SlideShare uma empresa Scribd logo
1 de 326
Peter M. Lance, PhD
MEASURE Evaluation
University of North Carolina at
Chapel Hill
September 13 and 15, 2016
Selection on Observables
Global, five-year, $180M cooperative agreement
Strategic objective:
To strengthen health information systems – the
capacity to gather, interpret, and use data – so
countries can make better decisions and sustain good
health outcomes over time.
Project overview
Improved country capacity to manage health
information systems, resources, and staff
Strengthened collection, analysis, and use of
routine health data
Methods, tools, and approaches improved and
applied to address health information challenges
and gaps
Increased capacity for rigorous evaluation
Phase IV Results Framework
Global footprint (more than 25 countries)
β€’ The program impact evaluation challenge
β€’ Randomization
β€’ Selection on observables
β€’ Within estimators
β€’ Instrumental variables
β€’ The program impact evaluation challenge
β€’ Randomization
β€’ Selection on observables
β€’ Within estimators
β€’ Instrumental variables
X
Y
P
X
Y
P
π‘Œ = 𝑃 βˆ™ π‘Œ1
+ 1 βˆ’ 𝑃 βˆ™ π‘Œ0
X
Y
P
X
Y
P
𝐸 π‘Œ1
βˆ’ π‘Œ0
= 𝐸 π‘Œ1 βˆ’ 𝐸 π‘Œ0
Average Y
across sample
of participants
βˆ’
Average Y
across sample
of nonβˆ’participants
𝐸 π‘Œ1
βˆ’ π‘Œ0
= 𝐸 π‘Œ1
βˆ’ 𝐸 π‘Œ0
Average Y
across sample
of participants
βˆ’
Average Y
across sample
of nonβˆ’participants
𝐸 π‘Œ1
βˆ’ π‘Œ0
= 𝐸 π‘Œ1
βˆ’ 𝐸 π‘Œ0
Average Y
across sample
of participants
βˆ’
Average Y
across sample
of nonβˆ’participants
𝐸 π‘Œ1
βˆ’ π‘Œ0
= 𝐸 π‘Œ1
βˆ’ 𝐸 π‘Œ0
Average Y
across sample
of participants
βˆ’
Average Y
across sample
of nonβˆ’participants
𝐸 π‘Œ1
βˆ’ π‘Œ0
= 𝐸 π‘Œ1
βˆ’ 𝐸 π‘Œ0
Average Y
across sample
of participants
βˆ’
Average Y
across sample
of nonβˆ’participants
𝐸 π‘Œ1
βˆ’ π‘Œ0
= 𝐸 π‘Œ1
βˆ’ 𝐸 π‘Œ0
Average Y
across sample
of participants
βˆ’
Average Y
across sample
of nonβˆ’participants
𝐸 π‘Œ1
βˆ’ π‘Œ0
= 𝐸 π‘Œ1
βˆ’ 𝐸 π‘Œ0
Average Y
across sample
of participants
βˆ’
Average Y
across sample
of nonβˆ’participants
𝐸 π‘Œ1 𝑃 = 1 = 𝐸 π‘Œ1 𝑃 = 0
= 𝐸 π‘Œ1
X
Y
P
𝐸 𝑋 𝑃 = 1 β‰  𝐸 𝑋 𝑃 = 0
X
Y
P
X
Y
P
𝐸 𝑋 𝑃 = 1 β‰  𝐸 𝑋 𝑃 = 0
𝐸 𝑋 𝑃 = 1 β‰  𝐸 𝑋 𝑃 = 0
𝐸 π‘Œ1 𝑃 = 1 β‰  𝐸 π‘Œ1 𝑃 = 0
𝐸 𝑋 𝑃 = 1 β‰  𝐸 𝑋 𝑃 = 0
𝐸 π‘Œ0 𝑃 = 1 β‰  𝐸 π‘Œ0 𝑃 = 0
𝐸 π‘Œ1
βˆ’ π‘Œ0
= 𝐸 π‘Œ1
βˆ’ 𝐸 π‘Œ0
Average Y
across sample
of participants
βˆ’
Average Y
across sample
of nonβˆ’participants
𝐸 π‘Œ1
𝑃 = 1 β‰  𝐸 π‘Œ1
𝑃 = 0
𝑋 = π‘₯βˆ—
𝐸 π‘Œ1
𝑃 = 1, 𝑋 = π‘₯βˆ—
= 𝐸 π‘Œ1
𝑃 = 0, 𝑋 = π‘₯βˆ—
𝐸 π‘Œ0
𝑃 = 1, 𝑋 = π‘₯βˆ—
= 𝐸 π‘Œ0
𝑃 = 0, 𝑋 = π‘₯βˆ—
𝐸 π‘Œ1
𝑃 = 1 β‰  𝐸 π‘Œ1
𝑃 = 0
𝑋 = π‘₯βˆ—
𝐸 π‘Œ1
𝑃 = 1, 𝑋 = π‘₯βˆ—
= 𝐸 π‘Œ1
𝑃 = 0, 𝑋 = π‘₯βˆ—
𝐸 π‘Œ0
𝑃 = 1, 𝑋 = π‘₯βˆ—
= 𝐸 π‘Œ0
𝑃 = 0, 𝑋 = π‘₯βˆ—
X
Y
P
𝐸 π‘Œ1
𝑃 = 1 β‰  𝐸 π‘Œ1
𝑃 = 0
𝑋 = π‘₯βˆ—
𝐸 π‘Œ1
𝑃 = 1, 𝑋 = π‘₯βˆ—
= 𝐸 π‘Œ1
𝑃 = 0, 𝑋 = π‘₯βˆ—
𝐸 π‘Œ0
𝑃 = 1, 𝑋 = π‘₯βˆ—
= 𝐸 π‘Œ0
𝑃 = 0, 𝑋 = π‘₯βˆ—
𝐸 π‘Œ1
𝑃 = 1 β‰  𝐸 π‘Œ1
𝑃 = 0
𝑋 = π‘₯βˆ—
𝐸 π‘Œ1
𝑃 = 1, 𝑋 = π‘₯βˆ—
= 𝐸 π‘Œ1
𝑃 = 0, 𝑋 = π‘₯βˆ—
𝐸 π‘Œ0
𝑃 = 1, 𝑋 = π‘₯βˆ—
= 𝐸 π‘Œ0
𝑃 = 0, 𝑋 = π‘₯βˆ—
𝐸 π‘Œ1
𝑃 = 1 β‰  𝐸 π‘Œ1
𝑃 = 0
𝑋 = π‘₯βˆ—
𝐸 π‘Œ1
𝑃 = 1, 𝑋 = π‘₯βˆ—
= 𝐸 π‘Œ1
𝑃 = 0, 𝑋 = π‘₯βˆ—
𝐸 π‘Œ0
𝑃 = 1, 𝑋 = π‘₯βˆ—
= 𝐸 π‘Œ0
𝑃 = 0, 𝑋 = π‘₯βˆ—
𝐸 π‘Œ1
βˆ’ π‘Œ0
= 𝐸 π‘Œ1
βˆ’ 𝐸 π‘Œ0
Average Y
across sample
of participants
βˆ’
Average Y
across sample
of nonβˆ’participants
𝐸 π‘Œ1
βˆ’ π‘Œ0
|𝑋 = π‘₯βˆ—
= 𝐸 π‘Œ1
|𝑋 = π‘₯βˆ—
βˆ’ 𝐸 π‘Œ0
|𝑋 = π‘₯βˆ—
Average Y
across sample
of participants
for whom
𝑋 = π‘₯βˆ—
βˆ’
Average Y
across sample
of participants
of nonβˆ’participants
𝑋 = π‘₯βˆ—
𝐸 π‘Œ1
βˆ’ π‘Œ0
= 𝐸 π‘Œ1 βˆ’ 𝐸 π‘Œ0
Population Participation Rate
Poor .4 .7
Middle .5 .3
Rich .1 .1
𝐸 π‘Œ1
βˆ’ π‘Œ0
= 𝐸 π‘Œ1 βˆ’ 𝐸 π‘Œ0
Population Participation
Rate
Poor .4 .7
Middle .5 .3
Rich .1 .1
𝐸 π‘Œ1
βˆ’ π‘Œ0
= 𝐸 π‘Œ1 βˆ’ 𝐸 π‘Œ0
Population Participation
Rate
Poor .4 .7
Middle .5 .3
Rich .1 .1
𝐸 π‘Œ1
βˆ’ π‘Œ0
= 𝐸 π‘Œ1 βˆ’ 𝐸 π‘Œ0
Participants
(𝑷 = 𝟏)
Non-
participants
(𝑷 = 𝟎)
Poor .64 .21
Middle .34 .63
Rich .02 .16
𝐸 π‘Œ1
βˆ’ π‘Œ0
= 𝐸 π‘Œ1 βˆ’ 𝐸 π‘Œ0
Participants
(𝑷 = 𝟏)
Non-
participants
(𝑷 = 𝟎)
Poor .64 .21
Middle .34 .63
Rich .02 .16
𝐸 π‘Œ1
βˆ’ π‘Œ0
= 𝐸 π‘Œ1 βˆ’ 𝐸 π‘Œ0
Participants
(𝑷 = 𝟏)
Non-
participants
(𝑷 = 𝟎)
Poor .64 .21
Middle .34 .63
Rich .02 .16
𝐸 π‘Œ1
βˆ’ π‘Œ0
= 𝐸 π‘Œ1 βˆ’ 𝐸 π‘Œ0
Participants
(𝑷 = 𝟏) Weight
Poor .64 .63
Middle .34 1.47
Rich .02 4.4
𝐸 π‘Œ1
βˆ’ π‘Œ0
= 𝐸 π‘Œ1 βˆ’ 𝐸 π‘Œ0
Participants
(𝑷 = 𝟏) Weight
Poor .64 .63
Middle .34 1.47
Rich .02 4.4
𝐸 π‘Œ1
βˆ’ π‘Œ0
= 𝐸 π‘Œ1 βˆ’ 𝐸 π‘Œ0
Participants
(𝑷 = 𝟏) Weight
Poor .64 .63
Middle .34 1.47
Rich .02 4.4
𝐸 π‘Œ1
βˆ’ π‘Œ0
= 𝐸 π‘Œ1 βˆ’ 𝐸 π‘Œ0
Participants
(𝑷 = 𝟏) Weight
Poor .64 .63
Middle .34 1.47
Rich .02 4.5
𝐸 π‘Œ1
βˆ’ π‘Œ0
= 𝐸 π‘Œ1
βˆ’ 𝐸 π‘Œ0
Average of Y
across a sample
of participants
=
𝑖=1
𝑛 𝑃
𝑀𝑖 βˆ— π‘Œπ‘–
𝑛 𝑃
π’˜π’Š=.63 if individual i is poor
π’˜π’Š=1.47 if individual i is middle class
π’˜π’Š=4.4 if individual i is rich
𝐸 π‘Œ1
βˆ’ π‘Œ0
= 𝐸 π‘Œ1
βˆ’ 𝐸 π‘Œ0
Average of Y
across a sample
of participants
=
𝑖=1
𝑛 𝑃
π‘Œπ‘–
𝑛 𝑃
π’˜π’Š=.63 if individual i is poor
π’˜π’Š=1.47 if individual i is middle class
π’˜π’Š=4.4 if individual i is rich
𝐸 π‘Œ1
βˆ’ π‘Œ0
= 𝐸 π‘Œ1
βˆ’ 𝐸 π‘Œ0
Average of Y
across a sample
of participants
=
𝑖=1
𝑛 𝑃
𝑀𝑖 βˆ™ π‘Œπ‘–
𝑖=1
𝑛 𝑃
𝑀𝑖
π’˜π’Š=.63 if individual i is poor
π’˜π’Š=1.47 if individual i is middle class
π’˜π’Š=4.4 if individual i is rich
𝐸 π‘Œ1
βˆ’ π‘Œ0
= 𝐸 π‘Œ1
βˆ’ 𝐸 π‘Œ0
Average of Y
across a sample
of participants
=
𝑖=1
𝑛 𝑃
π’˜π’Š βˆ™ π‘Œπ‘–
𝑖=1
𝑛 𝑃
𝑀𝑖
π’˜π’Š=.63 if individual i is poor
π’˜π’Š=1.47 if individual i is middle class
π’˜π’Š=4.4 if individual i is rich
𝐸 π‘Œ1
βˆ’ π‘Œ0
= 𝐸 π‘Œ1 βˆ’ 𝐸 π‘Œ0
Participants
(𝑷 = 𝟏) Weight
Poor .64 .63
Middle .34 1.47
Rich .02 4.5
𝐸 π‘Œ1
βˆ’ π‘Œ0
= 𝐸 π‘Œ1
βˆ’ 𝐸 π‘Œ0
Average of Y
across a sample
of participants
=
𝑖=1
𝑛 𝑃
𝑀𝑖 βˆ™ π‘Œπ‘–
𝑖=1
𝑛 𝑃
𝑀𝑖
π’˜π’Š=.63 if individual i is poor
π’˜π’Š=1.47 if individual i is middle
class
π’˜π’Š=4.5 if individual i is rich
𝐸 π‘Œ1
βˆ’ π‘Œ0
= 𝐸 π‘Œ1
βˆ’ 𝐸 π‘Œ0
. 4 βˆ—
Average Y
across sample
of participants
who are poor
+ .5 βˆ—
Average Y
across sample
of nonβˆ’participants
who are middle class
+.1 βˆ—
Average Y
across sample
of participants
who are rich
𝐸 π‘Œ1
βˆ’ π‘Œ0
= 𝐸 π‘Œ1
βˆ’ 𝐸 π‘Œ0
Average of Y
across a sample
of participants
=
𝑖=1
𝑛 𝑃
𝑀𝑖 βˆ™ π‘Œπ‘–
𝑖=1
𝑛 𝑃
𝑀𝑖
π’˜π’Š=.63 if individual i is poor
π’˜π’Š=1.47 if individual i is middle
class
π’˜π’Š=4.5 if individual i is rich
π‘Œ0 = 𝛽0 + πœ–
π‘Œ1
= 𝛽0 + 𝛽1 + πœ–
π‘Œ1 βˆ’ π‘Œ0
= 𝛽0 + 𝛽1 + πœ– βˆ’ 𝛽0 + πœ–
= 𝛽0 + 𝛽1 + πœ– βˆ’ 𝛽0 βˆ’ πœ–
= 𝛽1
π‘Œ1 βˆ’ π‘Œ0
= 𝛽0 + 𝛽1 + πœ– βˆ’ 𝛽0 + πœ–
= 𝛽0 + 𝛽1 + πœ– βˆ’ 𝛽0 βˆ’ πœ–
= 𝛽1
π‘Œ1 βˆ’ π‘Œ0
= 𝛽0 + 𝛽1 + πœ– βˆ’ 𝛽0 + πœ–
= 𝛽0 + 𝛽1 + πœ– βˆ’ 𝛽0 βˆ’ πœ–
= 𝛽1
π‘Œ1 βˆ’ π‘Œ0
= 𝛽0 + 𝛽1 + πœ– βˆ’ 𝛽0 + πœ–
= 𝛽0 + 𝛽1 + πœ– βˆ’ 𝛽0 βˆ’ πœ–
= 𝛽1
π‘Œ = 𝑃 βˆ™ π‘Œ1
+ 1 βˆ’ 𝑃 βˆ™ π‘Œ0
= 𝑃 βˆ— 𝛽0 + 𝛽1 + πœ– + 1 βˆ’ 𝑃 βˆ— 𝛽0 + πœ–
= 𝑃 βˆ— 𝛽0 + 𝑃 βˆ— 𝛽1 + 𝑃 βˆ— πœ–
+𝛽0 + πœ– βˆ’ 𝑃 βˆ— 𝛽0 βˆ’ 𝑃 βˆ— πœ–
= 𝛽0 + 𝑃 βˆ— 𝛽1 + πœ–
π‘Œ = 𝑃 βˆ™ π‘Œ1
+ 1 βˆ’ 𝑃 βˆ™ π‘Œ0
= 𝑃 βˆ™ 𝛽0 + 𝛽1 + πœ– + 1 βˆ’ 𝑃 βˆ™ 𝛽0 + πœ–
= 𝑃 βˆ— 𝛽0 + 𝑃 βˆ— 𝛽1 + 𝑃 βˆ— πœ–
+𝛽0 + πœ– βˆ’ 𝑃 βˆ— 𝛽0 βˆ’ 𝑃 βˆ— πœ–
= 𝛽0 + 𝑃 βˆ— 𝛽1 + πœ–
π‘Œ = 𝑃 βˆ™ π‘Œ1
+ 1 βˆ’ 𝑃 βˆ™ π‘Œ0
= 𝑃 βˆ™ 𝛽0 + 𝛽1 + πœ– + 1 βˆ’ 𝑃 βˆ™ 𝛽0 + πœ–
= 𝑃 βˆ™ 𝛽0 + 𝑃 βˆ™ 𝛽1 + 𝑃 βˆ™ πœ–
+𝛽0 + πœ– βˆ’ 𝑃 βˆ™ 𝛽0 βˆ’ 𝑃 βˆ™ πœ–
= 𝛽0 + 𝑃 βˆ— 𝛽1 + πœ–
π‘Œ = 𝑃 βˆ™ π‘Œ1
+ 1 βˆ’ 𝑃 βˆ™ π‘Œ0
= 𝑃 βˆ™ 𝛽0 + 𝛽1 + πœ– + 1 βˆ’ 𝑃 βˆ™ 𝛽0 + πœ–
= 𝑃 βˆ™ 𝛽0 + 𝑃 βˆ™ 𝛽1 + 𝑃 βˆ™ πœ–
+𝛽0 + πœ– βˆ’ 𝑃 βˆ™ 𝛽0 βˆ’ 𝑃 βˆ™ πœ–
= 𝛽0 + 𝑃 βˆ— 𝛽1 + πœ–
π‘Œ = 𝑃 βˆ™ π‘Œ1
+ 1 βˆ’ 𝑃 βˆ™ π‘Œ0
= 𝑃 βˆ™ 𝛽0 + 𝛽1 + πœ– + 1 βˆ’ 𝑃 βˆ™ 𝛽0 + πœ–
= 𝑃 βˆ™ 𝛽0 + 𝑃 βˆ™ 𝛽1 + 𝑃 βˆ™ πœ–
+𝛽0 + πœ– βˆ’ 𝑃 βˆ™ 𝛽0 βˆ’ 𝑃 βˆ™ πœ–
= 𝛽0 + 𝑃 βˆ— 𝛽1 + πœ–
π‘Œ = 𝑃 βˆ™ π‘Œ1
+ 1 βˆ’ 𝑃 βˆ™ π‘Œ0
= 𝑃 βˆ™ 𝛽0 + 𝛽1 + πœ– + 1 βˆ’ 𝑃 βˆ™ 𝛽0 + πœ–
= 𝑃 βˆ™ 𝛽0 + 𝑃 βˆ™ 𝛽1 + 𝑃 βˆ™ πœ–
+𝛽0 + πœ– βˆ’ 𝑃 βˆ™ 𝛽0 βˆ’ 𝑃 βˆ™ πœ–
= 𝛽0 + 𝑃 βˆ™ 𝛽1 + πœ–
π‘Œ = 𝛽0 + 𝑃 βˆ™ 𝛽1 + πœ–
π‘Œ = 𝛽0 + 𝛽1 βˆ™ 𝑃 + πœ–
π‘Œ0
= 𝛽0 + πœ–
π‘Œ1
= 𝛽0 + 𝛽1 + πœ–
π‘Œ = 𝑃 βˆ™ π‘Œ1
+ 1 βˆ’ 𝑃 βˆ™ π‘Œ0
π‘Œ0
= 𝛽0 + πœ–
π‘Œ1
= 𝛽0 + 𝛽1 + πœ–
π‘Œ = 𝑃 βˆ™ π‘Œ1
+ 1 βˆ’ 𝑃 βˆ™ π‘Œ0
π‘Œ0
= 𝛽0 + πœ–
π‘Œ1
= 𝛽0 + 𝛽1 + πœ–
π‘Œ = 𝑃 βˆ™ π‘Œ1
+ 1 βˆ’ 𝑃 βˆ™ π‘Œ0
π‘Œ = 𝛽0 + 𝛽1 βˆ™ 𝑃 + πœ–
π‘Œ0
= 𝛽0 + πœ–
π‘Œ1
= 𝛽0 + 𝛽1 + πœ–
π‘Œ = 𝑃 βˆ™ π‘Œ1
+ 1 βˆ’ 𝑃 βˆ™ π‘Œ0
π‘Œ = 𝛽0 + 𝛽1 βˆ™ 𝑃 + πœ–
π‘Œπ‘– = 𝛽0 + 𝛽1 βˆ™ 𝑃𝑖 + πœ–π‘–
𝑖 = 1, … , 𝑛
πœ–π‘– = π‘Œπ‘– βˆ’ 𝛽0 βˆ’ 𝛽1 βˆ™ 𝑃𝑖
π‘šπ‘–π‘› 𝛽0, 𝛽1
𝑖=1
𝑛
πœ€π‘–
2
= π‘šπ‘–π‘› 𝛽0, 𝛽1
𝑖=1
𝑛
