Matching and Regression to the Mean in Difference-in-Differences Analysis
被引:152
|
作者:
Daw, Jamie R.
论文数: 0引用数: 0
h-index: 0
机构:
Harvard Med Sch, Dept Hlth Care Policy, 180 Longwood Ave, Boston, MA 02115 USAHarvard Med Sch, Dept Hlth Care Policy, 180 Longwood Ave, Boston, MA 02115 USA
Daw, Jamie R.
[1
]
Hatfield, Laura A.
论文数: 0引用数: 0
h-index: 0
机构:
Harvard Med Sch, Dept Hlth Care Policy, 180 Longwood Ave, Boston, MA 02115 USAHarvard Med Sch, Dept Hlth Care Policy, 180 Longwood Ave, Boston, MA 02115 USA
Hatfield, Laura A.
[1
]
机构:
[1] Harvard Med Sch, Dept Hlth Care Policy, 180 Longwood Ave, Boston, MA 02115 USA
Objective Data Sources To demonstrate regression to the mean bias introduced by matching on preperiod variables in difference-in-differences studies. Simulated data. Study Design Principal Findings We performed a Monte Carlo simulation to estimate the effect of a placebo intervention on simulated longitudinal data for units in treatment and control groups using unmatched and matched difference-in-differences analyses. We varied the preperiod level and trend differences between the treatment and control groups, and the serial correlation of the matching variables. We assessed estimator bias as the mean absolute deviation of estimated program effects from the true value of zero. When preperiod outcome level is correlated with treatment assignment, an unmatched analysis is unbiased, but matching units on preperiod outcome levels produces biased estimates. The bias increases with greater preperiod level differences and weaker serial correlation in the outcome. This problem extends to matching on preperiod level of a time-varying covariate. When treatment assignment is correlated with preperiod trend only, the unmatched analysis is biased, and matching units on preperiod level or trend does not introduce additional bias. Conclusions Researchers should be aware of the threat of regression to the mean when constructing matched samples for difference-in-differences. We provide guidance on when to incorporate matching in this study design.
机构:
Stanford Univ, Grad Sch Business, Stanford, CA 94305 USA
NBER, Cambridge, MA 02138 USAStanford Univ, Grad Sch Business, Stanford, CA 94305 USA
Athey, Susan
Imbens, Guido W.
论文数: 0引用数: 0
h-index: 0
机构:
Stanford Univ, Grad Sch Business, Stanford, CA 94305 USA
NBER, Cambridge, MA 02138 USA
Stanford Univ, Dept Econ, Stanford, CA 94305 USA
SIEPR, Stanford, CA USAStanford Univ, Grad Sch Business, Stanford, CA 94305 USA
机构:
Harbor UCLA Med Ctr, Los Angeles Biomed Res Inst, Div Med Oncol & Hematol, Torrance, CA 90509 USADrexel Univ, Dept Epidemiol & Biostat, Dornsife Sch Publ Hlth, 3215 Market St, Philadelphia, PA 19104 USA
Pan, Kathy
Seguin-Fowler, Rebecca A.
论文数: 0引用数: 0
h-index: 0
机构:
Texas A&M Univ, Coll Agr & Life Sci, College Stn, TX USADrexel Univ, Dept Epidemiol & Biostat, Dornsife Sch Publ Hlth, 3215 Market St, Philadelphia, PA 19104 USA
机构:
Univ Sao Paulo, Dept Econ, Av Luciano Gualberto 908, Sao Paulo, SP, BrazilUniv Sao Paulo, Dept Econ, Av Luciano Gualberto 908, Sao Paulo, SP, Brazil
Chagas, Andre L. S.
Azzoni, Carlos R.
论文数: 0引用数: 0
h-index: 0
机构:
Univ Sao Paulo, Dept Econ, Av Luciano Gualberto 908, Sao Paulo, SP, BrazilUniv Sao Paulo, Dept Econ, Av Luciano Gualberto 908, Sao Paulo, SP, Brazil
Azzoni, Carlos R.
Almeida, Alexandre N.
论文数: 0引用数: 0
h-index: 0
机构:
Esalq Univ Sao Paulo, Dept Appl Econ, Ave Padua Dias 11, Piracicaba, SP, BrazilUniv Sao Paulo, Dept Econ, Av Luciano Gualberto 908, Sao Paulo, SP, Brazil