.

题目：Who benefited most clearly from quiet hours?

A. All employees equally.
B. Staff doing routine administrative work.
C. Programmers handling complex tasks.
D. New employees seeking quick help.

两层读取

总体层：productivity changed little。分组层：行政无明确收益；复杂任务程序员返工少、专注高；新员工沟通受阻。因此 C。

A 用总体平均掩盖组间差异；B 与 no clear benefit 冲突；D 把负面体验翻成收益。

平均值陷阱

“总体变化小”不等于“每组都没变化”。一组改善、一组受损可能在平均值中抵消。反过来，某组受益也不能推出政策适合所有人。

同义改写

completed fewer corrections 表示需要返工或修正的次数减少，可改写为 made fewer revisions；不能改成 completed fewer tasks。

独立练习

若公司只公布 overall productivity，遗漏了什么？

**答案：**不同工作类型和资历群体的异质效果，特别是程序员收益与新员工沟通成本。

独立动作验收

画“总体+三个群体”表格，分别写方向和证据。只回答 C 而不能解释平均值陷阱不通过。