π‘Œπ‘– βˆ’ 𝛽0 βˆ’ 𝛽1 βˆ™ 𝑃𝑖
2
πœ–π‘–
π‘Œ = 𝛽0 + 𝛽1 βˆ™ 𝑃 + πœ–
π‘Œ = 𝛽0 + 𝛽1 βˆ™ 𝑃 + πœ–
πœ–π‘– = π‘Œπ‘– βˆ’ 𝛽0 βˆ’ 𝛽1 βˆ™ 𝑃𝑖
π‘šπ‘–π‘› 𝛽0, 𝛽1
𝑖=1
𝑛
πœ€π‘–
2
= π‘šπ‘–π‘› 𝛽0, 𝛽1
𝑖=1
𝑛
π‘Œπ‘– βˆ’ 𝛽0 βˆ’ 𝛽1 βˆ™ 𝑃𝑖
2
πœ–π‘–
𝛽1 =
𝑖=1
𝑛
𝑃𝑖 βˆ’ 𝑃 βˆ™ π‘Œπ‘–
𝑖=1
𝑛
𝑃𝑖 βˆ’ 𝑃 2
where
𝑃 =
𝑖=1
𝑛
𝑃𝑖
𝑛
𝐸 𝛽1 = 𝛽1
An expected value for a random variable is the
average value from a large number of repetitions
of the experiment that random variable represents
An expected value is the true average of a random
variable across a population
Expected value
An expected value for a random variable is the
average value from a large number of repetitions
of the experiment that random variable represents
An expected value is the true average of a random
variable across a population
Expected value
An expected value is the true average of a random
variable across a population
𝐸 𝑋 = some true value
Expected value
π‘Œ = 𝛽0 + 𝛽1 βˆ™ 𝑃 + πœ–
𝐸 𝑐 = 𝑐
𝐸 𝑐 βˆ™ π‘Š = 𝑐 βˆ™ 𝐸 π‘Š
𝐸 π‘Š + 𝑍 = 𝐸 π‘Š + 𝐸 𝑍
𝐸 π‘Š βˆ’ 𝑍 = 𝐸 π‘Š βˆ’ 𝐸 𝑍
𝐸 π‘Ž βˆ™ π‘Š Β± 𝑏 βˆ™ 𝑍 = π‘Ž βˆ™ 𝐸 π‘Š Β± 𝑏 βˆ™ 𝐸 𝑍
Expectations: Properties
𝑬 𝒄 = 𝒄
𝐸 𝑐 βˆ™ π‘Š = 𝑐 βˆ™ 𝐸 π‘Š
𝐸 π‘Š + 𝑍 = 𝐸 π‘Š + 𝐸 𝑍
𝐸 π‘Š βˆ’ 𝑍 = 𝐸 π‘Š βˆ’ 𝐸 𝑍
𝐸 π‘Ž βˆ™ π‘Š Β± 𝑏 βˆ™ 𝑍 = π‘Ž βˆ™ 𝐸 π‘Š Β± 𝑏 βˆ™ 𝐸 𝑍
Expectations: Properties
𝐸 𝑐 = 𝑐
𝑬 𝒄 βˆ™ 𝑾 = 𝒄 βˆ™ 𝑬 𝑾
𝐸 π‘Š + 𝑍 = 𝐸 π‘Š + 𝐸 𝑍
𝐸 π‘Š βˆ’ 𝑍 = 𝐸 π‘Š βˆ’ 𝐸 𝑍
𝐸 π‘Ž βˆ™ π‘Š Β± 𝑏 βˆ™ 𝑍 = π‘Ž βˆ™ 𝐸 π‘Š Β± 𝑏 βˆ™ 𝐸 𝑍
Expectations: Properties
𝐸 𝑐 = 𝑐
𝐸 𝑐 βˆ™ π‘Š = 𝑐 βˆ™ 𝐸 π‘Š
𝑬 𝑾 + 𝒁 = 𝑬 𝑾 + 𝑬 𝒁
𝐸 π‘Š βˆ’ 𝑍 = 𝐸 π‘Š βˆ’ 𝐸 𝑍
𝐸 π‘Ž βˆ™ π‘Š Β± 𝑏 βˆ™ 𝑍 = π‘Ž βˆ™ 𝐸 π‘Š Β± 𝑏 βˆ™ 𝐸 𝑍
Expectations: Properties
𝐸 𝑐 = 𝑐
𝐸 𝑐 βˆ™ π‘Š = 𝑐 βˆ™ 𝐸 π‘Š
𝐸 π‘Š + 𝑍 = 𝐸 π‘Š + 𝐸 𝑍
𝑬 𝑾 βˆ’ 𝒁 = 𝑬 𝑾 βˆ’ 𝑬 𝒁
𝐸 π‘Ž βˆ™ π‘Š Β± 𝑏 βˆ™ 𝑍 = π‘Ž βˆ™ 𝐸 π‘Š Β± 𝑏 βˆ™ 𝐸 𝑍
Expectations: Properties
𝐸 𝑐 = 𝑐
𝐸 𝑐 βˆ™ π‘Š = 𝑐 βˆ™ 𝐸 π‘Š
𝐸 π‘Š + 𝑍 = 𝐸 π‘Š + 𝐸 𝑍
𝐸 π‘Š βˆ’ 𝑍 = 𝐸 π‘Š βˆ’ 𝐸 𝑍
𝑬 𝒂 βˆ™ 𝑾 Β± 𝒃 βˆ™ 𝒁 = 𝒂 βˆ™ 𝑬 𝑾 Β± 𝒃 βˆ™ 𝑬 𝒁
Expectations: Properties
𝐸 π‘Š βˆ™ 𝑍 β‰  𝐸 π‘Š βˆ™ 𝐸 𝑍
𝐸
π‘Š
𝑍
β‰ 
𝐸 π‘Š
𝐸 𝑍
𝐸 𝑓 π‘Š β‰  𝑓 𝐸 π‘Š
Expectations: Properties
𝑬 𝑾 βˆ™ 𝒁 β‰  𝑬 𝑾 βˆ™ 𝑬 𝒁
𝐸
π‘Š
𝑍
β‰ 
𝐸 π‘Š
𝐸 𝑍
𝐸 𝑓 π‘Š β‰  𝑓 𝐸 π‘Š
Expectations: Properties
𝐸 π‘Š βˆ™ 𝑍 β‰  𝐸 π‘Š βˆ™ 𝐸 𝑍
𝑬
𝑾
𝒁
β‰ 
𝑬 𝑾
𝑬 𝒁
𝐸 𝑓 π‘Š β‰  𝑓 𝐸 π‘Š
Expectations: Properties
𝐸 π‘Š βˆ™ 𝑍 β‰  𝐸 π‘Š βˆ™ 𝐸 𝑍
𝐸
π‘Š
𝑍
β‰ 
𝐸 π‘Š
𝐸 𝑍
𝑬 𝒇 𝑾 β‰  𝒇 𝑬 𝑾
Expectations: Properties
π‘Œπ‘– = 𝛽0 + 𝛽1 βˆ™ 𝑃𝑖 + πœ–π‘–
𝛽1 =
𝑖=1
𝑛
𝑃𝑖 βˆ’ 𝑃 βˆ™ π‘Œπ‘–
𝑖=1
𝑛
𝑃𝑖 βˆ’ 𝑃 2
π‘Œπ‘– = 𝛽0 + 𝛽1 βˆ™ 𝑃𝑖 + πœ–π‘–
𝛽1 =
𝑖=1
𝑛
𝑃𝑖 βˆ’ 𝑃 βˆ™ π‘Œπ‘–
𝑖=1
𝑛
𝑃𝑖 βˆ’ 𝑃 2
π‘Œπ‘– = 𝛽0 + 𝛽1 βˆ™ 𝑃𝑖 + πœ–π‘–
𝛽1 =
𝑖=1
𝑛
𝑃𝑖 βˆ’ 𝑃 βˆ™ π‘Œπ‘–
𝑖=1
𝑛
𝑃𝑖 βˆ’ 𝑃 2
𝐸 𝛽1 = 𝛽1
𝐸 𝛽1 = 𝛽1
𝛽1 =
𝑖=1
𝑛
𝑃𝑖 βˆ’ 𝑃 βˆ™ π‘Œπ‘–
𝑖=1
𝑛
𝑃𝑖 βˆ’ 𝑃 2
𝐸( 𝛽1) = 𝐸
𝑖=1
𝑛
𝑃𝑖 βˆ’ 𝑃 βˆ™ π‘Œπ‘–
𝑖=1
𝑛
𝑃𝑖 βˆ’ 𝑃 2
π‘Œπ‘– = 𝛽0 + 𝛽1 βˆ™ 𝑃𝑖 + πœ–π‘–
𝐸( 𝛽1) = 𝐸
𝑖=1
𝑛
𝑃𝑖 βˆ’ 𝑃 βˆ™ π‘Œπ‘–
𝑖=1
𝑛
𝑃𝑖 βˆ’ 𝑃 2
𝐸 𝛽1
= 𝐸
𝑖=1
𝑛
𝑃𝑖 βˆ’ 𝑃 βˆ™ 𝛽0 + 𝛽1 βˆ™ 𝑃𝑖 + πœ–π‘–
𝑖=1
𝑛
𝑃𝑖 βˆ’ 𝑃 2
𝐸 𝛽1
= 𝐸
𝑖=1
𝑛
𝛽0 βˆ™ 𝑃𝑖 βˆ’ 𝑃
𝑖=1
𝑛
𝑃𝑖 βˆ’ 𝑃 2
+
𝑖=1
𝑛
𝛽1 βˆ™ 𝑃𝑖 βˆ™ 𝑃𝑖 βˆ’ 𝑃
𝑖=1
𝑛
𝑃𝑖 βˆ’ 𝑃 2
𝐸 π‘Š + 𝑍 = 𝐸 π‘Š + 𝐸 𝑍
𝐸 π‘Š + 𝑍 + 𝑄 = 𝐸 π‘Š + 𝐸 𝑍 + 𝐸 𝑄
𝐸 π‘Š + 𝑍 = 𝐸 π‘Š + 𝐸 𝑍
𝐸 π‘Š + 𝑍 + 𝑄 = 𝐸 π‘Š + 𝐸 𝑍 + 𝐸 𝑄
𝐸 𝛽1
= 𝐸
𝑖=1
𝑛
𝛽0 βˆ™ 𝑃𝑖 βˆ’ 𝑃
𝑖=1
𝑛
𝑃𝑖 βˆ’ 𝑃 2
+
𝑖=1
𝑛
𝛽1 βˆ™ 𝑃𝑖 βˆ™ 𝑃𝑖 βˆ’ 𝑃
𝑖=1
𝑛
𝑃𝑖 βˆ’ 𝑃 2
𝐸 𝛽1
= 𝐸
𝑖=1
𝑛
𝛽0 βˆ™ 𝑃𝑖 βˆ’ 𝑃
𝑖=1
𝑛
𝑃𝑖 βˆ’ 𝑃 2
+
𝑖=1
𝑛
𝛽1 βˆ™ 𝑃𝑖 βˆ™ 𝑃𝑖 βˆ’ 𝑃
𝑖=1
𝑛
𝑃𝑖 βˆ’ 𝑃 2
𝐸 𝛽1
= 𝐸
𝑖=1
𝑛
𝛽0 βˆ™ 𝑃𝑖 βˆ’ 𝑃
𝑖=1
𝑛
𝑃𝑖 βˆ’ 𝑃 2
+
𝑖=1
𝑛
𝛽1 βˆ™ 𝑃𝑖 βˆ™ 𝑃𝑖 βˆ’ 𝑃
𝑖=1
𝑛
𝑃𝑖 βˆ’ 𝑃 2
𝐸 𝛽1
= 𝐸
𝑖=1
𝑛
𝛽0 βˆ™ 𝑃𝑖 βˆ’ 𝑃
𝑖=1
𝑛
𝑃𝑖 βˆ’ 𝑃 2
+
𝑖=1
𝑛
𝛽1 βˆ™ 𝑃𝑖 βˆ™ 𝑃𝑖 βˆ’ 𝑃
𝑖=1
𝑛
𝑃𝑖 βˆ’ 𝑃 2
𝐸 𝛽1
= 𝐸
𝑖=1
𝑛
𝛽0 βˆ™ 𝑃𝑖 βˆ’ 𝑃
𝑖=1
𝑛
𝑃𝑖 βˆ’ 𝑃 2
+
𝑖=1
𝑛
𝛽1 βˆ™ 𝑃𝑖 βˆ™ 𝑃𝑖 βˆ’ 𝑃
𝑖=1
𝑛
𝑃𝑖 βˆ’ 𝑃 2
𝐸 𝛽1
= 𝐸
𝑖=1
𝑛
𝛽0 βˆ™ 𝑃𝑖 βˆ’ 𝑃
𝑖=1
𝑛
𝑃𝑖 βˆ’ 𝑃 2
+ 𝐸
𝑖=1
𝑛
𝛽1 βˆ™ 𝑃𝑖 βˆ™ 𝑃𝑖 βˆ’ 𝑃
𝑖=1
𝑛
𝑃𝑖 βˆ’ 𝑃 2
+ 𝐸
𝑖=1
𝑛
πœ–π‘– βˆ™ 𝑃𝑖 βˆ’ 𝑃
𝑖=1
𝑛
𝑃𝑖 βˆ’ 𝑃 2
𝐸 𝛽1
= 𝐸
𝑖=1
𝑛
𝛽0 βˆ™ 𝑃𝑖 βˆ’ 𝑃
𝑖=1
𝑛
𝑃𝑖 βˆ’ 𝑃 2
+ 𝐸
𝑖=1
𝑛
𝛽1 βˆ™ 𝑃𝑖 βˆ™ 𝑃𝑖 βˆ’ 𝑃
𝑖=1
𝑛
𝑃𝑖 βˆ’ 𝑃 2
+ 𝐸
𝑖=1
𝑛
πœ–π‘– βˆ™ 𝑃𝑖 βˆ’ 𝑃
𝑖=1
𝑛
𝑃𝑖 βˆ’ 𝑃 2
𝐸 𝑐 βˆ™ π‘Š = 𝑐 βˆ™ 𝐸 π‘Š
𝐸 𝛽1
= 𝐸
𝑖=1
𝑛
𝛽0 βˆ™ 𝑃𝑖 βˆ’ 𝑃
𝑖=1
𝑛
𝑃𝑖 βˆ’ 𝑃 2
+ 𝐸
𝑖=1
𝑛
𝛽1 βˆ™ 𝑃𝑖 βˆ™ 𝑃𝑖 βˆ’ 𝑃
𝑖=1
𝑛
𝑃𝑖 βˆ’ 𝑃 2
+ 𝐸
𝑖=1
𝑛
πœ–π‘– βˆ™ 𝑃𝑖 βˆ’ 𝑃
𝑖=1
𝑛
𝑃𝑖 βˆ’ 𝑃 2
𝐸 𝛽1
= 𝛽0 βˆ™ 𝐸
𝑖=1
𝑛
𝑃𝑖 βˆ’ 𝑃
𝑖=1
𝑛
𝑃𝑖 βˆ’ 𝑃 2
+ 𝐸
𝑖=1
𝑛
𝛽1 βˆ™ 𝑃𝑖 βˆ™ 𝑃𝑖 βˆ’ 𝑃
𝑖=1
𝑛
𝑃𝑖 βˆ’ 𝑃 2
+ 𝐸
𝑖=1
𝑛
πœ–π‘– βˆ™ 𝑃𝑖 βˆ’ 𝑃
𝑖=1
𝑛
𝑃𝑖 βˆ’ 𝑃 2
𝐸 𝛽1
= 𝛽0 βˆ™ 𝐸
𝑖=1
𝑛
𝑃𝑖 βˆ’ 𝑃
𝑖=1
𝑛
𝑃𝑖 βˆ’ 𝑃 2
+ 𝐸
𝑖=1
𝑛
𝛽1 βˆ™ 𝑃𝑖 βˆ™ 𝑃𝑖 βˆ’ 𝑃
𝑖=1
𝑛
𝑃𝑖 βˆ’ 𝑃 2
+ 𝐸
𝑖=1
𝑛
πœ–π‘– βˆ™ 𝑃𝑖 βˆ’ 𝑃
𝑖=1
𝑛
𝑃𝑖 βˆ’ 𝑃 2
𝐸 𝛽1
= 𝛽0 βˆ— 𝐸
0
𝑖=1
𝑛
𝑃𝑖 βˆ’ 𝑃 2
+ 𝐸
𝑖=1
𝑛
𝛽1 βˆ™ 𝑃𝑖 βˆ™ 𝑃𝑖 βˆ’ 𝑃
𝑖=1
𝑛
𝑃𝑖 βˆ’ 𝑃 2
+ 𝐸
𝑖=1
𝑛
πœ–π‘– βˆ™ 𝑃𝑖 βˆ’ 𝑃
𝑖=1
𝑛
𝑃𝑖 βˆ’ 𝑃 2
𝐸 𝛽1
= 𝛽0 βˆ— 0
+𝐸
𝑖=1
𝑛
𝛽1 βˆ™ 𝑃𝑖 βˆ™ 𝑃𝑖 βˆ’ 𝑃
𝑖=1
𝑛
𝑃𝑖 βˆ’ 𝑃 2
+ 𝐸
𝑖=1
𝑛
πœ–π‘– βˆ™ 𝑃𝑖 βˆ’ 𝑃
𝑖=1
𝑛
𝑃𝑖 βˆ’ 𝑃 2
𝐸 𝛽1
= 𝐸
𝑖=1
𝑛
𝛽1 βˆ™ 𝑃𝑖 βˆ™ 𝑃𝑖 βˆ’ 𝑃
𝑖=1
𝑛
𝑃𝑖 βˆ’ 𝑃 2
+ 𝐸
𝑖=1
𝑛
πœ–π‘– βˆ™ 𝑃𝑖 βˆ’ 𝑃
𝑖=1
𝑛
𝑃𝑖 βˆ’ 𝑃 2
𝐸 𝛽1
= 𝐸
𝑖=1
𝑛
𝛽1 βˆ™ 𝑃𝑖 βˆ™ 𝑃𝑖 βˆ’ 𝑃
𝑖=1
𝑛
𝑃𝑖 βˆ’ 𝑃 2
+ 𝐸
𝑖=1
𝑛
πœ–π‘– βˆ™ 𝑃𝑖 βˆ’ 𝑃
𝑖=1
𝑛
𝑃𝑖 βˆ’ 𝑃 2
𝐸 𝛽1
= 𝛽1 βˆ™ 𝐸
𝑖=1
𝑛
𝑃𝑖 βˆ™ 𝑃𝑖 βˆ’ 𝑃
𝑖=1
𝑛
𝑃𝑖 βˆ’ 𝑃 2
+ 𝐸
𝑖=1
𝑛
πœ–π‘– βˆ™ 𝑃𝑖 βˆ’ 𝑃
𝑖=1
𝑛
𝑃𝑖 βˆ’ 𝑃 2
𝐸 𝛽1
= 𝛽1 βˆ™ 𝐸
𝑖=1
𝑛
𝑃𝑖 βˆ™ 𝑃𝑖 βˆ’ 𝑃
𝑖=1
𝑛
𝑃𝑖 βˆ’ 𝑃 2
+ 𝐸
𝑖=1
𝑛
πœ–π‘– βˆ™ 𝑃𝑖 βˆ’ 𝑃
𝑖=1
𝑛
𝑃𝑖 βˆ’ 𝑃 2
𝑖=1
𝑛
𝑃𝑖 βˆ™ 𝑃𝑖 βˆ’ 𝑃 =
𝑖=1
𝑛
𝑃𝑖 βˆ’ 𝑃 2
𝐸 𝛽1
= 𝛽1 βˆ— 𝐸
𝑖=1
𝑛
𝑃𝑖 βˆ™ 𝑃𝑖 βˆ’ 𝑃
𝑖=1
𝑛
𝑃𝑖 βˆ’ 𝑃 2
+ 𝐸
𝑖=1
𝑛
πœ–π‘– βˆ™ 𝑃𝑖 βˆ’ 𝑃
𝑖=1
𝑛
𝑃𝑖 βˆ’ 𝑃 2
𝐸 𝛽1
= 𝛽1 βˆ™ 𝐸
𝑖=1
𝑛
𝑃𝑖 βˆ’ 𝑃 2
𝑖=1
𝑛
𝑃𝑖 βˆ’ 𝑃 2
+ 𝐸
𝑖=1
𝑛
πœ–π‘– βˆ™ 𝑃𝑖 βˆ’ 𝑃
𝑖=1
𝑛
𝑃𝑖 βˆ’ 𝑃 2
𝐸 𝛽1
= 𝛽1 βˆ™ 𝐸 1
+𝐸
𝑖=1
𝑛
πœ–π‘– βˆ™ 𝑃𝑖 βˆ’ 𝑃
𝑖=1
𝑛
𝑃𝑖 βˆ’ 𝑃 2
𝐸 𝛽1
= 𝛽1 βˆ™ 1
+𝐸
𝑖=1
𝑛
πœ–π‘– βˆ™ 𝑃𝑖 βˆ’ 𝑃
𝑖=1
𝑛
𝑃𝑖 βˆ’ 𝑃 2
𝐸 𝛽1
= 𝛽1
+𝐸
𝑖=1
𝑛
πœ–π‘– βˆ™ 𝑃𝑖 βˆ’ 𝑃
𝑖=1
𝑛
𝑃𝑖 βˆ’ 𝑃 2
𝐸 𝛽1
= 𝛽1
+𝐸
𝑖=1
𝑛
πœ–π‘– βˆ™ 𝑃𝑖 βˆ’ 𝑃
𝑖=1
𝑛
𝑃𝑖 βˆ’ 𝑃 2
𝐸 𝛽1
= 𝛽1
+0
𝐸 𝛽1
= 𝛽1
𝐸 𝛽1
= 𝛽1
+𝐸
𝑖=1
𝑛
πœ–π‘– βˆ™ 𝑃𝑖 βˆ’ 𝑃
𝑖=1
𝑛
𝑃𝑖 βˆ’ 𝑃 2
𝐸 π‘Š βˆ™ 𝑍 β‰  𝐸 π‘Š βˆ™ 𝐸 𝑍
Expectations: Properties
𝐸 𝛽1
= 𝛽1
+𝐸
𝑖=1
𝑛
πœ–π‘– βˆ™ 𝑃𝑖 βˆ’ 𝑃
𝑖=1
𝑛
𝑃𝑖 βˆ’ 𝑃 2
𝐸 𝛽1
= 𝛽1
+
𝑖=1
𝑛
𝐸
πœ–π‘– βˆ™ 𝑃𝑖 βˆ’ 𝑃
𝑖=1
𝑛
𝑃𝑖 βˆ’ 𝑃 2
𝐸 𝛽1
= 𝛽1
+
𝑖=1
𝑛
𝐸
𝑖=1
𝑛
𝑃𝑖 βˆ’ 𝑃
𝑖=1
𝑛
𝑃𝑖 βˆ’ 𝑃 2
βˆ™ 𝐸 πœ–π‘–
𝐸 𝛽1
= 𝛽1
+
𝑖=1
𝑛
𝐸
𝑖=1
𝑛
𝑃𝑖 βˆ’ 𝑃
𝑖=1
𝑛
𝑃𝑖 βˆ’ 𝑃 2
βˆ™ 0
𝐸 𝛽1
= 𝛽1
+0
𝐸 𝛽1
= 𝛽1
𝐸 𝛽1
= 𝛽1
π‘Œπ‘– = 𝛽0 + 𝛽1 βˆ™ 𝑃𝑖 + πœ–π‘–
𝐸 𝛽1
= 𝛽1
π‘Œπ‘– = 𝛽0 + 𝛽1 βˆ™ 𝑃𝑖 + πœ–π‘–
𝐸 𝛽1
= 𝛽1
π‘Œπ‘– = 𝛽0 + 𝛽1 βˆ™ 𝑃𝑖 + πœ–π‘–
Peter M. Lance, PhD
MEASURE Evaluation
University of North Carolina at
Chapel Hill
September 13 and 15, 2016
Selection on Observables: Part
Deux
π‘Œ0
= 𝛽0 + πœ–
π‘Œ1
= 𝛽0 + 𝛽1 + πœ–
π‘Œ = 𝑃 βˆ™ π‘Œ1
+ 1 βˆ’ 𝑃 βˆ™ π‘Œ0
π‘Œ = 𝛽0 + 𝛽1 βˆ™ 𝑃 + πœ–
π‘Œ0
= 𝛽0 + πœ–
π‘Œ1
= 𝛽0 + 𝛽1 + πœ–
π‘Œ = 𝑃 βˆ™ π‘Œ1
+ 1 βˆ’ 𝑃 βˆ™ π‘Œ0
π‘Œ = 𝛽0 + 𝛽1 βˆ™ 𝑃 + πœ–
π‘Œ0
= 𝛽0 + πœ–
π‘Œ1
= 𝛽0 + 𝛽1 + πœ–
π‘Œ = 𝑃 βˆ™ π‘Œ1
+ 1 βˆ’ 𝑃 βˆ™ π‘Œ0
π‘Œ = 𝛽0 + 𝛽1 βˆ™ 𝑃 + πœ–
𝐸 𝛽1
= 𝛽1
+𝐸
𝑖=1
𝑛
πœ–π‘– βˆ™ 𝑃𝑖 βˆ’ 𝑃
𝑖=1
𝑛
𝑃𝑖 βˆ’ 𝑃 2
𝐸 𝛽1
= 𝛽1
+𝐸
𝑖=1
𝑛
πœ–π‘– βˆ™ 𝑃𝑖 βˆ’ 𝑃
𝑖=1
𝑛
𝑃𝑖 βˆ’ 𝑃 2
𝐸 𝛽1
= 𝛽1
π‘Œπ‘– = 𝛽0 + 𝛽1 βˆ™ 𝑃𝑖 + πœ–π‘–
𝐸 𝛽1
= 𝛽1
π‘Œπ‘– = 𝛽0 + 𝛽1 βˆ™ 𝑃𝑖 + πœ–π‘–
𝐸 𝛽1
= 𝛽1
π‘Œπ‘– = 𝛽0 + 𝛽1 βˆ™ 𝑃𝑖 + πœ–π‘–
π‘Œ0 = 𝛽0 + πœ€
π‘Œ1
= 𝛽0 + 𝛽1 + πœ€
π‘Œ0 = 𝛽0 + 𝛽2 βˆ™ π‘₯ + πœ€
π‘Œ1
= 𝛽0 + 𝛽1 + 𝛽2 βˆ™ π‘₯ + πœ€
π‘Œ1 βˆ’ π‘Œ0
= 𝛽0 + 𝛽1 + πœ– βˆ’ 𝛽0 + πœ–
= 𝛽0 + 𝛽1 + πœ– βˆ’ 𝛽0 βˆ’ πœ–
= 𝛽1
π‘Œ1 βˆ’ π‘Œ0
= 𝛽0 + 𝛽1 + 𝛽2 βˆ— π‘₯ + πœ€
βˆ’ 𝛽0 + 𝛽2 βˆ— π‘₯ + πœ€
= 𝛽0 + 𝛽1 + πœ– βˆ’ 𝛽0 βˆ’ πœ–
= 𝛽1
π‘Œ1 βˆ’ π‘Œ0
= 𝛽1
π‘Œ = 𝑃 βˆ™ π‘Œ1
+ 1 βˆ’ 𝑃 βˆ™ π‘Œ0
= 𝑃 βˆ— 𝛽0 + 𝛽1 + πœ– + 1 βˆ’ 𝑃 βˆ— 𝛽0 + πœ–
= 𝑃 βˆ— 𝛽0 + 𝑃 βˆ— 𝛽1 + 𝑃 βˆ— πœ–
+𝛽0 + πœ– βˆ’ 𝑃 βˆ— 𝛽0 βˆ’ 𝑃 βˆ— πœ–
= 𝛽0 + 𝑃 βˆ— 𝛽1 + πœ–
π‘Œ = 𝑃 βˆ™ π‘Œ1
+ 1 βˆ’ 𝑃 βˆ™ π‘Œ0
= 𝑃 βˆ™ 𝛽0 + 𝛽1 + 𝛽2 βˆ™ π‘₯ + πœ€
+ 1 βˆ’ 𝑃 βˆ™ 𝛽0 + 𝛽2 βˆ™ π‘₯ + πœ€
= 𝑃 βˆ— 𝛽0 + 𝑃 βˆ— 𝛽1 + 𝑃 βˆ— πœ–
+𝛽0 + πœ– βˆ’ 𝑃 βˆ— 𝛽0 βˆ’ 𝑃 βˆ— πœ–
π‘Œ = 𝛽0 + 𝛽1 βˆ™ 𝑃 + 𝛽2 βˆ™ π‘₯ + πœ€
Cost of participation
𝐢 = 𝜌0 + 𝜌1 βˆ™ π‘₯
Cost of participation
𝐢 = 𝜌0 + 𝜌1 βˆ™ π‘₯
Benefit-Cost>0
π‘Œ1
βˆ’ π‘Œ0
βˆ’ C > 0
𝛽1 βˆ’ 𝐢 > 0
𝛽1 βˆ’ 𝛾0 + 𝛾1 βˆ— π‘₯ > 0
Benefit-Cost>0
π‘Œ1 βˆ’ π‘Œ0 βˆ’ C > 0
𝛽1 βˆ’ 𝐢 > 0
𝛽1 βˆ’ 𝛾0 + 𝛾1 βˆ— π‘₯ > 0
Benefit-Cost>0
π‘Œ1 βˆ’ π‘Œ0 βˆ’ C > 0
𝛽1 βˆ’ 𝐢 > 0
𝛽1 βˆ’ 𝛾0 + 𝛾1 βˆ— π‘₯ > 0
Benefit-Cost>0
π‘Œ1 βˆ’ π‘Œ0 βˆ’ C > 0
𝛽1 βˆ’ 𝐢 > 0
𝛽1 βˆ’ 𝜌0 + 𝜌1 βˆ™ π‘₯ > 0
π‘₯ 𝑃
X
Y
P
Benefit-Cost>0
𝛽1 βˆ’ 𝜌0 + 𝜌1 βˆ™ π‘₯ > 0
πœ€
Benefit-Cost>0
𝛽1 βˆ’ 𝜌0 + 𝜌1 βˆ™ π‘₯ > 0
πœ€
Benefit-Cost>0
𝛽1 βˆ’ 𝜌0 + 𝜌1 βˆ™ π‘₯ > 0
πœ€
P and 𝜺 are independent
True model:
π‘Œ = 𝛽0 + 𝛽1 βˆ™ 𝑃 + 𝛽2 βˆ™ π‘₯ + πœ€
We actually attempt to estimate:
π‘Œ = 𝜏0 + 𝜏1 βˆ™ 𝑃 + πœ–
Error term now contains:
𝛽2 βˆ™ π‘₯
True model:
π‘Œ = 𝛽0 + 𝛽1 βˆ™ 𝑃 + 𝛽2 βˆ™ π‘₯ + πœ€
We actually attempt to estimate:
π‘Œ = 𝜏0 + 𝜏1 βˆ™ 𝑃 + πœ–
Error term now contains:
𝛽2 βˆ™ π‘₯
True model:
π‘Œ = 𝛽0 + 𝛽1 βˆ™ 𝑃 + 𝛽2 βˆ™ π‘₯ + πœ€
