mw-identification
Use when the empirical identification strategy is the bottleneck for a Management-World manuscript — quasi-experimental designs (DID, IV, RDD, DML, event study) with Chinese policy shocks. Stress-tests the design before drafting tables.
npx skills add brycewang-stanford/Awesome-Journal-Skills --skill mw-identification --agent claude-code
Same command for any agent — swap --agent for codex, cursor, copilot.
Weekly change comes from our own snapshots, not the repository page — it measures attention, not adoption.
# 因果识别策略(mw-identification) ## 触发时机 - 实证主体仅有 OLS + 控制变量,担心被退稿 - DID 用了 TWFE 但没回应近年异质性处理批评(Goodman-Bacon, de Chaisemartin, Sun-Abraham, Callaway-Sant'Anna) - IV 第一阶段 F 弱 / 工具变量内生性疑虑 - 准备用双重机器学习但不确定怎么报告 ## 设计优先级 《管理世界》编委的偏好排序(强 → 弱): 1. **政策冲击 + DID(含 staggered / continuous treatment)** 2. **断点回归(清晰的政策门槛)** 3. **工具变量(强工具 + 排他性论证)** 4. **倾向得分匹配 + DID** 5. **合成控制法** 6. **双重机器学习 / 因果森林** 7. OLS + 内生性讨论(容易退稿) ## 分支路径 ### 分支 A:DID - 是否 staggered?→ 必须用 Goodman-Bacon 分解 + Callaway-Sant'Anna 或 Sun-Abraham - 平行趋势检验:事件研究图必须画 - 安慰剂:随机分配处理组 500–1000 次 - 是否汇报 Bacon 分解的"坏比较"权重? ### 分支 B:IV - 第一阶段 F **必须 ≥ 10**(弱工具 → 用 Anderson-Rubin 或 weak-IV-robust CI) - 排他性论证至少需要 3 段:理论 / 制度 / 安慰剂 - 是否报告了 reduced form? ### 分支 C:RDD - 是否做了 McCrary 检验? - 带宽:是否使用最优带宽(Calonico-Cattaneo-Titiunik)+ 至少 3 个带宽稳健性? - 协变量平滑性检验 ### 分支 D:DML - 报告 sample-split 数 + cross-fitting - 报告 nuisance 函数选择(lasso / random forest / xgboost) - 至少给出 5 种不同 ML 学习器的稳健性 ## 执行桥(StatsPAI / Stata MCP) 把设计**跑出来并审计**,而不是只做描述。完整映射见 [`execution-with-mcp`](../../../shared-resources/empirical-methods/execution-with-mcp.md)。《管理世界》重中国情境实证 + 政策可操作;识别 + 经济量级,定性/案例另循其标准。 - `detect_design` → `recommend` → 用 `as_handle=true` 拟合 → `audit_result` 列出尚欠的检查。 - **观
- 触发时机
- 设计优先级
- 分支路径
- 分支 A:DID
- 分支 B:IV
- 分支 C:RDD
- 分支 D:DML
- 执行桥(StatsPAI / Stata MCP)
- 必查清单
- 反模式
- 本刊识别策略决策表
- 常见退稿模式(识别维度)
- 微型走查示例:数字化转型与企业全要素生产率
- 审稿人追问模式 + 本刊语境修法
What does the mw-identification skill do?
Use when the empirical identification strategy is the bottleneck for a Management-World manuscript — quasi-experimental designs (DID, IV, RDD, DML, event study) with Chinese policy shocks. Stress-tests the design before drafting tables.
How do I install it?
Run `npx skills add brycewang-stanford/Awesome-Journal-Skills --skill mw-identification --agent claude-code` — it drops the skill into your project so the agent can pick it up. Swap the --agent value for codex, cursor or copilot if you use one of those.
Where does this skill come from?
From brycewang-stanford/Awesome-Journal-Skills, a repository with 909 stars. We read it straight from the repository tree rather than a submitted listing, so what you see here is what is actually published.
Is a popular skill a good skill?
Not necessarily. Stars measure attention, not adoption — a repository can trend for a week and be abandoned. That is why we show the weekly change from our own snapshots next to the total, instead of a single flattering number.