Documenti di Didattica
Documenti di Professioni
Documenti di Cultura
† Harvard ‡ Chicago
October 1, 2018
Imai, Jiang, and Malani (HU/UC) Two-Stage Randomized Controlled Trials Harvard (October 1, 2018) 1 / 22
Methodological Motivation: Two-stage RCTs
Imai, Jiang, and Malani (HU/UC) Two-Stage Randomized Controlled Trials Harvard (October 1, 2018) 2 / 22
Empirical Motivation: Indian Health Insurance Experiment
Imai, Jiang, and Malani (HU/UC) Two-Stage Randomized Controlled Trials Harvard (October 1, 2018) 3 / 22
Study Design
Time line:
1 September 2013 – February 2014: Baseline survey
2 April – May 2015: Enrollment
3 September 2016 – January 2017: Endline survey
Imai, Jiang, and Malani (HU/UC) Two-Stage Randomized Controlled Trials Harvard (October 1, 2018) 4 / 22
Causal Inference and Interference between Units
Imai, Jiang, and Malani (HU/UC) Two-Stage Randomized Controlled Trials Harvard (October 1, 2018) 5 / 22
Two-stage Randomized Experiments
Individuals (households): i = 1, 2, . . . , N
Blocks (villages): j = 1, 2, . . . , J
Size of block j: nj where N = Jj=1 nj
P
Imai, Jiang, and Malani (HU/UC) Two-Stage Randomized Controlled Trials Harvard (October 1, 2018) 6 / 22
Intention-to-Treat Analysis: Causal Quantities of Interest
Average outcome under the treatment Zij = z and the assignment
mechanism Aj = a:
X
Y ij (z, a) = Yij (Zij = z, Z−i,j = z−i,j )Pa (Z−i,j = z−i,j | Zij = z)
z−i,j
Halloran and Struchiner (1995), Sobel (2006), Hudgens and Halloran (2008)
Imai, Jiang, and Malani (HU/UC) Two-Stage Randomized Controlled Trials Harvard (October 1, 2018) 8 / 22
Complier Average Direct Effect
Goal: Estimate the treatment effect rather than the ITT effect
Use randomized encouragement as an instrument
1 Monotonicity: Dij (1, z−i,j ) ≥ Dij (0, z−i,j ) for any z−i,j
2 Exclusion restriction: Yij (zj , dj ) = Yij (z0j , dj ) for any zj and z0j
Imai, Jiang, and Malani (HU/UC) Two-Stage Randomized Controlled Trials Harvard (October 1, 2018) 9 / 22
Key Identification Assumption
Imai, Jiang, and Malani (HU/UC) Two-Stage Randomized Controlled Trials Harvard (October 1, 2018) 10 / 22
Scenario I: No Spillover Effect of the Treatment Receipt on
the Outcome
' '
Z2j / D2j / Y2j
A @
.. .. ..
. . .
Znj j / Dn j / Yn j
j j
Imai, Jiang, and Malani (HU/UC) Two-Stage Randomized Controlled Trials Harvard (October 1, 2018) 11 / 22
Scenario II: No Spillover Effect of the Treatment
Assignment on the Treatment Receipt
Dij (zij , z−i,j ) = Dij (zij , z0−i,j ) (Kang and Imbens, 2016)
' '
Z2j / D2j / Y2j
A @
.. .. ..
. . .
Znj j / Dn j / Yn j
j j
Imai, Jiang, and Malani (HU/UC) Two-Stage Randomized Controlled Trials Harvard (October 1, 2018) 12 / 22
Scenario III: Limited Spillover Effect of the Treatment
Assignment on the Treatment Receipt
If Dij (1, z−i,j ) = Dij (0, z−i,j ) for any given z−i,j ,
then Di 0 j (1, z−i,j ) = Di 0 j (0, z−i,j ) for all i 0 6= i
Z1j / Y1j
7D
E 1j 7 E
' '
Z2j / D2j / Y2j
A @
.. .. ..
. . .
Znj j / Dn j / Yn j
j j
Imai, Jiang, and Malani (HU/UC) Two-Stage Randomized Controlled Trials Harvard (October 1, 2018) 13 / 22
Identification and Consistent Estimation
ADEY (a)
lim CADE(z, a) = lim
nj →∞ nj →∞ ADED (a)
d Y (a)
ADE p
D
−→ lim CADE(z, a)
nj →∞,J→∞
ADE
d (a)
Imai, Jiang, and Malani (HU/UC) Two-Stage Randomized Controlled Trials Harvard (October 1, 2018) 14 / 22
Randomization Inference
CADE(a)
PJ Pnj
j=1 i=1 {Yij (1, a) − Yij (0, a)}1{Dij (1, a) = 1, Dij (0, a) = 0}
= PJ Pnj
j=1 i=1 1{Dij (1, a) = 1, Dij (0, a) = 0}
The model:
1
X 1
X
Yij = αa 1{Aj = a} + βa Dij 1{Aj = a} + ij
|{z}
a=0 a=0 CADE
1
X 1
X
Dij = γa 1{Aj = a} + δa Zij 1{Aj = a} + ηij
a=0 a=0
Compliance status:
complier if Dij (1, a) = 1, Dij (0, a) = 0
Cij (a) = always − taker if Dij (1, a) = Dij (0, a) = 1
never − taker if Dij (1, a) = Dij (0, a) = 0
Imai, Jiang, and Malani (HU/UC) Two-Stage Randomized Controlled Trials Harvard (October 1, 2018) 18 / 22
No spillover effect of treatment on outcome
i.i.d.
Yij (dij = 0) ∼ N (0, 1)
indep.
Yij (1) − Yij (0) ∼ N (θj , σ 2 )
indep.
θj ∼ N (θ, ω 2 )
Imai, Jiang, and Malani (HU/UC) Two-Stage Randomized Controlled Trials Harvard (October 1, 2018) 19 / 22
Results: Both Spillover Effects Present
n=10, J=250 n=250, J=10 n=50, J=50
●
0.95
0.95
0.95
● ● ●
● ● ●
● ● ●
● ●
●
Coverage rates for CADE(1)
●
●
● ●
●
0.85
0.85
0.85
●
0.75
0.75
0.75
● Proposed ●
● HC2
Cluster
●
0.65
0.65
0.65
Cluster HC2
1.0
1.0
● ●
● ● ●
● ● ●
●
0.9
0.9
0.9
● ●
Coverage rates for CADE(0)
● ●
● ●
● ●
●
●
0.8
0.8
0.8
●
●
0.7
0.7
0.7
●
0.6
0.6
0.6
●
0.5
0.5
0.5
●
Intracluster
Imai, Jiang, and Malani correlation
(HU/UC) coefficient Intracluster correlation
Two-Stage Randomized coefficientTrials
Controlled Intracluster
Harvard (Octobercorrelation coefficient
1, 2018) 20 / 22
Results: Indian Health Insurance Experiment
Imai, Jiang, and Malani (HU/UC) Two-Stage Randomized Controlled Trials Harvard (October 1, 2018) 21 / 22
Concluding Remarks
In social science research,
1 people interact with each other interference
2 people don’t follow instructions noncompliance
Our contributions:
1 Identification condition for complier average direct effects
2 Consistent estimator for CADE and its variance
3 Connections to regression and instrumental variables
4 Application to the India health insurance experiment
5 Implementation as part of R package experiment