We actually attempt to estimate:
π‘Œ = 𝜏0 + 𝜏1 βˆ™ 𝑃 + πœ–
Error term now contains:
𝛽2 βˆ™ π‘₯
𝐸( 𝜏1) = 𝐸
𝑖=1
𝑛
𝑃𝑖 βˆ’ 𝑃 βˆ™ π‘Œπ‘–
𝑖=1
𝑛
𝑃𝑖 βˆ’ 𝑃 2
𝐸 𝜏1
= 𝐸
𝑖=1
𝑛
𝑃𝑖 βˆ’ 𝑃 βˆ™ 𝛽0 + 𝛽1 βˆ™ 𝑃𝑖 + 𝛽2 βˆ™ π‘₯𝑖 + πœ€π‘–
𝑖=1
𝑛
𝑃𝑖 βˆ’ 𝑃 2
𝐸 𝜏1
= 𝐸
𝑖=1
𝑛
𝛽0 βˆ™ 𝑃𝑖 βˆ’ 𝑃
𝑖=1
𝑛
𝑃𝑖 βˆ’ 𝑃 2
+
𝑖=1
𝑛
𝛽1 βˆ™ 𝑃𝑖 βˆ™ 𝑃𝑖 βˆ’ 𝑃
𝑖=1
𝑛
𝑃𝑖 βˆ’ 𝑃 2
+
𝑖=1
𝑛
𝛽2 βˆ™ π‘₯𝑖 βˆ™ 𝑃𝑖 βˆ’ 𝑃
𝑖=1
𝑛
𝑃𝑖 βˆ’ 𝑃 2
+
𝑖=1
𝑛
πœ€π‘– βˆ™ 𝑃𝑖 βˆ’ 𝑃
𝑖=1
𝑛
𝑃𝑖 βˆ’ 𝑃 2
𝐸 𝜏1
= 𝐸
𝑖=1
𝑛
𝛽0 βˆ™ 𝑃𝑖 βˆ’ 𝑃
𝑖=1
𝑛
𝑃𝑖 βˆ’ 𝑃 2
+ 𝐸
𝑖=1
𝑛
𝛽1 βˆ™ 𝑃𝑖 βˆ™ 𝑃𝑖 βˆ’ 𝑃
𝑖=1
𝑛
𝑃𝑖 βˆ’ 𝑃 2
+𝐸
𝑖=1
𝑛
𝛽2 βˆ™ π‘₯𝑖 βˆ™ 𝑃𝑖 βˆ’ 𝑃
𝑖=1
𝑛
𝑃𝑖 βˆ’ 𝑃 2
+𝐸
𝑖=1
𝑛
πœ€π‘– βˆ™ 𝑃𝑖 βˆ’ 𝑃
𝑖=1
𝑛
𝑃𝑖 βˆ’ 𝑃 2
𝐸 𝜏1
= 𝐸
𝑖=1
𝑛
𝛽0 βˆ™ 𝑃𝑖 βˆ’ 𝑃
𝑖=1
𝑛
𝑃𝑖 βˆ’ 𝑃 2
+ 𝐸
𝑖=1
𝑛
𝛽1 βˆ™ 𝑃𝑖 βˆ™ 𝑃𝑖 βˆ’ 𝑃
𝑖=1
𝑛
𝑃𝑖 βˆ’ 𝑃 2
+𝐸
𝑖=1
𝑛
𝛽2 βˆ™ π‘₯𝑖 βˆ™ 𝑃𝑖 βˆ’ 𝑃
𝑖=1
𝑛
𝑃𝑖 βˆ’ 𝑃 2
+𝐸
𝑖=1
𝑛
πœ€π‘– βˆ™ 𝑃𝑖 βˆ’ 𝑃
𝑖=1
𝑛
𝑃𝑖 βˆ’ 𝑃 2
𝐸 𝜏1
= 0
+𝐸
𝑖=1
𝑛
𝛽1 βˆ™ 𝑃𝑖 βˆ™ 𝑃𝑖 βˆ’ 𝑃
𝑖=1
𝑛
𝑃𝑖 βˆ’ 𝑃 2
+𝐸
𝑖=1
𝑛
𝛽2 βˆ™ π‘₯𝑖 βˆ™ 𝑃𝑖 βˆ’ 𝑃
𝑖=1
𝑛
𝑃𝑖 βˆ’ 𝑃 2
+𝐸
𝑖=1
𝑛
πœ€π‘– βˆ™ 𝑃𝑖 βˆ’ 𝑃
𝑖=1
𝑛
𝑃𝑖 βˆ’ 𝑃 2
𝐸 𝜏1
= 𝐸
𝑖=1
𝑛
𝛽1 βˆ™ 𝑃𝑖 βˆ™ 𝑃𝑖 βˆ’ 𝑃
𝑖=1
𝑛
𝑃𝑖 βˆ’ 𝑃 2
+𝐸
𝑖=1
𝑛
𝛽2 βˆ™ π‘₯𝑖 βˆ™ 𝑃𝑖 βˆ’ 𝑃
𝑖=1
𝑛
𝑃𝑖 βˆ’ 𝑃 2
+𝐸
𝑖=1
𝑛
πœ€π‘– βˆ™ 𝑃𝑖 βˆ’ 𝑃
𝑖=1
𝑛
𝑃𝑖 βˆ’ 𝑃 2
𝐸 𝜏1
= 𝐸
𝑖=1
𝑛
𝛽1 βˆ™ 𝑃𝑖 βˆ™ 𝑃𝑖 βˆ’ 𝑃
𝑖=1
𝑛
𝑃𝑖 βˆ’ 𝑃 2
+𝐸
𝑖=1
𝑛
𝛽2 βˆ™ π‘₯𝑖 βˆ™ 𝑃𝑖 βˆ’ 𝑃
𝑖=1
𝑛
𝑃𝑖 βˆ’ 𝑃 2
+𝐸
𝑖=1
𝑛
πœ€π‘– βˆ™ 𝑃𝑖 βˆ’ 𝑃
𝑖=1
𝑛
𝑃𝑖 βˆ’ 𝑃 2
𝐸 𝜏1
= 𝛽1
+𝐸
𝑖=1
𝑛
𝛽2 βˆ™ π‘₯𝑖 βˆ™ 𝑃𝑖 βˆ’ 𝑃
𝑖=1
𝑛
𝑃𝑖 βˆ’ 𝑃 2
+𝐸
𝑖=1
𝑛
πœ€π‘– βˆ™ 𝑃𝑖 βˆ’ 𝑃
𝑖=1
𝑛
𝑃𝑖 βˆ’ 𝑃 2
𝐸 𝜏1
= 𝛽1
+𝐸
𝑖=1
𝑛
𝛽2 βˆ™ π‘₯𝑖 βˆ™ 𝑃𝑖 βˆ’ 𝑃
𝑖=1
𝑛
𝑃𝑖 βˆ’ 𝑃 2
+𝐸
𝑖=1
𝑛
πœ€π‘– βˆ™ 𝑃𝑖 βˆ’ 𝑃
𝑖=1
𝑛
𝑃𝑖 βˆ’ 𝑃 2
𝐸 𝜏1
= 𝛽1
+𝐸
𝑖=1
𝑛
𝛽2 βˆ™ π‘₯𝑖 βˆ™ 𝑃𝑖 βˆ’ 𝑃
𝑖=1
𝑛
𝑃𝑖 βˆ’ 𝑃 2
+0
𝐸 𝜏1
= 𝛽1
+𝐸
𝑖=1
𝑛
𝛽2 βˆ™ π‘₯𝑖 βˆ™ 𝑃𝑖 βˆ’ 𝑃
𝑖=1
𝑛
𝑃𝑖 βˆ’ 𝑃 2
𝐸 𝜏1
= 𝛽1
+𝐸
𝑖=1
𝑛
𝛽2 βˆ™ π‘₯𝑖 βˆ™ 𝑃𝑖 βˆ’ 𝑃
𝑖=1
𝑛
𝑃𝑖 βˆ’ 𝑃 2
𝐸 𝑐 βˆ™ π‘Š = 𝑐 βˆ™ 𝐸 π‘Š
𝐸 𝜏1
= 𝛽1
+𝛽2 βˆ™ 𝐸
𝑖=1
𝑛
π‘₯𝑖 βˆ™ 𝑃𝑖 βˆ’ 𝑃
𝑖=1
𝑛
𝑃𝑖 βˆ’ 𝑃 2
𝐸 𝜏1
= 𝛽1
+𝛽2 βˆ™ 𝐸
𝑖=1
𝑛
π‘₯𝑖 βˆ™ 𝑃𝑖 βˆ’ 𝑃
𝑖=1
𝑛
𝑃𝑖 βˆ’ 𝑃 2
𝐸 𝜏1
= 𝛽1
+𝛽2 βˆ™ 𝐸
𝑖=1
𝑛
π‘₯𝑖 βˆ™ 𝑃𝑖 βˆ’ 𝑃
𝑖=1
𝑛
𝑃𝑖 βˆ’ 𝑃 2
π‘₯𝑖= 𝛾0 + 𝛾1 βˆ™ 𝑃𝑖 + πœ—π‘–
𝐸 𝜏1
= 𝛽1
+𝛽2 βˆ™ 𝐸
𝑖=1
𝑛
π‘₯𝑖 βˆ™ 𝑃𝑖 βˆ’ 𝑃
𝑖=1
𝑛
𝑃𝑖 βˆ’ 𝑃 2
π‘₯𝑖= 𝛾0 + 𝛾1 βˆ™ 𝑃𝑖 + πœ—π‘–
𝐸 𝜏1
= 𝛽1
+𝛽2 βˆ™ 𝐸
𝑖=1
𝑛
π‘₯𝑖 βˆ™ 𝑃𝑖 βˆ’ 𝑃
𝑖=1
𝑛
𝑃𝑖 βˆ’ 𝑃 2
π‘₯𝑖= 𝛾0 + 𝛾1 βˆ™ 𝑃𝑖 + πœ—π‘–
𝐸 𝜏1
= 𝛽1
+𝛽2 βˆ™ 𝛾1
𝑖=1
𝑛
π‘₯
𝑖=1
𝑛
π‘₯
π‘₯𝑖= 𝛾0 + 𝛾1 βˆ™ 𝑃𝑖 + πœ—π‘–
True model:
π‘Œ = 𝛽0 + 𝛽1 βˆ™ 𝑃 + 𝛽2 βˆ™ π‘₯ + πœ€
We actually attempt to estimate:
π‘Œ = 𝜏0 + 𝜏1 βˆ™ 𝑃 + πœ–
Error term now contains:
𝛽2 βˆ™ π‘₯
𝐸 𝜏1 = 𝛽1 + 𝛽2 βˆ™ 𝐸
𝑖=1
𝑛
π‘₯𝑖 βˆ™ 𝑃𝑖 βˆ’ 𝑃
𝑖=1
𝑛
𝑃𝑖 βˆ’ 𝑃 2
𝐸 𝜏1 = 𝛽1 + 𝛽2 βˆ™ 𝛾1
𝐸 𝜏1 = 𝛽1 + 𝛽2 βˆ™ 𝛾1
The Actual
Causal Effect of
P on y
𝐸 𝜏1 = 𝛽1 + 𝛽2 βˆ™ 𝛾1
The actual
causal effect of
P on y
The actual causal effect of
the omitted variable X on
Y
𝐸 𝜏1 = 𝛽1 + 𝛽2 βˆ™ 𝛾1
The actual
causal effect of
P on y
Th”Effect” of P on x:
π‘₯𝑖= 𝛾0 + 𝛾1 βˆ™ 𝑃𝑖 + πœ—π‘–
The actual causal effect of
the omitted variable X on
Y
True model:
π‘Œ = 𝛽0 + 𝛽1 βˆ™ 𝑃 + 𝛽2 βˆ™ π‘₯ + πœ€
We actually attempt to estimate:
π‘Œ = 𝜏0 + 𝜏1 βˆ™ 𝑃 + πœ–
Error term now contains:
𝛽2 βˆ™ π‘₯
True model:
π‘Œ = 𝛽0 + 𝛽1 βˆ™ 𝑃 + 𝛽2 βˆ™ π‘₯ + πœ€
We actually attempt to estimate:
π‘Œ = 𝜏0 + 𝜏1 βˆ™ 𝑃 + πœ–
Error term now contains:
𝛽2 βˆ™ π‘₯
True model:
π‘Œ = 𝛽0 + 𝛽1 βˆ™ 𝑃 + 𝛽2 βˆ™ π‘₯ + πœ€
We actually attempt to estimate:
π‘Œ = 𝜏0 + 𝜏1 βˆ™ 𝑃 + πœ–
Error term now contains:
𝛽2 βˆ™ π‘₯
True model:
π‘Œ = 𝛽0 + 𝛽1 βˆ™ 𝑃 + 𝛽2 βˆ™ π‘₯ + πœ€
We actually attempt to estimate:
π‘Œ = 𝜏0 + 𝜏1 βˆ™ 𝑃 + πœ–
Error term now contains:
𝛽2 βˆ™ π‘₯
True model:
π‘Œ = 𝛽0 + 𝛽1 βˆ™ 𝑃 + 𝛽2 βˆ™ π‘₯ + πœ€
We actually attempt to estimate:
π‘Œ = 𝜏0 + 𝜏1 βˆ™ 𝑃 + πœ–
𝑬 𝝉 𝟏 β‰  𝜷
𝐸 𝜏1 = 𝛽1 + 𝛽2 βˆ™ 𝛾1
The actual
causal effect of
P on y
Th”Effect” of P on x:
π‘₯𝑖= 𝛾0 + 𝛾1 βˆ™ 𝑃𝑖 + πœ—π‘–
The actual causal effect of
the omitted variable X on
Y
𝐸 𝜏1 = 𝛽1 + 𝛽2 βˆ™ 𝛾1
The actual
causal effect of
P on y
Th”Effect” of P on x:
π‘₯𝑖= 𝛾0 + 𝛾1 βˆ™ 𝑃𝑖 + πœ—π‘–
The actual causal effect of
the omitted variable X on
Y
π‘Œ0
= 𝛽0 + 𝛽2 βˆ™ π‘₯1 + 𝛽3 βˆ™ π‘₯2 + 𝛽4 βˆ™ π‘₯2 + πœ€
π‘Œ1
= 𝛽0 + 𝛽1 + 𝛽2 βˆ™ π‘₯1 + 𝛽3 βˆ™ π‘₯2 + 𝛽4 βˆ™ π‘₯2 + πœ€
π‘Œ0
= 𝛽0 + 𝛽2 βˆ™ π‘₯1 + 𝛽3 βˆ™ π‘₯2 + 𝛽4 βˆ™ π‘₯2 + πœ€
π‘Œ1
= 𝛽0 + 𝛽1 + 𝛽2 βˆ™ π‘₯1 + 𝛽3 βˆ™ π‘₯2 + 𝛽4 βˆ™ π‘₯3 + πœ€
π‘Œ1 βˆ’ π‘Œ0
= 𝛽0 + 𝛽1 + πœ– βˆ’ 𝛽0 + πœ–
= 𝛽0 + 𝛽1 + πœ– βˆ’ 𝛽0 βˆ’ πœ–
= 𝛽1
π‘Œ1
βˆ’ π‘Œ0
= 𝛽0 + 𝛽1 + 𝛽2 βˆ™ π‘₯1 + 𝛽3 βˆ™ π‘₯2 + 𝛽4 βˆ™ π‘₯2 + πœ€
βˆ’ 𝛽0 + 𝛽2 βˆ™ π‘₯1 + 𝛽3 βˆ™ π‘₯2 + 𝛽4 βˆ™ π‘₯2 + πœ€
= 𝛽0 + 𝛽1 + πœ– βˆ’ 𝛽0 βˆ’ πœ–
= 𝛽1
π‘Œ1 βˆ’ π‘Œ0
= 𝛽1
π‘Œ = 𝑃 βˆ™ π‘Œ1
+ 1 βˆ’ 𝑃 βˆ™ π‘Œ0
= 𝑃 βˆ— 𝛽0 + 𝛽1 + πœ– + 1 βˆ’ 𝑃 βˆ— 𝛽0 + πœ–
= 𝑃 βˆ— 𝛽0 + 𝑃 βˆ— 𝛽1 + 𝑃 βˆ— πœ–
+𝛽0 + πœ– βˆ’ 𝑃 βˆ— 𝛽0 βˆ’ 𝑃 βˆ— πœ–
= 𝛽0 + 𝑃 βˆ— 𝛽1 + πœ–
π‘Œ = 𝑃 βˆ™ π‘Œ1
+ 1 βˆ’ 𝑃 βˆ™ π‘Œ0
= 𝑃 βˆ™ 𝛽0 + 𝛽1 + 𝛽2 βˆ™ π‘₯1 + 𝛽3 βˆ™ π‘₯2 + 𝛽4 βˆ™ π‘₯3 + πœ€
+ 1 βˆ’ 𝑃
βˆ™ 𝛽0 + 𝛽2 βˆ™ π‘₯ + 𝛽2 βˆ™ π‘₯1 + 𝛽3 βˆ™ π‘₯2 + 𝛽4 βˆ™ π‘₯3 + πœ€
π‘Œ = 𝛽0 + 𝛽1 βˆ™ 𝑃 + 𝛽2 βˆ™ π‘₯1 + 𝛽3 βˆ™ π‘₯2 + 𝛽4 βˆ™ π‘₯3 + πœ€
Cost of participation
𝐢 = 𝜌0 + 𝜌1 βˆ™ π‘₯1 + 𝜌2 βˆ™ π‘₯2 + 𝜌3 βˆ™ π‘₯3
True model:
π‘Œ = 𝛽0 + 𝛽1 βˆ™ 𝑃 + 𝛽2 βˆ™ π‘₯1 + 𝛽3 βˆ™ π‘₯2 + 𝛽4 βˆ™ π‘₯3 + πœ€
We actually attempt to estimate:
π‘Œ = 𝜏0 + 𝜏1 βˆ™ 𝑃 + πœ–
𝐸 𝜏1 = 𝛽1 + 𝛽2 βˆ™ 𝛾11 + 𝛽3 βˆ™ 𝛾21 + 𝛽4 βˆ™ 𝛾31
𝐸 𝜏1 = 𝛽1 + 𝛽2 βˆ™ 𝛾11 + 𝛽3 βˆ™ 𝛾21 + 𝛽4 βˆ™ 𝛾31
The actual
causal effect of
P on y
𝐸 𝜏1 = 𝛽1 + 𝛽2 βˆ™ 𝛾11 + 𝛽3 βˆ™ 𝛾21 + 𝛽4 βˆ™ 𝛾31
The actual
causal effect of
P on y
Th”Effect” of P on x1:
π‘₯1𝑖= 𝛾10 + 𝛾11 βˆ™ 𝑃𝑖 + πœ—1𝑖
The
actual
causal
effect of
the
omitted
variable
X1 on Y
𝐸 𝜏1 = 𝛽1 + 𝛽2 βˆ™ 𝛾11 + 𝛽3 βˆ™ 𝛾21 + 𝛽4 βˆ™ 𝛾31
The actual
causal effect of
P on y
”Effect” of P on x1:
π‘₯1𝑖= 𝛾10 + 𝛾11 βˆ™ 𝑃𝑖 + πœ—1𝑖
The
actual
causal
effect of
the
omitted
variable
X1 on Y
The actual causal effect
of the omitted variable
X2 on Y
Th”Effect” of P on x2:
π‘₯2𝑖= 𝛾20 + 𝛾21 βˆ™ 𝑃𝑖 + πœ—2𝑖
𝐸 𝜏1 = 𝛽1 + 𝛽2 βˆ™ 𝛾11 + 𝛽3 βˆ™ 𝛾21 + 𝛽4 βˆ™ 𝛾31
The actual
causal effect of
P on y
”Effect” of P on x1:
π‘₯1𝑖= 𝛾10 + 𝛾11 βˆ™ 𝑃𝑖 + πœ—1𝑖
The
actual
causal
effect of
the
omitted
variable
X1 on Y
The actual causal effect
of the omitted variable
X2 on Y
”Effect” of P on x2:
π‘₯2𝑖= 𝛾20 + 𝛾21 βˆ™ 𝑃𝑖 + πœ—2𝑖
Th”Effect” of P on x1:
π‘₯3𝑖= 𝛾30 + 𝛾31 βˆ™ 𝑃𝑖 + πœ—3𝑖
The actual causal effect of
the omitted variable X3 on Y
𝐸 𝜏1 = 𝛽1 + 𝛽2 βˆ™ 𝛾11 + 𝛽3 βˆ™ 𝛾21 + 𝛽4 βˆ™ 𝛾31
The actual
causal effect of
P on y
”Effect” of P on x1:
π‘₯1𝑖= 𝛾10 + 𝛾11 βˆ™ 𝑃𝑖 + πœ—1𝑖
The
actual
causal
effect of
the
omitted
variable
X1 on Y
The actual causal effect
of the omitted variable
X2 on Y
”Effect” of P on x2:
π‘₯2𝑖= 𝛾20 + 𝛾21 βˆ™ 𝑃𝑖 + πœ—2𝑖
Th”Effect” of P on x3:
π‘₯3𝑖= 𝛾30 + 𝛾31 βˆ™ 𝑃𝑖 + πœ—3𝑖
The actual causal effect of
the omitted variable X3 on Y
𝐸 𝜏1 = 𝛽1 + 𝛽2 βˆ™ 𝛾11 + 𝛽2 βˆ™ 𝛾21 + 𝛽3 βˆ™ 𝛾31
Potential outcomes
π‘Œ0
= 𝛽0 + 𝛽2 βˆ™ π‘₯1 + 𝛽3 βˆ™ π‘₯2 + πœ€
π‘Œ1
= 𝛽0 + 𝛽1 + 𝛽2 βˆ™ π‘₯1 + 𝛽3 βˆ™ π‘₯2 + πœ€
Costs of participation
𝐢 = 𝜌0 + 𝜌2 βˆ™ π‘₯2 + 𝜌3 βˆ™ π‘₯3
X2
Y
P
X1
X3
𝐸 𝜏1 = 𝛽1 + 𝛽2 βˆ™ 𝛾11 + 𝛽2 βˆ™ 𝛾21 + 𝛽3 βˆ™ 𝛾31
𝐸 𝜏1 = 𝛽1 + 𝛽2 βˆ™ 𝛾11 + 𝛽2 βˆ™ 𝛾21 + 𝛽3 βˆ™ 𝛾31
𝐸 𝜏1 = 𝛽1 + 𝛽2 βˆ™ 𝛾21 + 𝛽3 βˆ™ 𝛾31
X2
Y
P
X1
X3
𝐸 𝜏1 = 𝛽1 + 𝛽2 βˆ™ 𝛾21 + 𝛽3 βˆ™ 𝛾31
𝐸 𝜏1 = 𝛽1 + 𝛽2 βˆ™ 𝛾21
X2
Y
P
X1
X3
𝐸 𝜏1 = 𝛽1 + 𝛽2 βˆ™ 𝛾21
X2
Y
P
X1
X3
π‘Œ = 𝛽0 + 𝛽1 βˆ™ 𝑃 + 𝛽2 βˆ™ π‘₯1 + 𝛽3 βˆ™ π‘₯2 + 𝛽4 βˆ™ π‘₯3 + πœ€
π‘Œ = 𝛽0 + 𝛽1 βˆ™ 𝑃 + 𝛽2 βˆ™ π‘₯ + πœ€
𝐸 𝛽1 = 𝛽1
π‘Œ = 𝛽0 + 𝛽1 βˆ™ 𝑃 + 𝛽2 βˆ™ π‘₯ + πœ€
Regress π‘Œ on 𝑃 and π‘₯
𝐸 𝛽1 = 𝛽1
π‘Œ = 𝛽0 + 𝛽1 βˆ™ 𝑃 + 𝛽2 βˆ™ π‘₯ + πœ€
Regress π‘Œ on 𝑃 and π‘₯
𝐸 𝛽1 = 𝛽1
π‘Œ = 𝛽0 + 𝛽1 βˆ™ 𝑃 + 𝛽2 βˆ™ π‘₯ + πœ€
Regress π‘Œ on 𝑃 and π‘₯ and 𝑍
𝐸 𝛽1 β‰  𝛽1
π‘Œ = 𝛽0 + 𝛽1 βˆ™ 𝑃 + 𝛽2 βˆ™ π‘₯ + πœ€
Regress π‘Œ on 𝑃 and π‘₯ and 𝑍
𝐸 𝛽1 β‰  𝛽1
Bad controls
X2
Y
P
X1
X3
X2
Y
P
X1
X3
Z
π‘Œ = 𝛽0 + 𝛽1 βˆ™ 𝑃 + 𝛽𝑧 βˆ™ 𝑍 + πœ”
1.Is correlated with the regressor of
interest
2.Is correlated with the error term
π‘Œ = 𝛽0 + 𝛽1 βˆ™ 𝑃 + 𝛽𝑧 βˆ™ 𝑍 + πœ”
1.Is correlated with the regressor of
interest
2.Is correlated with the error term
π‘Œ = 𝛽0 + 𝛽1 βˆ™ 𝑃 + 𝛽𝑧 βˆ™ 𝑍 + πœ”
1.Is correlated with the regressor of
interest
2.Is correlated with the error term
π‘Œ = 𝛽0 + 𝛽1 βˆ™ 𝑃 + 𝛽𝑧 βˆ™ 𝑍 + πœ”
1.Is correlated with the regressor of
interest
2.Is correlated with the error term
Matching
X
Y
P
1.
𝐸 π‘Œ1
|𝑋 = 𝑐 β‰  𝐸 π‘Œ1
|𝑋 = π‘˜
𝐸 π‘Œ0
|𝑋 = 𝑐 β‰  𝐸 π‘Œ0
|𝑋 = π‘˜
2.
π‘ƒπ‘Ÿ 𝑃 = 1|𝑋 = 𝑐 β‰  π‘ƒπ‘Ÿ 𝑃 = 1|𝑋 = π‘˜
𝐸 𝑋|𝑃 = 1 β‰  𝐸 𝑋|𝑃 = 0
𝑋 =
0 if male
1 if female
𝐸 π‘Œ1
|𝑋 𝐸 π‘Œ0
|𝑋 𝐸 π‘Œ1
|𝑋 βˆ’ 𝐸 π‘Œ 0
|𝑋
𝑋 = 0 6 4 2
𝑋 = 1 5 1 4
𝐸 π‘Œ1
|𝑋 𝐸 π‘Œ0
|𝑋 𝐸 π‘Œ1
|𝑋 βˆ’ 𝐸 π‘Œ 0
|𝑋
𝑋 = 0 6 4 2
𝑋 = 1 5 1 4
𝐸 π‘Œ1
βˆ’ π‘Œ0
|𝑋 = 0 βˆ™ π‘ƒπ‘Ÿ 𝑋 = 0
