METHOD · mth_did · v1
Difference-in-Differences
双重差分
Estimates treatment effects from changes over time between treated and comparison groups under design-specific identifying assumptions.
方法身份
- 规范名称
- Difference-in-Differences
- 别名
- DiD、DID、difference in differences
- 方法族
- causal inference / panel quasi-experiment
以上名称和分类来自此版本资料卡,不代表方法之间存在已核验的关系或适用性。
适用问题
- policy/intervention effects with pre/post outcomes and a comparison group
- staggered treatment adoption with modern group-time estimators
数据条件
- repeated outcomes over time
- treated and comparison units/groups
- treatment timing observed
识别与模型假设
- parallel trends in relevant untreated potential outcomes
- no anticipation as defined by design
- stable measurement and correctly defined treatment timing
常见失败
- naive TWFE/event-study interpretation under heterogeneous staggered effects
- treating insignificant pre-trends as proof
- post-result sample/treatment changes without provenance
不能支持的结论
- causal claims without credible comparison trends
- unit-level treatment effects
- transportability beyond studied population without evidence
诊断与核对
- pre-treatment trends
- appropriate event-study diagnostics
- group-time ATT heterogeneity
- placebo/alternative comparison checks
可比较的替代方法
- Synthetic Control
- Interrupted Time Series
- RDD
- RCT
软件与实现
- R did
- Stata/R modern DiD tools
来源与参考
- Difference-in-Differences with multiple time periods
Brantly Callaway; Pedro H. C. Sant’Anna · 2021
打开原始来源 ↗
来源链接用于核对。来源可访问不代表已获全文或代码再分发授权。
对象关系
关系连接明确的对象版本。前置知识指学习或实现知识;搭配关系是编辑建议,尚未统计真实使用频率。
此对象版本尚未收录明确版本的关系。
关联研究主题
这些旧版关系没有绑定方法版本,只保留供回溯。当前版本的明确版本关系见“对象关系”;编辑建议不代表已验证适用性。
Football Governance →
Policy/league reforms may be evaluated with quasi-experimental panels when assumptions fit
编辑关联 · 待领域复核Policy Evaluation →
Core policy-evaluation design under parallel-trends assumptions
编辑关联 · 待领域复核Education Policy and Inequality →
Policy reforms over groups/time
编辑关联 · 待领域复核Labor Markets and Employment Policy →
Policy/employment timing
编辑关联 · 待领域复核