+𝐸 π‘Œ1
βˆ’ π‘Œ0
|𝑋 = 1 βˆ™ π‘ƒπ‘Ÿ 𝑋 = 1
= 2 βˆ™ .5 + 4 βˆ™ .5
= 3
𝐸 π‘Œ1
|𝑋 𝐸 π‘Œ0
|𝑋 𝐸 π‘Œ1
|𝑋 βˆ’ 𝐸 π‘Œ 0
|𝑋
𝑋 = 0 6 4 2
𝑋 = 1 5 1 4
𝐸 π‘Œ1
|𝑋 = 0 βˆ™ π‘ƒπ‘Ÿ 𝑋 = 0
+𝐸 π‘Œ1
|𝑋 = 1 βˆ™ π‘ƒπ‘Ÿ 𝑋 = 1
= 6 βˆ™ .5 + 5 βˆ™ .5
= 5.5
𝐸 π‘Œ1
|𝑋 𝐸 π‘Œ0
|𝑋 𝐸 π‘Œ1
|𝑋 βˆ’ 𝐸 π‘Œ 0
|𝑋
𝑋 = 0 6 4 2
𝑋 = 1 5 1 4
𝐸 π‘Œ0
|𝑋 = 0 βˆ™ π‘ƒπ‘Ÿ 𝑋 = 0
+𝐸 π‘Œ0
|𝑋 = 1 βˆ™ π‘ƒπ‘Ÿ 𝑋 = 1
= 4 βˆ™ .5 + 1 βˆ™ .5
= 2.5
𝐸 𝑃 = 1|𝑋 𝐸 𝑃 = 0|𝑋
𝑋 = 0 .1 .9
𝑋 = 1 .5 .5
𝐸 𝑃 = 1|𝑋 = 0 βˆ™ π‘ƒπ‘Ÿ 𝑋 = 0
+𝐸 𝑃 = 1|𝑋 = 1 βˆ™ π‘ƒπ‘Ÿ 𝑋 = 1
= .1 βˆ™ .5 + .5 βˆ™ .5
= .3
𝐸 𝑃 = 1|𝑋 𝐸 𝑃 = 0|𝑋
𝑋 = 0 .1 .9
𝑋 = 1 .5 .5
𝐸 𝑃 = 1|𝑋 = 0 βˆ™ π‘ƒπ‘Ÿ 𝑋 = 0
+𝐸 𝑃 = 1|𝑋 = 1 βˆ™ π‘ƒπ‘Ÿ 𝑋 = 1
= .1 βˆ™ .5 + .5 βˆ™ .5
= .3
𝑛 = 1000
𝑖=1
𝑛
𝑋𝑖
𝑛
=
𝑖=1
1000
𝑋𝑖
1000
β‰ˆ 500
𝑖=1
𝑛
𝑃𝑖
𝑛
=
𝑖=1
1000
𝑃𝑖
1000
β‰ˆ 300
𝑛 = 1000
𝑖=1
𝑛
𝑋𝑖
𝑛
=
𝑖=1
1000
𝑋𝑖
1000
β‰ˆ .5
𝑖=1
𝑛
𝑃𝑖
𝑛
=
𝑖=1
1000
𝑃𝑖
1000
β‰ˆ 300
𝑛 = 1000
𝑖=1
𝑛
𝑋𝑖
𝑛
=
𝑖=1
1000
𝑋𝑖
1000
β‰ˆ .5
𝑖=1
𝑛
𝑃𝑖
𝑛
=
𝑖=1
1000
𝑃𝑖
1000
β‰ˆ .3
𝑗=1
300
π‘Œπ‘—
300
: 6 βˆ™ .16667 + 5 βˆ™ .8333 = 5.16652
𝑗=1
300
π‘Œπ‘—
300
: 6 βˆ™ .16667 + 5 βˆ™ .8333 = 5.16652
𝑗=1
700
π‘Œπ‘—
700
: 4 βˆ™ .6428 + 1 βˆ™ .3571 = 2.9286
?
𝑖 = 237
𝑋237 = 0
𝑃237 = 1
π‘Œ237 = π‘Œ237
1
= 5.3
𝑖 = 237
𝑋237 = 0
𝑃237 = 1
π‘Œ237 = π‘Œ237
1
= 5.3
𝑖 = 237
𝑋237 = 0
𝑃237 = 1
π‘Œ237 = π‘Œ237
1
= 5.3
𝑖 = 237
𝑋237 = 0
𝑃237 = 1
π‘Œ237 = π‘Œ237
1
= 5.3
𝑖 = 237
𝑋237 = 0
𝑃237 = 1
π‘Œ237 = π‘Œ237
1
= 6.3
𝐼237 = π‘Œ237
1
βˆ’ π‘Œ237
0
= 6.3βˆ’?
π‘Œ237
0
π‘Œ237
0
= π‘Œ766
0
π‘Œ237
0
=
π‘Œ48
0
+ π‘Œ109
0
+ π‘Œ418
0
+ π‘Œ505
0
+ π‘Œ919
0
5
𝐼𝑖 = π‘Œπ‘–
1
βˆ’ π‘Œπ‘–
0
𝐴𝑇𝐸 =
𝑖=1
𝑛
𝐼𝑖
𝑛
π‘˜=0
𝐾
𝐸 𝐼|𝑋 = π‘˜ βˆ™ π‘ƒπ‘Ÿ 𝑋 = π‘˜
=
π‘˜=0
𝐾
𝐸 π‘Œ1
βˆ’ π‘Œ0
|𝑋 = π‘˜ βˆ™ π‘ƒπ‘Ÿ 𝑋 = π‘˜
π‘˜=0
𝐾
𝐸 𝐼|𝑋 = π‘˜, 𝑃 = 1 βˆ™ π‘ƒπ‘Ÿ 𝑋 = π‘˜|𝑃 = 1
=
π‘˜=0
𝐾
𝐸 π‘Œ1
βˆ’ π‘Œ0
|𝑋 = π‘˜, 𝑃 = 1 βˆ™ π‘ƒπ‘Ÿ 𝑋 = π‘˜|𝑃 = 1
𝐼237 = π‘Œ237
1
βˆ’ π‘Œ237
0
= 6.3βˆ’?
π‘˜=0
𝐾
𝐸 𝐼|𝑋 = π‘˜ βˆ™ π‘ƒπ‘Ÿ 𝑋 = π‘˜
=
π‘˜=0
𝐾
𝐸 π‘Œ1
βˆ’ π‘Œ0
|𝑋 = π‘˜ βˆ™ π‘ƒπ‘Ÿ 𝑋 = π‘˜
X
Y
P
2X Gender-male/female
5X SES-5 quintiles
4X Age-broken up into 4 categories
2X Sector-Urban/Rural
6X Religion-Christian, Muslim,
Hindu, Buddhist, Jewish,
Traditional.
6X Occupation-6 occupation types
2 Insurance-Insured/not insured
____
5,760 potential β€œtypes”
2X Gender-male/female
5X SES-5 quintiles
4X Age-broken up into 4 categories
2X Sector-Urban/Rural
6X Religion-Christian, Muslim,
Hindu, Buddhist, Jewish,
Traditional.
6X Occupation-6 occupation types
2 Insurance-Insured/not insured
____
2 potential β€œtypes”
2X Gender-male/female
5X SES-5 quintiles
4X Age-broken up into 4 categories
2X Sector-Urban/Rural
6X Religion-Christian, Muslim,
Hindu, Buddhist, Jewish,
Traditional.
6X Occupation-6 occupation types
2 Insurance-Insured/not insured
____
10 potential β€œtypes”
2X Gender-male/female
5X SES-5 quintiles
4X Age-broken up into 4 categories
2X Sector-Urban/Rural
6X Religion-Christian, Muslim,
Hindu, Buddhist, Jewish,
Traditional.
6X Occupation-6 occupation types
2 Insurance-Insured/not insured
____
40 potential β€œtypes”
2X Gender-male/female
5X SES-5 quintiles
4X Age-broken up into 4 categories
2X Sector-Urban/Rural
6X Religion-Christian, Muslim,
Hindu, Buddhist, Jewish,
Traditional.
6X Occupation-6 occupation types
2 Insurance-Insured/not insured
____
80 potential β€œtypes”
2X Gender-male/female
5X SES-5 quintiles
4X Age-broken up into 4 categories
2X Sector-Urban/Rural
6X Religion-Christian, Muslim,
Hindu, Buddhist, Jewish,
Traditional.
6X Occupation-6 occupation types
2 Insurance-Insured/not insured
____
480 potential β€œtypes”
2X Gender-male/female
5X SES-5 quintiles
4X Age-broken up into 4 categories
2X Sector-Urban/Rural
6X Religion-Christian, Muslim,
Hindu, Buddhist, Jewish,
Traditional.
6X Occupation-6 occupation types
2 Insurance-Insured/not insured
____
2,880 potential β€œtypes”
2X Gender-male/female
5X SES-5 quintiles
4X Age-broken up into 4 categories
2X Sector-Urban/Rural
6X Religion-Christian, Muslim,
Hindu, Buddhist, Jewish,
Traditional.
6X Occupation-6 occupation types
2 Insurance-Insured/not insured
____
5,760 potential β€œtypes”
2X Gender-male/female
5X SES-5 quintiles
4X Age-broken up into 4 categories
2X Sector-Urban/Rural
6X Religion-Christian, Muslim,
Hindu, Buddhist, Jewish,
Traditional.
6X Occupation-6 occupation types
2 Insurance-Insured/not insured
____
5,760 potential β€œtypes”
Pr(𝑃 = 1|𝑋)
𝐼237 = π‘Œ237
1
βˆ’ π‘Œ237
0
= 6.3βˆ’?
𝐼237 = π‘Œ237
1
βˆ’ π‘Œ237
0
= 6.3βˆ’?
Pr(𝑃 = 1|𝑋)
π‘ƒπ‘Ÿ(𝑃 = 1|𝑋)
π‘ƒπ‘Ÿ 𝑃 = 1 𝑋 =
𝑒π‘₯𝑝 𝛽0 + 𝛽1 βˆ™ 𝑋
1 + 𝑒π‘₯𝑝 𝛽0 + 𝛽1 βˆ™ 𝑋
The basics of the method
1. Pool your sample of participants and non-participants
and define various characteristics 𝑋.
2. Estimate the probability of program participation
conditional on
Pr 𝑃 = 1|𝑋
3. Compute the propensity score using the fitted binary
participation model.
π‘ƒπ‘Ÿ 𝑃𝑖 = 1|𝑋𝑖
4. Find a counterfactual outcome for each individual by
identifying some individual who did experience the
counterfactual conditions and had a similar propensity score
Matching approaches
1. Nearest neighbor
2. Caliper
3. (Budget) Caliper
4. Weighting
Matching approaches
1. Nearest neighbor
2. Caliper
3. (Budget) Caliper
4. Weighting
Balancing property
Pr 𝑃 = 1|𝑋
Propensity score
0
1
P=0 P=1
Common
support
Propensity score
0
1
P=0 P=1
Failure of
common
support
Other applications of the
propensity score
1.Weighting
2.Regression
Other applications of the
propensity score
1.Weighting
2.Regression
𝐸 𝑃 = 1|𝑋 𝐸 𝑃 = 0|𝑋
𝑋 = 0 .1 .9
𝑋 = 1 .5 .5
𝐸 𝑃 = 1|𝑋 = 0 βˆ™ π‘ƒπ‘Ÿ 𝑋 = 0
+𝐸 𝑃 = 1|𝑋 = 1 βˆ™ π‘ƒπ‘Ÿ 𝑋 = 1
= .1 βˆ™ .5 + .5 βˆ™ .5
= .3
1,000 observations
𝐸 𝑃 = 1|𝑋 𝐸 𝑃 = 0|𝑋
𝑋 = 0 .1 .9
𝑋 = 1 .5 .5
𝐸 𝑃 = 1|𝑋 = 0 βˆ™ π‘ƒπ‘Ÿ 𝑋 = 0
+𝐸 𝑃 = 1|𝑋 = 1 βˆ™ π‘ƒπ‘Ÿ 𝑋 = 1
= .1 βˆ™ .5 + .5 βˆ™ .5
= .3
1,000 observations
300 participants
𝐸 𝑃 = 1|𝑋 𝐸 𝑃 = 0|𝑋
𝑋 = 0 .1 .9
𝑋 = 1 .5 .5
𝐸 𝑃 = 1|𝑋 = 0 βˆ™ π‘ƒπ‘Ÿ 𝑋 = 0
+𝐸 𝑃 = 1|𝑋 = 1 βˆ™ π‘ƒπ‘Ÿ 𝑋 = 1
= .1 βˆ™ .5 + .5 βˆ™ .5
= .3
1,000 observations
300 participants 700 non-participants
𝐸 𝑃 = 1|𝑋 𝐸 𝑃 = 0|𝑋
𝑋 = 0 .1 .9
𝑋 = 1 .5 .5
𝐸 𝑃 = 1|𝑋 = 0 βˆ™ π‘ƒπ‘Ÿ 𝑋 = 0
+𝐸 𝑃 = 1|𝑋 = 1 βˆ™ π‘ƒπ‘Ÿ 𝑋 = 1
= .1 βˆ™ .5 + .5 βˆ™ .5
= .3
1,000 observations
300 participants 700 non-participants
50 men 250 women
𝐸 𝑃 = 1|𝑋 𝐸 𝑃 = 0|𝑋
𝑋 = 0 .1 .9
𝑋 = 1 .5 .5
𝐸 𝑃 = 1|𝑋 = 0 βˆ™ π‘ƒπ‘Ÿ 𝑋 = 0
+𝐸 𝑃 = 1|𝑋 = 1 βˆ™ π‘ƒπ‘Ÿ 𝑋 = 1
= .1 βˆ™ .5 + .5 βˆ™ .5
= .3
1,000 observations
300 participants 700 non-participants
50 men 250 women 450 men 250 women
𝐸 𝑃 = 1|𝑋 𝐸 𝑃 = 0|𝑋
𝑋 = 0 .1 .9
𝑋 = 1 .5 .5
𝐸 𝑃 = 1|𝑋 = 0 βˆ™ π‘ƒπ‘Ÿ 𝑋 = 0
+𝐸 𝑃 = 1|𝑋 = 1 βˆ™ π‘ƒπ‘Ÿ 𝑋 = 1
= .1 βˆ™ .5 + .5 βˆ™ .5
= .3
1,000 observations
300 participants 700 non-participants
500 men 500 women 450 men 250 women
𝐸 𝑃 = 1|𝑋 𝐸 𝑃 = 0|𝑋
𝑋 = 0 .1 .9
𝑋 = 1 .5 .5
𝐸 𝑃 = 1|𝑋 = 0 βˆ™ π‘ƒπ‘Ÿ 𝑋 = 0
+𝐸 𝑃 = 1|𝑋 = 1 βˆ™ π‘ƒπ‘Ÿ 𝑋 = 1
= .1 βˆ™ .5 + .5 βˆ™ .5
= .3
1,000 observations
300 participants 700 non-participants
500 men 500 women 500 men 500 women
π‘Šπ‘Œπ‘– =
𝑃𝑖 βˆ™ π‘Œπ‘–
π‘ƒπ‘Ÿ 𝑃𝑖 = 1|𝑋𝑖
βˆ’
1 βˆ’ 𝑃𝑖 βˆ™ π‘Œπ‘–
1 βˆ’ π‘ƒπ‘Ÿ 𝑃𝑖 = 1|𝑋𝑖
𝐴𝑇𝐸 =
𝑖=1
𝑛
π‘Šπ‘Œπ‘–
𝑖=1
𝑛 𝑃𝑖
π‘ƒπ‘Ÿ 𝑃𝑖 = 1|𝑋𝑖
+
1 βˆ’ 𝑃𝑖
1 βˆ’ π‘ƒπ‘Ÿ 𝑃𝑖 = 1|𝑋𝑖
π‘Šπ‘Œπ‘– =
𝑃𝑖 βˆ™ π‘Œπ‘–
π‘ƒπ‘Ÿ 𝑃𝑖 = 1|𝑋𝑖
βˆ’
1 βˆ’ 𝑃𝑖 βˆ™ π‘Œπ‘–
1 βˆ’ π‘ƒπ‘Ÿ 𝑃𝑖 = 1|𝑋𝑖
𝐴𝑇𝐸 =
𝑖=1
𝑛
π‘Šπ‘Œπ‘–
π’Š=𝟏
𝒏 π‘·π’Š
𝑷𝒓 π‘·π’Š = 𝟏|π‘Ώπ’Š
+
𝟏 βˆ’ π‘·π’Š
𝟏 βˆ’ 𝑷𝒓 π‘·π’Š = 𝟏|π‘Ώπ’Š
π‘Šπ‘Œπ‘– =
𝑃𝑖 βˆ™ π‘Œπ‘–
π‘ƒπ‘Ÿ 𝑃𝑖 = 1|𝑋𝑖
βˆ’
1 βˆ’ 𝑃𝑖 βˆ™ π‘Œπ‘–
1 βˆ’ π‘ƒπ‘Ÿ 𝑃𝑖 = 1|𝑋𝑖
𝐴𝑇𝐸 =
𝑖=1
𝑛
π‘Šπ‘Œπ‘–
𝑖=1
𝑛 𝑃𝑖
π‘ƒπ‘Ÿ 𝑃𝑖 = 1|𝑋𝑖
+
1 βˆ’ 𝑃𝑖
1 βˆ’ π‘ƒπ‘Ÿ 𝑃𝑖 = 1|𝑋𝑖
Other applications of the
propensity score
1.Weighting
2.Regression
Regress π‘Œπ‘– on 𝑃𝑖 and π‘ƒπ‘Ÿ 𝑃𝑖 = 1|𝑋𝑖
Regress π‘Œπ‘– on 𝑃𝑖 and π‘ƒπ‘Ÿ 𝑃𝑖 = 1|𝑋𝑖
Conclusion
Links:
The manual:
http://www.measureevaluation.org/resources/publications/ms-
14-87-en
The webinar introducing the manual:
http://www.measureevaluation.org/resources/webinars/metho
ds-for-program-impact-evaluation
My email:
pmlance@email.unc.edu
MEASURE Evaluation is funded by the U.S. Agency
for International Development (USAID) under terms
of Cooperative Agreement AID-OAA-L-14-00004 and
implemented by the Carolina Population Center, University
of North Carolina at Chapel Hill in partnership with ICF
International, John Snow, Inc., Management Sciences for
Health, Palladium Group, and Tulane University. The views
expressed in this presentation do not necessarily reflect
the views of USAID or the United States government.
www.measureevaluation.org

Mais conteΓΊdo relacionado

Destaque

Daftar isi
Daftar isiDaftar isi
Daftar isineni teh
Β 
Bagian 3
Bagian 3Bagian 3
Bagian 3neni teh
Β 
Bagian 2
Bagian 2Bagian 2
Bagian 2neni teh
Β 
Bagian 4
Bagian 4Bagian 4
Bagian 4neni teh
Β 
Bagian 5
Bagian 5Bagian 5
Bagian 5neni teh
Β 
Daftar pustaka
Daftar pustakaDaftar pustaka
Daftar pustakaneni teh
Β 
Addressing Complexity in the Impact Evaluation of the Cross-Border Health Int...
Addressing Complexity in the Impact Evaluation of the Cross-Border Health Int...Addressing Complexity in the Impact Evaluation of the Cross-Border Health Int...
Addressing Complexity in the Impact Evaluation of the Cross-Border Health Int...MEASURE Evaluation
Β 
Measuring Ethnic and Sexual Identities: Lessons from Two Studies in Central A...
Measuring Ethnic and Sexual Identities: Lessons from Two Studies in Central A...Measuring Ethnic and Sexual Identities: Lessons from Two Studies in Central A...
Measuring Ethnic and Sexual Identities: Lessons from Two Studies in Central A...MEASURE Evaluation
Β 
Using the PLACE Method to Inform Decision Making
Using the PLACE Method to Inform Decision MakingUsing the PLACE Method to Inform Decision Making
Using the PLACE Method to Inform Decision MakingMEASURE Evaluation
Β 
Community Trace and Verify in Tanzania
Community Trace and Verify in TanzaniaCommunity Trace and Verify in Tanzania
Community Trace and Verify in TanzaniaMEASURE Evaluation
Β 
Interoperability & Crowdsourcing: Can these improve the management of ANC pro...
Interoperability & Crowdsourcing: Can these improve the management of ANC pro...Interoperability & Crowdsourcing: Can these improve the management of ANC pro...
Interoperability & Crowdsourcing: Can these improve the management of ANC pro...MEASURE Evaluation
Β 
The Prevalence, Experience and Management of Pain
The Prevalence, Experience and Management of PainThe Prevalence, Experience and Management of Pain
The Prevalence, Experience and Management of PainMEASURE Evaluation
Β 
Including AIDS-affected young people in OVC research: Challenges and opportu...
Including AIDS-affected young people in OVC research:  Challenges and opportu...Including AIDS-affected young people in OVC research:  Challenges and opportu...
Including AIDS-affected young people in OVC research: Challenges and opportu...MEASURE Evaluation
Β 
Monitoring Scale-up of Health Practices and Interventions
Monitoring Scale-up of Health Practices and InterventionsMonitoring Scale-up of Health Practices and Interventions
Monitoring Scale-up of Health Practices and InterventionsMEASURE Evaluation
Β 
Measuring Success in Repositioning Family Planning
Measuring Success in Repositioning Family PlanningMeasuring Success in Repositioning Family Planning
Measuring Success in Repositioning Family PlanningMEASURE Evaluation
Β 
RHINO Forum: How can RHIS improve the delivery of HIV/AIDS services?
RHINO Forum: How can RHIS improve the delivery of HIV/AIDS services?RHINO Forum: How can RHIS improve the delivery of HIV/AIDS services?
RHINO Forum: How can RHIS improve the delivery of HIV/AIDS services? MEASURE Evaluation
Β 
Assessing HIV Service: Use and Information Systems for Key Populations in Nam...
Assessing HIV Service: Use and Information Systems for Key Populations in Nam...Assessing HIV Service: Use and Information Systems for Key Populations in Nam...
Assessing HIV Service: Use and Information Systems for Key Populations in Nam...MEASURE Evaluation
Β 
Integration as a Health Systems Strengthening Intervention: Case Studies from...
Integration as a Health Systems Strengthening Intervention: Case Studies from...Integration as a Health Systems Strengthening Intervention: Case Studies from...
Integration as a Health Systems Strengthening Intervention: Case Studies from...MEASURE Evaluation
Β 
RHINO Forum Kickoff: iHRIS Open Source HR Information Solutions
RHINO Forum Kickoff: iHRIS Open Source HR Information SolutionsRHINO Forum Kickoff: iHRIS Open Source HR Information Solutions
RHINO Forum Kickoff: iHRIS Open Source HR Information SolutionsRoutine Health Information Network
Β 
Assessment of Constraints to Data Use
Assessment of Constraints to Data UseAssessment of Constraints to Data Use
Assessment of Constraints to Data UseMEASURE Evaluation
Β 

Destaque (20)

Daftar isi
Daftar isiDaftar isi
Daftar isi
Β 
Bagian 3
Bagian 3Bagian 3
Bagian 3
Β 
Bagian 2
Bagian 2Bagian 2
Bagian 2
Β 
Bagian 4
Bagian 4Bagian 4
Bagian 4
Β 
Bagian 5
Bagian 5Bagian 5
Bagian 5
Β 
Daftar pustaka
Daftar pustakaDaftar pustaka
Daftar pustaka
Β 
Addressing Complexity in the Impact Evaluation of the Cross-Border Health Int...
Addressing Complexity in the Impact Evaluation of the Cross-Border Health Int...Addressing Complexity in the Impact Evaluation of the Cross-Border Health Int...
Addressing Complexity in the Impact Evaluation of the Cross-Border Health Int...
Β 
Measuring Ethnic and Sexual Identities: Lessons from Two Studies in Central A...
Measuring Ethnic and Sexual Identities: Lessons from Two Studies in Central A...Measuring Ethnic and Sexual Identities: Lessons from Two Studies in Central A...
Measuring Ethnic and Sexual Identities: Lessons from Two Studies in Central A...
Β 
Using the PLACE Method to Inform Decision Making
Using the PLACE Method to Inform Decision MakingUsing the PLACE Method to Inform Decision Making
Using the PLACE Method to Inform Decision Making
Β 
Community Trace and Verify in Tanzania
Community Trace and Verify in TanzaniaCommunity Trace and Verify in Tanzania
Community Trace and Verify in Tanzania
Β 
Interoperability & Crowdsourcing: Can these improve the management of ANC pro...
Interoperability & Crowdsourcing: Can these improve the management of ANC pro...Interoperability & Crowdsourcing: Can these improve the management of ANC pro...
Interoperability & Crowdsourcing: Can these improve the management of ANC pro...
Β 
The Prevalence, Experience and Management of Pain
The Prevalence, Experience and Management of PainThe Prevalence, Experience and Management of Pain
The Prevalence, Experience and Management of Pain
Β 
Including AIDS-affected young people in OVC research: Challenges and opportu...
Including AIDS-affected young people in OVC research:  Challenges and opportu...Including AIDS-affected young people in OVC research:  Challenges and opportu...
Including AIDS-affected young people in OVC research: Challenges and opportu...
Β 
Monitoring Scale-up of Health Practices and Interventions
Monitoring Scale-up of Health Practices and InterventionsMonitoring Scale-up of Health Practices and Interventions
Monitoring Scale-up of Health Practices and Interventions
Β 
Measuring Success in Repositioning Family Planning
Measuring Success in Repositioning Family PlanningMeasuring Success in Repositioning Family Planning
Measuring Success in Repositioning Family Planning
Β 
RHINO Forum: How can RHIS improve the delivery of HIV/AIDS services?
RHINO Forum: How can RHIS improve the delivery of HIV/AIDS services?RHINO Forum: How can RHIS improve the delivery of HIV/AIDS services?
RHINO Forum: How can RHIS improve the delivery of HIV/AIDS services?
Β 
Assessing HIV Service: Use and Information Systems for Key Populations in Nam...
Assessing HIV Service: Use and Information Systems for Key Populations in Nam...Assessing HIV Service: Use and Information Systems for Key Populations in Nam...
Assessing HIV Service: Use and Information Systems for Key Populations in Nam...
Β 
Integration as a Health Systems Strengthening Intervention: Case Studies from...
Integration as a Health Systems Strengthening Intervention: Case Studies from...Integration as a Health Systems Strengthening Intervention: Case Studies from...
Integration as a Health Systems Strengthening Intervention: Case Studies from...
Β 
RHINO Forum Kickoff: iHRIS Open Source HR Information Solutions
RHINO Forum Kickoff: iHRIS Open Source HR Information SolutionsRHINO Forum Kickoff: iHRIS Open Source HR Information Solutions
RHINO Forum Kickoff: iHRIS Open Source HR Information Solutions
Β 
Assessment of Constraints to Data Use
Assessment of Constraints to Data UseAssessment of Constraints to Data Use
Assessment of Constraints to Data Use
Β 

Semelhante a Selection on Observables

SUEC 高中 Adv Maths (Irrational Part 3)
SUEC 高中 Adv Maths (Irrational Part 3)SUEC 高中 Adv Maths (Irrational Part 3)
SUEC 高中 Adv Maths (Irrational Part 3)tungwc
Β 
SUEC 高中 Adv Maths (2 Roots Part 2)
SUEC 高中 Adv Maths (2 Roots Part 2)SUEC 高中 Adv Maths (2 Roots Part 2)
SUEC 高中 Adv Maths (2 Roots Part 2)tungwc
Β 
SUEC 高中 Adv Maths (Biquadratic Equation, Method of Changing the Variable, Rec...
SUEC 高中 Adv Maths (Biquadratic Equation, Method of Changing the Variable, Rec...SUEC 高中 Adv Maths (Biquadratic Equation, Method of Changing the Variable, Rec...
SUEC 高中 Adv Maths (Biquadratic Equation, Method of Changing the Variable, Rec...tungwc
Β 
SUEC 高中 Adv Maths (2 Roots Part 1)
SUEC 高中 Adv Maths (2 Roots Part 1)SUEC 高中 Adv Maths (2 Roots Part 1)
SUEC 高中 Adv Maths (2 Roots Part 1)tungwc
Β 
SUEC 高中 Adv Maths (Quadratic Equation in One Variable)
SUEC 高中 Adv Maths (Quadratic Equation in One Variable)SUEC 高中 Adv Maths (Quadratic Equation in One Variable)
SUEC 高中 Adv Maths (Quadratic Equation in One Variable)tungwc
Β 
Cheatsheet - FΓ³rmulas de FΓ­sica para FΓ­sica General y FΓ­sica de Ingenieros
Cheatsheet - FΓ³rmulas de FΓ­sica para FΓ­sica General y FΓ­sica de IngenierosCheatsheet - FΓ³rmulas de FΓ­sica para FΓ­sica General y FΓ­sica de Ingenieros
Cheatsheet - FΓ³rmulas de FΓ­sica para FΓ­sica General y FΓ­sica de IngenierosJose Perez
Β 
SUEC 高中 Adv Maths (Irrational Part 2)
SUEC 高中 Adv Maths (Irrational Part 2)SUEC 高中 Adv Maths (Irrational Part 2)
SUEC 高中 Adv Maths (Irrational Part 2)tungwc
Β 
Carbohydrate Metabolism
Carbohydrate MetabolismCarbohydrate Metabolism
Carbohydrate MetabolismVedantPatel100
Β 
Pembuktian sifat sifat logaritma
Pembuktian sifat sifat logaritmaPembuktian sifat sifat logaritma
Pembuktian sifat sifat logaritmaFranxisca Kurniawati
Β 
Ecuacion diferencial de la forma u=ax+bx+c
Ecuacion diferencial de la forma u=ax+bx+cEcuacion diferencial de la forma u=ax+bx+c
Ecuacion diferencial de la forma u=ax+bx+cEduardo Pila
Β 
Math-7-Lesson-9-Irrational-Numbers.pdf
Math-7-Lesson-9-Irrational-Numbers.pdfMath-7-Lesson-9-Irrational-Numbers.pdf
Math-7-Lesson-9-Irrational-Numbers.pdfVelodonaTancio
Β 
Lipid Metabolism
Lipid MetabolismLipid Metabolism
Lipid MetabolismVedantPatel100
Β 
12 Proof 1 = 2
12 Proof 1 = 212 Proof 1 = 2
12 Proof 1 = 2AhmedRaza283
Β 
SUEC 高中 Adv Maths (Polynomial Function)
SUEC 高中 Adv Maths (Polynomial Function)SUEC 高中 Adv Maths (Polynomial Function)
SUEC 高中 Adv Maths (Polynomial Function)tungwc
Β 

Semelhante a Selection on Observables (20)

SUEC 高中 Adv Maths (Irrational Part 3)
SUEC 高中 Adv Maths (Irrational Part 3)SUEC 高中 Adv Maths (Irrational Part 3)
SUEC 高中 Adv Maths (Irrational Part 3)
Β 
SUEC 高中 Adv Maths (2 Roots Part 2)
SUEC 高中 Adv Maths (2 Roots Part 2)SUEC 高中 Adv Maths (2 Roots Part 2)
SUEC 高中 Adv Maths (2 Roots Part 2)
Β 
SUEC 高中 Adv Maths (Biquadratic Equation, Method of Changing the Variable, Rec...
SUEC 高中 Adv Maths (Biquadratic Equation, Method of Changing the Variable, Rec...SUEC 高中 Adv Maths (Biquadratic Equation, Method of Changing the Variable, Rec...
SUEC 高中 Adv Maths (Biquadratic Equation, Method of Changing the Variable, Rec...
Β 
SUEC 高中 Adv Maths (2 Roots Part 1)
SUEC 高中 Adv Maths (2 Roots Part 1)SUEC 高中 Adv Maths (2 Roots Part 1)
SUEC 高中 Adv Maths (2 Roots Part 1)
Β 
SUEC 高中 Adv Maths (Quadratic Equation in One Variable)
SUEC 高中 Adv Maths (Quadratic Equation in One Variable)SUEC 高中 Adv Maths (Quadratic Equation in One Variable)
SUEC 高中 Adv Maths (Quadratic Equation in One Variable)
Β 
Cheatsheet - FΓ³rmulas de FΓ­sica para FΓ­sica General y FΓ­sica de Ingenieros
Cheatsheet - FΓ³rmulas de FΓ­sica para FΓ­sica General y FΓ­sica de IngenierosCheatsheet - FΓ³rmulas de FΓ­sica para FΓ­sica General y FΓ­sica de Ingenieros
Cheatsheet - FΓ³rmulas de FΓ­sica para FΓ­sica General y FΓ­sica de Ingenieros
Β 
SUEC 高中 Adv Maths (Irrational Part 2)
SUEC 高中 Adv Maths (Irrational Part 2)SUEC 高中 Adv Maths (Irrational Part 2)
SUEC 高中 Adv Maths (Irrational Part 2)
Β 
Carbohydrate Metabolism
Carbohydrate MetabolismCarbohydrate Metabolism
Carbohydrate Metabolism
Β 
Pembuktian sifat sifat logaritma
Pembuktian sifat sifat logaritmaPembuktian sifat sifat logaritma
Pembuktian sifat sifat logaritma
Β 
Sets
SetsSets
Sets
Β 
Relativity
RelativityRelativity
Relativity
Β 
Ecuacion diferencial de la forma u=ax+bx+c
Ecuacion diferencial de la forma u=ax+bx+cEcuacion diferencial de la forma u=ax+bx+c
Ecuacion diferencial de la forma u=ax+bx+c
Β 
08.sdcd_ransformada_z
08.sdcd_ransformada_z08.sdcd_ransformada_z
08.sdcd_ransformada_z
Β 
Photosynthesis
PhotosynthesisPhotosynthesis
Photosynthesis
Β 
07.5.scd_ejemplos
07.5.scd_ejemplos07.5.scd_ejemplos
07.5.scd_ejemplos
Β 
Math-7-Lesson-9-Irrational-Numbers.pdf
Math-7-Lesson-9-Irrational-Numbers.pdfMath-7-Lesson-9-Irrational-Numbers.pdf
Math-7-Lesson-9-Irrational-Numbers.pdf
Β 
Lipid Metabolism
Lipid MetabolismLipid Metabolism
Lipid Metabolism
Β 
04.mdsd_laplace_fourier
04.mdsd_laplace_fourier04.mdsd_laplace_fourier
04.mdsd_laplace_fourier
Β 
12 Proof 1 = 2
12 Proof 1 = 212 Proof 1 = 2
12 Proof 1 = 2
Β 
SUEC 高中 Adv Maths (Polynomial Function)
SUEC 高中 Adv Maths (Polynomial Function)SUEC 高中 Adv Maths (Polynomial Function)
SUEC 高中 Adv Maths (Polynomial Function)
Β 

Mais de MEASURE Evaluation

Managing missing values in routinely reported data: One approach from the Dem...
Managing missing values in routinely reported data: One approach from the Dem...Managing missing values in routinely reported data: One approach from the Dem...
Managing missing values in routinely reported data: One approach from the Dem...MEASURE Evaluation
Β 
Use of Routine Data for Economic Evaluations
Use of Routine Data for Economic EvaluationsUse of Routine Data for Economic Evaluations
Use of Routine Data for Economic EvaluationsMEASURE Evaluation
Β 
Routine data use in evaluation: practical guidance
Routine data use in evaluation: practical guidanceRoutine data use in evaluation: practical guidance
Routine data use in evaluation: practical guidanceMEASURE Evaluation
Β 
Tuberculosis/HIV Mobility Study: Objectives and Background
Tuberculosis/HIV Mobility Study: Objectives and BackgroundTuberculosis/HIV Mobility Study: Objectives and Background
Tuberculosis/HIV Mobility Study: Objectives and BackgroundMEASURE Evaluation
Β 
How to improve the capabilities of health information systems to address emer...
How to improve the capabilities of health information systems to address emer...How to improve the capabilities of health information systems to address emer...
How to improve the capabilities of health information systems to address emer...MEASURE Evaluation
Β 
LCI Evaluation Uganda Organizational Network Analysis
LCI Evaluation Uganda Organizational Network AnalysisLCI Evaluation Uganda Organizational Network Analysis
LCI Evaluation Uganda Organizational Network AnalysisMEASURE Evaluation
Β 
Using Organizational Network Analysis to Plan and Evaluate Global Health Prog...
Using Organizational Network Analysis to Plan and Evaluate Global Health Prog...Using Organizational Network Analysis to Plan and Evaluate Global Health Prog...
Using Organizational Network Analysis to Plan and Evaluate Global Health Prog...MEASURE Evaluation
Β 
Understanding Referral Networks for Adolescent Girls and Young Women
Understanding Referral Networks for Adolescent Girls and Young WomenUnderstanding Referral Networks for Adolescent Girls and Young Women
Understanding Referral Networks for Adolescent Girls and Young WomenMEASURE Evaluation
Β 
Data for Impact: Lessons Learned in Using the Ripple Effects Mapping Method
Data for Impact: Lessons Learned in Using the Ripple Effects Mapping MethodData for Impact: Lessons Learned in Using the Ripple Effects Mapping Method
Data for Impact: Lessons Learned in Using the Ripple Effects Mapping MethodMEASURE Evaluation
Β 
Local Capacity Initiative (LCI) Evaluation
Local Capacity Initiative (LCI) EvaluationLocal Capacity Initiative (LCI) Evaluation
Local Capacity Initiative (LCI) EvaluationMEASURE Evaluation
Β 
Development and Validation of a Reproductive Empowerment Scale
Development and Validation of a Reproductive Empowerment ScaleDevelopment and Validation of a Reproductive Empowerment Scale
Development and Validation of a Reproductive Empowerment ScaleMEASURE Evaluation
Β 
Sustaining the Impact: MEASURE Evaluation Conversation on Maternal and Child ...
Sustaining the Impact: MEASURE Evaluation Conversation on Maternal and Child ...Sustaining the Impact: MEASURE Evaluation Conversation on Maternal and Child ...
Sustaining the Impact: MEASURE Evaluation Conversation on Maternal and Child ...MEASURE Evaluation
Β 
Using Most Significant Change in a Mixed-Methods Evaluation in Uganda
Using Most Significant Change in a Mixed-Methods Evaluation in UgandaUsing Most Significant Change in a Mixed-Methods Evaluation in Uganda
Using Most Significant Change in a Mixed-Methods Evaluation in UgandaMEASURE Evaluation
Β 
Lessons Learned In Using the Most Significant Change Technique in Evaluation
Lessons Learned In Using the Most Significant Change Technique in EvaluationLessons Learned In Using the Most Significant Change Technique in Evaluation
Lessons Learned In Using the Most Significant Change Technique in EvaluationMEASURE Evaluation
Β 
Malaria Data Quality and Use in Selected Centers of Excellence in Madagascar:...
Malaria Data Quality and Use in Selected Centers of Excellence in Madagascar:...Malaria Data Quality and Use in Selected Centers of Excellence in Madagascar:...
Malaria Data Quality and Use in Selected Centers of Excellence in Madagascar:...MEASURE Evaluation
Β 
Evaluating National Malaria Programs’ Impact in Moderate- and Low-Transmissio...
Evaluating National Malaria Programs’ Impact in Moderate- and Low-Transmissio...Evaluating National Malaria Programs’ Impact in Moderate- and Low-Transmissio...
Evaluating National Malaria Programs’ Impact in Moderate- and Low-Transmissio...MEASURE Evaluation
Β 
Improved Performance of the Malaria Surveillance, Monitoring, and Evaluation ...
Improved Performance of the Malaria Surveillance, Monitoring, and Evaluation ...Improved Performance of the Malaria Surveillance, Monitoring, and Evaluation ...
Improved Performance of the Malaria Surveillance, Monitoring, and Evaluation ...MEASURE Evaluation
Β 
Lessons learned in using process tracing for evaluation
Lessons learned in using process tracing for evaluationLessons learned in using process tracing for evaluation
Lessons learned in using process tracing for evaluationMEASURE Evaluation
Β 
Use of Qualitative Comparative Analysis in the Assessment of the Actionable D...
Use of Qualitative Comparative Analysis in the Assessment of the Actionable D...Use of Qualitative Comparative Analysis in the Assessment of the Actionable D...
Use of Qualitative Comparative Analysis in the Assessment of the Actionable D...MEASURE Evaluation
Β 
Sustaining the Impact: MEASURE Evaluation Conversation on Health Informatics
Sustaining the Impact: MEASURE Evaluation Conversation on Health InformaticsSustaining the Impact: MEASURE Evaluation Conversation on Health Informatics
Sustaining the Impact: MEASURE Evaluation Conversation on Health InformaticsMEASURE Evaluation
Β 

Mais de MEASURE Evaluation (20)

Managing missing values in routinely reported data: One approach from the Dem...
Managing missing values in routinely reported data: One approach from the Dem...Managing missing values in routinely reported data: One approach from the Dem...
Managing missing values in routinely reported data: One approach from the Dem...
Β 
Use of Routine Data for Economic Evaluations
Use of Routine Data for Economic EvaluationsUse of Routine Data for Economic Evaluations
Use of Routine Data for Economic Evaluations
Β 
Routine data use in evaluation: practical guidance
Routine data use in evaluation: practical guidanceRoutine data use in evaluation: practical guidance
Routine data use in evaluation: practical guidance
Β 
Tuberculosis/HIV Mobility Study: Objectives and Background
Tuberculosis/HIV Mobility Study: Objectives and BackgroundTuberculosis/HIV Mobility Study: Objectives and Background
Tuberculosis/HIV Mobility Study: Objectives and Background
Β 
How to improve the capabilities of health information systems to address emer...
How to improve the capabilities of health information systems to address emer...How to improve the capabilities of health information systems to address emer...
How to improve the capabilities of health information systems to address emer...
Β 
LCI Evaluation Uganda Organizational Network Analysis
LCI Evaluation Uganda Organizational Network AnalysisLCI Evaluation Uganda Organizational Network Analysis
LCI Evaluation Uganda Organizational Network Analysis
Β 
Using Organizational Network Analysis to Plan and Evaluate Global Health Prog...
Using Organizational Network Analysis to Plan and Evaluate Global Health Prog...Using Organizational Network Analysis to Plan and Evaluate Global Health Prog...
Using Organizational Network Analysis to Plan and Evaluate Global Health Prog...
Β 
Understanding Referral Networks for Adolescent Girls and Young Women
Understanding Referral Networks for Adolescent Girls and Young WomenUnderstanding Referral Networks for Adolescent Girls and Young Women
Understanding Referral Networks for Adolescent Girls and Young Women
Β 
Data for Impact: Lessons Learned in Using the Ripple Effects Mapping Method
Data for Impact: Lessons Learned in Using the Ripple Effects Mapping MethodData for Impact: Lessons Learned in Using the Ripple Effects Mapping Method
Data for Impact: Lessons Learned in Using the Ripple Effects Mapping Method
Β 
Local Capacity Initiative (LCI) Evaluation
Local Capacity Initiative (LCI) EvaluationLocal Capacity Initiative (LCI) Evaluation
Local Capacity Initiative (LCI) Evaluation
Β 
Development and Validation of a Reproductive Empowerment Scale
Development and Validation of a Reproductive Empowerment ScaleDevelopment and Validation of a Reproductive Empowerment Scale
Development and Validation of a Reproductive Empowerment Scale
Β 
Sustaining the Impact: MEASURE Evaluation Conversation on Maternal and Child ...
Sustaining the Impact: MEASURE Evaluation Conversation on Maternal and Child ...Sustaining the Impact: MEASURE Evaluation Conversation on Maternal and Child ...
Sustaining the Impact: MEASURE Evaluation Conversation on Maternal and Child ...
Β 
Using Most Significant Change in a Mixed-Methods Evaluation in Uganda
Using Most Significant Change in a Mixed-Methods Evaluation in UgandaUsing Most Significant Change in a Mixed-Methods Evaluation in Uganda
Using Most Significant Change in a Mixed-Methods Evaluation in Uganda
Β 
Lessons Learned In Using the Most Significant Change Technique in Evaluation
Lessons Learned In Using the Most Significant Change Technique in EvaluationLessons Learned In Using the Most Significant Change Technique in Evaluation
Lessons Learned In Using the Most Significant Change Technique in Evaluation
Β 
Malaria Data Quality and Use in Selected Centers of Excellence in Madagascar:...
Malaria Data Quality and Use in Selected Centers of Excellence in Madagascar:...Malaria Data Quality and Use in Selected Centers of Excellence in Madagascar:...
Malaria Data Quality and Use in Selected Centers of Excellence in Madagascar:...
Β 
Evaluating National Malaria Programs’ Impact in Moderate- and Low-Transmissio...
Evaluating National Malaria Programs’ Impact in Moderate- and Low-Transmissio...Evaluating National Malaria Programs’ Impact in Moderate- and Low-Transmissio...
Evaluating National Malaria Programs’ Impact in Moderate- and Low-Transmissio...
Β 
Improved Performance of the Malaria Surveillance, Monitoring, and Evaluation ...
Improved Performance of the Malaria Surveillance, Monitoring, and Evaluation ...Improved Performance of the Malaria Surveillance, Monitoring, and Evaluation ...
Improved Performance of the Malaria Surveillance, Monitoring, and Evaluation ...
Β 
Lessons learned in using process tracing for evaluation
Lessons learned in using process tracing for evaluationLessons learned in using process tracing for evaluation
Lessons learned in using process tracing for evaluation
Β 
Use of Qualitative Comparative Analysis in the Assessment of the Actionable D...
Use of Qualitative Comparative Analysis in the Assessment of the Actionable D...Use of Qualitative Comparative Analysis in the Assessment of the Actionable D...
Use of Qualitative Comparative Analysis in the Assessment of the Actionable D...
Β 
Sustaining the Impact: MEASURE Evaluation Conversation on Health Informatics
Sustaining the Impact: MEASURE Evaluation Conversation on Health InformaticsSustaining the Impact: MEASURE Evaluation Conversation on Health Informatics
Sustaining the Impact: MEASURE Evaluation Conversation on Health Informatics
Β 

Último

Ecosystem Interactions Class Discussion Presentation in Blue Green Lined Styl...
Ecosystem Interactions Class Discussion Presentation in Blue Green Lined Styl...Ecosystem Interactions Class Discussion Presentation in Blue Green Lined Styl...
Ecosystem Interactions Class Discussion Presentation in Blue Green Lined Styl...fonyou31
Β 
The basics of sentences session 2pptx copy.pptx
The basics of sentences session 2pptx copy.pptxThe basics of sentences session 2pptx copy.pptx
The basics of sentences session 2pptx copy.pptxheathfieldcps1
Β 
APM Welcome, APM North West Network Conference, Synergies Across Sectors
APM Welcome, APM North West Network Conference, Synergies Across SectorsAPM Welcome, APM North West Network Conference, Synergies Across Sectors
APM Welcome, APM North West Network Conference, Synergies Across SectorsAssociation for Project Management
Β 
Russian Call Girls in Andheri Airport Mumbai WhatsApp 9167673311 πŸ’ž Full Nigh...
Russian Call Girls in Andheri Airport Mumbai WhatsApp  9167673311 πŸ’ž Full Nigh...Russian Call Girls in Andheri Airport Mumbai WhatsApp  9167673311 πŸ’ž Full Nigh...
Russian Call Girls in Andheri Airport Mumbai WhatsApp 9167673311 πŸ’ž Full Nigh...Pooja Nehwal
Β 
POINT- BIOCHEMISTRY SEM 2 ENZYMES UNIT 5.pptx
POINT- BIOCHEMISTRY SEM 2 ENZYMES UNIT 5.pptxPOINT- BIOCHEMISTRY SEM 2 ENZYMES UNIT 5.pptx
POINT- BIOCHEMISTRY SEM 2 ENZYMES UNIT 5.pptxSayali Powar
Β 
Sports & Fitness Value Added Course FY..
Sports & Fitness Value Added Course FY..Sports & Fitness Value Added Course FY..
Sports & Fitness Value Added Course FY..Disha Kariya
Β 
Kisan Call Centre - To harness potential of ICT in Agriculture by answer farm...
Kisan Call Centre - To harness potential of ICT in Agriculture by answer farm...Kisan Call Centre - To harness potential of ICT in Agriculture by answer farm...
Kisan Call Centre - To harness potential of ICT in Agriculture by answer farm...Krashi Coaching
Β 
Presentation by Andreas Schleicher Tackling the School Absenteeism Crisis 30 ...
Presentation by Andreas Schleicher Tackling the School Absenteeism Crisis 30 ...Presentation by Andreas Schleicher Tackling the School Absenteeism Crisis 30 ...
Presentation by Andreas Schleicher Tackling the School Absenteeism Crisis 30 ...EduSkills OECD
Β 
Separation of Lanthanides/ Lanthanides and Actinides
Separation of Lanthanides/ Lanthanides and ActinidesSeparation of Lanthanides/ Lanthanides and Actinides
Separation of Lanthanides/ Lanthanides and ActinidesFatimaKhan178732
Β 
Accessible design: Minimum effort, maximum impact
Accessible design: Minimum effort, maximum impactAccessible design: Minimum effort, maximum impact
Accessible design: Minimum effort, maximum impactdawncurless
Β 
Organic Name Reactions for the students and aspirants of Chemistry12th.pptx
Organic Name Reactions  for the students and aspirants of Chemistry12th.pptxOrganic Name Reactions  for the students and aspirants of Chemistry12th.pptx
Organic Name Reactions for the students and aspirants of Chemistry12th.pptxVS Mahajan Coaching Centre
Β 
Advanced Views - Calendar View in Odoo 17
Advanced Views - Calendar View in Odoo 17Advanced Views - Calendar View in Odoo 17
Advanced Views - Calendar View in Odoo 17Celine George
Β 
Measures of Central Tendency: Mean, Median and Mode
Measures of Central Tendency: Mean, Median and ModeMeasures of Central Tendency: Mean, Median and Mode
Measures of Central Tendency: Mean, Median and ModeThiyagu K
Β 
Arihant handbook biology for class 11 .pdf
Arihant handbook biology for class 11 .pdfArihant handbook biology for class 11 .pdf
Arihant handbook biology for class 11 .pdfchloefrazer622
Β 
Mastering the Unannounced Regulatory Inspection
Mastering the Unannounced Regulatory InspectionMastering the Unannounced Regulatory Inspection
Mastering the Unannounced Regulatory InspectionSafetyChain Software
Β 
JAPAN: ORGANISATION OF PMDA, PHARMACEUTICAL LAWS & REGULATIONS, TYPES OF REGI...
JAPAN: ORGANISATION OF PMDA, PHARMACEUTICAL LAWS & REGULATIONS, TYPES OF REGI...JAPAN: ORGANISATION OF PMDA, PHARMACEUTICAL LAWS & REGULATIONS, TYPES OF REGI...
JAPAN: ORGANISATION OF PMDA, PHARMACEUTICAL LAWS & REGULATIONS, TYPES OF REGI...anjaliyadav012327
Β 
Activity 01 - Artificial Culture (1).pdf
Activity 01 - Artificial Culture (1).pdfActivity 01 - Artificial Culture (1).pdf
Activity 01 - Artificial Culture (1).pdfciinovamais
Β 

Último (20)

Ecosystem Interactions Class Discussion Presentation in Blue Green Lined Styl...
Ecosystem Interactions Class Discussion Presentation in Blue Green Lined Styl...Ecosystem Interactions Class Discussion Presentation in Blue Green Lined Styl...
Ecosystem Interactions Class Discussion Presentation in Blue Green Lined Styl...
Β 
CΓ³digo Creativo y Arte de Software | Unidad 1
CΓ³digo Creativo y Arte de Software | Unidad 1CΓ³digo Creativo y Arte de Software | Unidad 1
CΓ³digo Creativo y Arte de Software | Unidad 1
Β 
The basics of sentences session 2pptx copy.pptx
The basics of sentences session 2pptx copy.pptxThe basics of sentences session 2pptx copy.pptx
The basics of sentences session 2pptx copy.pptx
Β 
APM Welcome, APM North West Network Conference, Synergies Across Sectors
APM Welcome, APM North West Network Conference, Synergies Across SectorsAPM Welcome, APM North West Network Conference, Synergies Across Sectors
APM Welcome, APM North West Network Conference, Synergies Across Sectors
Β 
Russian Call Girls in Andheri Airport Mumbai WhatsApp 9167673311 πŸ’ž Full Nigh...
Russian Call Girls in Andheri Airport Mumbai WhatsApp  9167673311 πŸ’ž Full Nigh...Russian Call Girls in Andheri Airport Mumbai WhatsApp  9167673311 πŸ’ž Full Nigh...
Russian Call Girls in Andheri Airport Mumbai WhatsApp 9167673311 πŸ’ž Full Nigh...
Β 
POINT- BIOCHEMISTRY SEM 2 ENZYMES UNIT 5.pptx
POINT- BIOCHEMISTRY SEM 2 ENZYMES UNIT 5.pptxPOINT- BIOCHEMISTRY SEM 2 ENZYMES UNIT 5.pptx
POINT- BIOCHEMISTRY SEM 2 ENZYMES UNIT 5.pptx
Β 
Sports & Fitness Value Added Course FY..
Sports & Fitness Value Added Course FY..Sports & Fitness Value Added Course FY..
Sports & Fitness Value Added Course FY..
Β 
Kisan Call Centre - To harness potential of ICT in Agriculture by answer farm...
Kisan Call Centre - To harness potential of ICT in Agriculture by answer farm...Kisan Call Centre - To harness potential of ICT in Agriculture by answer farm...
Kisan Call Centre - To harness potential of ICT in Agriculture by answer farm...
Β 
Mattingly "AI & Prompt Design: Structured Data, Assistants, & RAG"
Mattingly "AI & Prompt Design: Structured Data, Assistants, & RAG"Mattingly "AI & Prompt Design: Structured Data, Assistants, & RAG"
Mattingly "AI & Prompt Design: Structured Data, Assistants, & RAG"
Β 
Presentation by Andreas Schleicher Tackling the School Absenteeism Crisis 30 ...
Presentation by Andreas Schleicher Tackling the School Absenteeism Crisis 30 ...Presentation by Andreas Schleicher Tackling the School Absenteeism Crisis 30 ...
Presentation by Andreas Schleicher Tackling the School Absenteeism Crisis 30 ...
Β 
Separation of Lanthanides/ Lanthanides and Actinides
Separation of Lanthanides/ Lanthanides and ActinidesSeparation of Lanthanides/ Lanthanides and Actinides
Separation of Lanthanides/ Lanthanides and Actinides
Β 
Accessible design: Minimum effort, maximum impact
Accessible design: Minimum effort, maximum impactAccessible design: Minimum effort, maximum impact
Accessible design: Minimum effort, maximum impact
Β 
Organic Name Reactions for the students and aspirants of Chemistry12th.pptx
Organic Name Reactions  for the students and aspirants of Chemistry12th.pptxOrganic Name Reactions  for the students and aspirants of Chemistry12th.pptx
Organic Name Reactions for the students and aspirants of Chemistry12th.pptx
Β 
Mattingly "AI & Prompt Design: The Basics of Prompt Design"
Mattingly "AI & Prompt Design: The Basics of Prompt Design"Mattingly "AI & Prompt Design: The Basics of Prompt Design"
Mattingly "AI & Prompt Design: The Basics of Prompt Design"
Β 
Advanced Views - Calendar View in Odoo 17
Advanced Views - Calendar View in Odoo 17Advanced Views - Calendar View in Odoo 17
Advanced Views - Calendar View in Odoo 17
Β 
Measures of Central Tendency: Mean, Median and Mode
Measures of Central Tendency: Mean, Median and ModeMeasures of Central Tendency: Mean, Median and Mode
Measures of Central Tendency: Mean, Median and Mode
Β 
Arihant handbook biology for class 11 .pdf
Arihant handbook biology for class 11 .pdfArihant handbook biology for class 11 .pdf
Arihant handbook biology for class 11 .pdf
Β 
Mastering the Unannounced Regulatory Inspection
Mastering the Unannounced Regulatory InspectionMastering the Unannounced Regulatory Inspection
Mastering the Unannounced Regulatory Inspection
Β 
JAPAN: ORGANISATION OF PMDA, PHARMACEUTICAL LAWS & REGULATIONS, TYPES OF REGI...
JAPAN: ORGANISATION OF PMDA, PHARMACEUTICAL LAWS & REGULATIONS, TYPES OF REGI...JAPAN: ORGANISATION OF PMDA, PHARMACEUTICAL LAWS & REGULATIONS, TYPES OF REGI...
JAPAN: ORGANISATION OF PMDA, PHARMACEUTICAL LAWS & REGULATIONS, TYPES OF REGI...
Β 
Activity 01 - Artificial Culture (1).pdf
Activity 01 - Artificial Culture (1).pdfActivity 01 - Artificial Culture (1).pdf
Activity 01 - Artificial Culture (1).pdf
Β 

Selection on Observables

  • 1. Peter M. Lance, PhD MEASURE Evaluation University of North Carolina at Chapel Hill September 13 and 15, 2016 Selection on Observables
  • 2. Global, five-year, $180M cooperative agreement Strategic objective: To strengthen health information systems – the capacity to gather, interpret, and use data – so countries can make better decisions and sustain good health outcomes over time. Project overview
  • 3. Improved country capacity to manage health information systems, resources, and staff Strengthened collection, analysis, and use of routine health data Methods, tools, and approaches improved and applied to address health information challenges and gaps Increased capacity for rigorous evaluation Phase IV Results Framework
  • 4. Global footprint (more than 25 countries)
  • 5. β€’ The program impact evaluation challenge β€’ Randomization β€’ Selection on observables β€’ Within estimators β€’ Instrumental variables
  • 6. β€’ The program impact evaluation challenge β€’ Randomization β€’ Selection on observables β€’ Within estimators β€’ Instrumental variables
  • 7.
  • 8.
  • 10. X Y P π‘Œ = 𝑃 βˆ™ π‘Œ1 + 1 βˆ’ 𝑃 βˆ™ π‘Œ0
  • 11. X Y P
  • 12. X Y P
  • 13. 𝐸 π‘Œ1 βˆ’ π‘Œ0 = 𝐸 π‘Œ1 βˆ’ 𝐸 π‘Œ0 Average Y across sample of participants βˆ’ Average Y across sample of nonβˆ’participants
  • 14. 𝐸 π‘Œ1 βˆ’ π‘Œ0 = 𝐸 π‘Œ1 βˆ’ 𝐸 π‘Œ0 Average Y across sample of participants βˆ’ Average Y across sample of nonβˆ’participants
  • 15. 𝐸 π‘Œ1 βˆ’ π‘Œ0 = 𝐸 π‘Œ1 βˆ’ 𝐸 π‘Œ0 Average Y across sample of participants βˆ’ Average Y across sample of nonβˆ’participants
  • 16. 𝐸 π‘Œ1 βˆ’ π‘Œ0 = 𝐸 π‘Œ1 βˆ’ 𝐸 π‘Œ0 Average Y across sample of participants βˆ’ Average Y across sample of nonβˆ’participants
  • 17. 𝐸 π‘Œ1 βˆ’ π‘Œ0 = 𝐸 π‘Œ1 βˆ’ 𝐸 π‘Œ0 Average Y across sample of participants βˆ’ Average Y across sample of nonβˆ’participants
  • 18. 𝐸 π‘Œ1 βˆ’ π‘Œ0 = 𝐸 π‘Œ1 βˆ’ 𝐸 π‘Œ0 Average Y across sample of participants βˆ’ Average Y across sample of nonβˆ’participants
  • 19. 𝐸 π‘Œ1 βˆ’ π‘Œ0 = 𝐸 π‘Œ1 βˆ’ 𝐸 π‘Œ0 Average Y across sample of participants βˆ’ Average Y across sample of nonβˆ’participants
  • 20. 𝐸 π‘Œ1 𝑃 = 1 = 𝐸 π‘Œ1 𝑃 = 0 = 𝐸 π‘Œ1
  • 21. X Y P
  • 22. 𝐸 𝑋 𝑃 = 1 β‰  𝐸 𝑋 𝑃 = 0
  • 23. X Y P
  • 24. X Y P
  • 25. 𝐸 𝑋 𝑃 = 1 β‰  𝐸 𝑋 𝑃 = 0
  • 26. 𝐸 𝑋 𝑃 = 1 β‰  𝐸 𝑋 𝑃 = 0 𝐸 π‘Œ1 𝑃 = 1 β‰  𝐸 π‘Œ1 𝑃 = 0
  • 27. 𝐸 𝑋 𝑃 = 1 β‰  𝐸 𝑋 𝑃 = 0 𝐸 π‘Œ0 𝑃 = 1 β‰  𝐸 π‘Œ0 𝑃 = 0
  • 28. 𝐸 π‘Œ1 βˆ’ π‘Œ0 = 𝐸 π‘Œ1 βˆ’ 𝐸 π‘Œ0 Average Y across sample of participants βˆ’ Average Y across sample of nonβˆ’participants
  • 29. 𝐸 π‘Œ1 𝑃 = 1 β‰  𝐸 π‘Œ1 𝑃 = 0 𝑋 = π‘₯βˆ— 𝐸 π‘Œ1 𝑃 = 1, 𝑋 = π‘₯βˆ— = 𝐸 π‘Œ1 𝑃 = 0, 𝑋 = π‘₯βˆ— 𝐸 π‘Œ0 𝑃 = 1, 𝑋 = π‘₯βˆ— = 𝐸 π‘Œ0 𝑃 = 0, 𝑋 = π‘₯βˆ—
  • 30. 𝐸 π‘Œ1 𝑃 = 1 β‰  𝐸 π‘Œ1 𝑃 = 0 𝑋 = π‘₯βˆ— 𝐸 π‘Œ1 𝑃 = 1, 𝑋 = π‘₯βˆ— = 𝐸 π‘Œ1 𝑃 = 0, 𝑋 = π‘₯βˆ— 𝐸 π‘Œ0 𝑃 = 1, 𝑋 = π‘₯βˆ— = 𝐸 π‘Œ0 𝑃 = 0, 𝑋 = π‘₯βˆ—
  • 31. X Y P
  • 32. 𝐸 π‘Œ1 𝑃 = 1 β‰  𝐸 π‘Œ1 𝑃 = 0 𝑋 = π‘₯βˆ— 𝐸 π‘Œ1 𝑃 = 1, 𝑋 = π‘₯βˆ— = 𝐸 π‘Œ1 𝑃 = 0, 𝑋 = π‘₯βˆ— 𝐸 π‘Œ0 𝑃 = 1, 𝑋 = π‘₯βˆ— = 𝐸 π‘Œ0 𝑃 = 0, 𝑋 = π‘₯βˆ—
  • 33. 𝐸 π‘Œ1 𝑃 = 1 β‰  𝐸 π‘Œ1 𝑃 = 0 𝑋 = π‘₯βˆ— 𝐸 π‘Œ1 𝑃 = 1, 𝑋 = π‘₯βˆ— = 𝐸 π‘Œ1 𝑃 = 0, 𝑋 = π‘₯βˆ— 𝐸 π‘Œ0 𝑃 = 1, 𝑋 = π‘₯βˆ— = 𝐸 π‘Œ0 𝑃 = 0, 𝑋 = π‘₯βˆ—
  • 34. 𝐸 π‘Œ1 𝑃 = 1 β‰  𝐸 π‘Œ1 𝑃 = 0 𝑋 = π‘₯βˆ— 𝐸 π‘Œ1 𝑃 = 1, 𝑋 = π‘₯βˆ— = 𝐸 π‘Œ1 𝑃 = 0, 𝑋 = π‘₯βˆ— 𝐸 π‘Œ0 𝑃 = 1, 𝑋 = π‘₯βˆ— = 𝐸 π‘Œ0 𝑃 = 0, 𝑋 = π‘₯βˆ—
  • 35. 𝐸 π‘Œ1 βˆ’ π‘Œ0 = 𝐸 π‘Œ1 βˆ’ 𝐸 π‘Œ0 Average Y across sample of participants βˆ’ Average Y across sample of nonβˆ’participants
  • 36. 𝐸 π‘Œ1 βˆ’ π‘Œ0 |𝑋 = π‘₯βˆ— = 𝐸 π‘Œ1 |𝑋 = π‘₯βˆ— βˆ’ 𝐸 π‘Œ0 |𝑋 = π‘₯βˆ— Average Y across sample of participants for whom 𝑋 = π‘₯βˆ— βˆ’ Average Y across sample of participants of nonβˆ’participants 𝑋 = π‘₯βˆ—
  • 37. 𝐸 π‘Œ1 βˆ’ π‘Œ0 = 𝐸 π‘Œ1 βˆ’ 𝐸 π‘Œ0 Population Participation Rate Poor .4 .7 Middle .5 .3 Rich .1 .1
  • 38. 𝐸 π‘Œ1 βˆ’ π‘Œ0 = 𝐸 π‘Œ1 βˆ’ 𝐸 π‘Œ0 Population Participation Rate Poor .4 .7 Middle .5 .3 Rich .1 .1
  • 39. 𝐸 π‘Œ1 βˆ’ π‘Œ0 = 𝐸 π‘Œ1 βˆ’ 𝐸 π‘Œ0 Population Participation Rate Poor .4 .7 Middle .5 .3 Rich .1 .1
  • 40. 𝐸 π‘Œ1 βˆ’ π‘Œ0 = 𝐸 π‘Œ1 βˆ’ 𝐸 π‘Œ0 Participants (𝑷 = 𝟏) Non- participants (𝑷 = 𝟎) Poor .64 .21 Middle .34 .63 Rich .02 .16
  • 41. 𝐸 π‘Œ1 βˆ’ π‘Œ0 = 𝐸 π‘Œ1 βˆ’ 𝐸 π‘Œ0 Participants (𝑷 = 𝟏) Non- participants (𝑷 = 𝟎) Poor .64 .21 Middle .34 .63 Rich .02 .16
  • 42. 𝐸 π‘Œ1 βˆ’ π‘Œ0 = 𝐸 π‘Œ1 βˆ’ 𝐸 π‘Œ0 Participants (𝑷 = 𝟏) Non- participants (𝑷 = 𝟎) Poor .64 .21 Middle .34 .63 Rich .02 .16
  • 43. 𝐸 π‘Œ1 βˆ’ π‘Œ0 = 𝐸 π‘Œ1 βˆ’ 𝐸 π‘Œ0 Participants (𝑷 = 𝟏) Weight Poor .64 .63 Middle .34 1.47 Rich .02 4.4
  • 44. 𝐸 π‘Œ1 βˆ’ π‘Œ0 = 𝐸 π‘Œ1 βˆ’ 𝐸 π‘Œ0 Participants (𝑷 = 𝟏) Weight Poor .64 .63 Middle .34 1.47 Rich .02 4.4
  • 45. 𝐸 π‘Œ1 βˆ’ π‘Œ0 = 𝐸 π‘Œ1 βˆ’ 𝐸 π‘Œ0 Participants (𝑷 = 𝟏) Weight Poor .64 .63 Middle .34 1.47 Rich .02 4.4
  • 46. 𝐸 π‘Œ1 βˆ’ π‘Œ0 = 𝐸 π‘Œ1 βˆ’ 𝐸 π‘Œ0 Participants (𝑷 = 𝟏) Weight Poor .64 .63 Middle .34 1.47 Rich .02 4.5
  • 47. 𝐸 π‘Œ1 βˆ’ π‘Œ0 = 𝐸 π‘Œ1 βˆ’ 𝐸 π‘Œ0 Average of Y across a sample of participants = 𝑖=1 𝑛 𝑃 𝑀𝑖 βˆ— π‘Œπ‘– 𝑛 𝑃 π’˜π’Š=.63 if individual i is poor π’˜π’Š=1.47 if individual i is middle class π’˜π’Š=4.4 if individual i is rich
  • 48. 𝐸 π‘Œ1 βˆ’ π‘Œ0 = 𝐸 π‘Œ1 βˆ’ 𝐸 π‘Œ0 Average of Y across a sample of participants = 𝑖=1 𝑛 𝑃 π‘Œπ‘– 𝑛 𝑃 π’˜π’Š=.63 if individual i is poor π’˜π’Š=1.47 if individual i is middle class π’˜π’Š=4.4 if individual i is rich
  • 49. 𝐸 π‘Œ1 βˆ’ π‘Œ0 = 𝐸 π‘Œ1 βˆ’ 𝐸 π‘Œ0 Average of Y across a sample of participants = 𝑖=1 𝑛 𝑃 𝑀𝑖 βˆ™ π‘Œπ‘– 𝑖=1 𝑛 𝑃 𝑀𝑖 π’˜π’Š=.63 if individual i is poor π’˜π’Š=1.47 if individual i is middle class π’˜π’Š=4.4 if individual i is rich
  • 50. 𝐸 π‘Œ1 βˆ’ π‘Œ0 = 𝐸 π‘Œ1 βˆ’ 𝐸 π‘Œ0 Average of Y across a sample of participants = 𝑖=1 𝑛 𝑃 π’˜π’Š βˆ™ π‘Œπ‘– 𝑖=1 𝑛 𝑃 𝑀𝑖 π’˜π’Š=.63 if individual i is poor π’˜π’Š=1.47 if individual i is middle class π’˜π’Š=4.4 if individual i is rich
  • 51. 𝐸 π‘Œ1 βˆ’ π‘Œ0 = 𝐸 π‘Œ1 βˆ’ 𝐸 π‘Œ0 Participants (𝑷 = 𝟏) Weight Poor .64 .63 Middle .34 1.47 Rich .02 4.5
  • 52. 𝐸 π‘Œ1 βˆ’ π‘Œ0 = 𝐸 π‘Œ1 βˆ’ 𝐸 π‘Œ0 Average of Y across a sample of participants = 𝑖=1 𝑛 𝑃 𝑀𝑖 βˆ™ π‘Œπ‘– 𝑖=1 𝑛 𝑃 𝑀𝑖 π’˜π’Š=.63 if individual i is poor π’˜π’Š=1.47 if individual i is middle class π’˜π’Š=4.5 if individual i is rich
  • 53. 𝐸 π‘Œ1 βˆ’ π‘Œ0 = 𝐸 π‘Œ1 βˆ’ 𝐸 π‘Œ0 . 4 βˆ— Average Y across sample of participants who are poor + .5 βˆ— Average Y across sample of nonβˆ’participants who are middle class +.1 βˆ— Average Y across sample of participants who are rich
  • 54. 𝐸 π‘Œ1 βˆ’ π‘Œ0 = 𝐸 π‘Œ1 βˆ’ 𝐸 π‘Œ0 Average of Y across a sample of participants = 𝑖=1 𝑛 𝑃 𝑀𝑖 βˆ™ π‘Œπ‘– 𝑖=1 𝑛 𝑃 𝑀𝑖 π’˜π’Š=.63 if individual i is poor π’˜π’Š=1.47 if individual i is middle class π’˜π’Š=4.5 if individual i is rich
  • 55.
  • 56. π‘Œ0 = 𝛽0 + πœ– π‘Œ1 = 𝛽0 + 𝛽1 + πœ–
  • 57. π‘Œ1 βˆ’ π‘Œ0 = 𝛽0 + 𝛽1 + πœ– βˆ’ 𝛽0 + πœ– = 𝛽0 + 𝛽1 + πœ– βˆ’ 𝛽0 βˆ’ πœ– = 𝛽1
  • 58. π‘Œ1 βˆ’ π‘Œ0 = 𝛽0 + 𝛽1 + πœ– βˆ’ 𝛽0 + πœ– = 𝛽0 + 𝛽1 + πœ– βˆ’ 𝛽0 βˆ’ πœ– = 𝛽1
  • 59. π‘Œ1 βˆ’ π‘Œ0 = 𝛽0 + 𝛽1 + πœ– βˆ’ 𝛽0 + πœ– = 𝛽0 + 𝛽1 + πœ– βˆ’ 𝛽0 βˆ’ πœ– = 𝛽1
  • 60. π‘Œ1 βˆ’ π‘Œ0 = 𝛽0 + 𝛽1 + πœ– βˆ’ 𝛽0 + πœ– = 𝛽0 + 𝛽1 + πœ– βˆ’ 𝛽0 βˆ’ πœ– = 𝛽1
  • 61. π‘Œ = 𝑃 βˆ™ π‘Œ1 + 1 βˆ’ 𝑃 βˆ™ π‘Œ0 = 𝑃 βˆ— 𝛽0 + 𝛽1 + πœ– + 1 βˆ’ 𝑃 βˆ— 𝛽0 + πœ– = 𝑃 βˆ— 𝛽0 + 𝑃 βˆ— 𝛽1 + 𝑃 βˆ— πœ– +𝛽0 + πœ– βˆ’ 𝑃 βˆ— 𝛽0 βˆ’ 𝑃 βˆ— πœ– = 𝛽0 + 𝑃 βˆ— 𝛽1 + πœ–
  • 62. π‘Œ = 𝑃 βˆ™ π‘Œ1 + 1 βˆ’ 𝑃 βˆ™ π‘Œ0 = 𝑃 βˆ™ 𝛽0 + 𝛽1 + πœ– + 1 βˆ’ 𝑃 βˆ™ 𝛽0 + πœ– = 𝑃 βˆ— 𝛽0 + 𝑃 βˆ— 𝛽1 + 𝑃 βˆ— πœ– +𝛽0 + πœ– βˆ’ 𝑃 βˆ— 𝛽0 βˆ’ 𝑃 βˆ— πœ– = 𝛽0 + 𝑃 βˆ— 𝛽1 + πœ–
  • 63. π‘Œ = 𝑃 βˆ™ π‘Œ1 + 1 βˆ’ 𝑃 βˆ™ π‘Œ0 = 𝑃 βˆ™ 𝛽0 + 𝛽1 + πœ– + 1 βˆ’ 𝑃 βˆ™ 𝛽0 + πœ– = 𝑃 βˆ™ 𝛽0 + 𝑃 βˆ™ 𝛽1 + 𝑃 βˆ™ πœ– +𝛽0 + πœ– βˆ’ 𝑃 βˆ™ 𝛽0 βˆ’ 𝑃 βˆ™ πœ– = 𝛽0 + 𝑃 βˆ— 𝛽1 + πœ–
  • 64. π‘Œ = 𝑃 βˆ™ π‘Œ1 + 1 βˆ’ 𝑃 βˆ™ π‘Œ0 = 𝑃 βˆ™ 𝛽0 + 𝛽1 + πœ– + 1 βˆ’ 𝑃 βˆ™ 𝛽0 + πœ– = 𝑃 βˆ™ 𝛽0 + 𝑃 βˆ™ 𝛽1 + 𝑃 βˆ™ πœ– +𝛽0 + πœ– βˆ’ 𝑃 βˆ™ 𝛽0 βˆ’ 𝑃 βˆ™ πœ– = 𝛽0 + 𝑃 βˆ— 𝛽1 + πœ–
  • 65. π‘Œ = 𝑃 βˆ™ π‘Œ1 + 1 βˆ’ 𝑃 βˆ™ π‘Œ0 = 𝑃 βˆ™ 𝛽0 + 𝛽1 + πœ– + 1 βˆ’ 𝑃 βˆ™ 𝛽0 + πœ– = 𝑃 βˆ™ 𝛽0 + 𝑃 βˆ™ 𝛽1 + 𝑃 βˆ™ πœ– +𝛽0 + πœ– βˆ’ 𝑃 βˆ™ 𝛽0 βˆ’ 𝑃 βˆ™ πœ– = 𝛽0 + 𝑃 βˆ— 𝛽1 + πœ–
  • 66. π‘Œ = 𝑃 βˆ™ π‘Œ1 + 1 βˆ’ 𝑃 βˆ™ π‘Œ0 = 𝑃 βˆ™ 𝛽0 + 𝛽1 + πœ– + 1 βˆ’ 𝑃 βˆ™ 𝛽0 + πœ– = 𝑃 βˆ™ 𝛽0 + 𝑃 βˆ™ 𝛽1 + 𝑃 βˆ™ πœ– +𝛽0 + πœ– βˆ’ 𝑃 βˆ™ 𝛽0 βˆ’ 𝑃 βˆ™ πœ– = 𝛽0 + 𝑃 βˆ™ 𝛽1 + πœ–
  • 67. π‘Œ = 𝛽0 + 𝑃 βˆ™ 𝛽1 + πœ–
  • 68. π‘Œ = 𝛽0 + 𝛽1 βˆ™ 𝑃 + πœ–
  • 69. π‘Œ0 = 𝛽0 + πœ– π‘Œ1 = 𝛽0 + 𝛽1 + πœ– π‘Œ = 𝑃 βˆ™ π‘Œ1 + 1 βˆ’ 𝑃 βˆ™ π‘Œ0
  • 70. π‘Œ0 = 𝛽0 + πœ– π‘Œ1 = 𝛽0 + 𝛽1 + πœ– π‘Œ = 𝑃 βˆ™ π‘Œ1 + 1 βˆ’ 𝑃 βˆ™ π‘Œ0
  • 71. π‘Œ0 = 𝛽0 + πœ– π‘Œ1 = 𝛽0 + 𝛽1 + πœ– π‘Œ = 𝑃 βˆ™ π‘Œ1 + 1 βˆ’ 𝑃 βˆ™ π‘Œ0
  • 72. π‘Œ = 𝛽0 + 𝛽1 βˆ™ 𝑃 + πœ–
  • 73. π‘Œ0 = 𝛽0 + πœ– π‘Œ1 = 𝛽0 + 𝛽1 + πœ– π‘Œ = 𝑃 βˆ™ π‘Œ1 + 1 βˆ’ 𝑃 βˆ™ π‘Œ0
  • 74. π‘Œ = 𝛽0 + 𝛽1 βˆ™ 𝑃 + πœ–
  • 75. π‘Œπ‘– = 𝛽0 + 𝛽1 βˆ™ 𝑃𝑖 + πœ–π‘– 𝑖 = 1, … , 𝑛
  • 76. πœ–π‘– = π‘Œπ‘– βˆ’ 𝛽0 βˆ’ 𝛽1 βˆ™ 𝑃𝑖 π‘šπ‘–π‘› 𝛽0, 𝛽1 𝑖=1 𝑛 πœ€π‘– 2 = π‘šπ‘–π‘› 𝛽0, 𝛽1 𝑖=1 𝑛 π‘Œπ‘– βˆ’ 𝛽0 βˆ’ 𝛽1 βˆ™ 𝑃𝑖 2 πœ–π‘–
  • 77. π‘Œ = 𝛽0 + 𝛽1 βˆ™ 𝑃 + πœ–
  • 78. π‘Œ = 𝛽0 + 𝛽1 βˆ™ 𝑃 + πœ–
  • 79. πœ–π‘– = π‘Œπ‘– βˆ’ 𝛽0 βˆ’ 𝛽1 βˆ™ 𝑃𝑖 π‘šπ‘–π‘› 𝛽0, 𝛽1 𝑖=1 𝑛 πœ€π‘– 2 = π‘šπ‘–π‘› 𝛽0, 𝛽1 𝑖=1 𝑛 π‘Œπ‘– βˆ’ 𝛽0 βˆ’ 𝛽1 βˆ™ 𝑃𝑖 2 πœ–π‘–
  • 80. 𝛽1 = 𝑖=1 𝑛 𝑃𝑖 βˆ’ 𝑃 βˆ™ π‘Œπ‘– 𝑖=1 𝑛 𝑃𝑖 βˆ’ 𝑃 2 where 𝑃 = 𝑖=1 𝑛 𝑃𝑖 𝑛
  • 81. 𝐸 𝛽1 = 𝛽1
  • 82. An expected value for a random variable is the average value from a large number of repetitions of the experiment that random variable represents An expected value is the true average of a random variable across a population Expected value
  • 83. An expected value for a random variable is the average value from a large number of repetitions of the experiment that random variable represents An expected value is the true average of a random variable across a population Expected value
  • 84. An expected value is the true average of a random variable across a population 𝐸 𝑋 = some true value Expected value
  • 85. π‘Œ = 𝛽0 + 𝛽1 βˆ™ 𝑃 + πœ–
  • 86. 𝐸 𝑐 = 𝑐 𝐸 𝑐 βˆ™ π‘Š = 𝑐 βˆ™ 𝐸 π‘Š 𝐸 π‘Š + 𝑍 = 𝐸 π‘Š + 𝐸 𝑍 𝐸 π‘Š βˆ’ 𝑍 = 𝐸 π‘Š βˆ’ 𝐸 𝑍 𝐸 π‘Ž βˆ™ π‘Š Β± 𝑏 βˆ™ 𝑍 = π‘Ž βˆ™ 𝐸 π‘Š Β± 𝑏 βˆ™ 𝐸 𝑍 Expectations: Properties
  • 87. 𝑬 𝒄 = 𝒄 𝐸 𝑐 βˆ™ π‘Š = 𝑐 βˆ™ 𝐸 π‘Š 𝐸 π‘Š + 𝑍 = 𝐸 π‘Š + 𝐸 𝑍 𝐸 π‘Š βˆ’ 𝑍 = 𝐸 π‘Š βˆ’ 𝐸 𝑍 𝐸 π‘Ž βˆ™ π‘Š Β± 𝑏 βˆ™ 𝑍 = π‘Ž βˆ™ 𝐸 π‘Š Β± 𝑏 βˆ™ 𝐸 𝑍 Expectations: Properties
  • 88. 𝐸 𝑐 = 𝑐 𝑬 𝒄 βˆ™ 𝑾 = 𝒄 βˆ™ 𝑬 𝑾 𝐸 π‘Š + 𝑍 = 𝐸 π‘Š + 𝐸 𝑍 𝐸 π‘Š βˆ’ 𝑍 = 𝐸 π‘Š βˆ’ 𝐸 𝑍 𝐸 π‘Ž βˆ™ π‘Š Β± 𝑏 βˆ™ 𝑍 = π‘Ž βˆ™ 𝐸 π‘Š Β± 𝑏 βˆ™ 𝐸 𝑍 Expectations: Properties
  • 89. 𝐸 𝑐 = 𝑐 𝐸 𝑐 βˆ™ π‘Š = 𝑐 βˆ™ 𝐸 π‘Š 𝑬 𝑾 + 𝒁 = 𝑬 𝑾 + 𝑬 𝒁 𝐸 π‘Š βˆ’ 𝑍 = 𝐸 π‘Š βˆ’ 𝐸 𝑍 𝐸 π‘Ž βˆ™ π‘Š Β± 𝑏 βˆ™ 𝑍 = π‘Ž βˆ™ 𝐸 π‘Š Β± 𝑏 βˆ™ 𝐸 𝑍 Expectations: Properties
  • 90. 𝐸 𝑐 = 𝑐 𝐸 𝑐 βˆ™ π‘Š = 𝑐 βˆ™ 𝐸 π‘Š 𝐸 π‘Š + 𝑍 = 𝐸 π‘Š + 𝐸 𝑍 𝑬 𝑾 βˆ’ 𝒁 = 𝑬 𝑾 βˆ’ 𝑬 𝒁 𝐸 π‘Ž βˆ™ π‘Š Β± 𝑏 βˆ™ 𝑍 = π‘Ž βˆ™ 𝐸 π‘Š Β± 𝑏 βˆ™ 𝐸 𝑍 Expectations: Properties
  • 91. 𝐸 𝑐 = 𝑐 𝐸 𝑐 βˆ™ π‘Š = 𝑐 βˆ™ 𝐸 π‘Š 𝐸 π‘Š + 𝑍 = 𝐸 π‘Š + 𝐸 𝑍 𝐸 π‘Š βˆ’ 𝑍 = 𝐸 π‘Š βˆ’ 𝐸 𝑍 𝑬 𝒂 βˆ™ 𝑾 Β± 𝒃 βˆ™ 𝒁 = 𝒂 βˆ™ 𝑬 𝑾 Β± 𝒃 βˆ™ 𝑬 𝒁 Expectations: Properties
  • 92. 𝐸 π‘Š βˆ™ 𝑍 β‰  𝐸 π‘Š βˆ™ 𝐸 𝑍 𝐸 π‘Š 𝑍 β‰  𝐸 π‘Š 𝐸 𝑍 𝐸 𝑓 π‘Š β‰  𝑓 𝐸 π‘Š Expectations: Properties
  • 93. 𝑬 𝑾 βˆ™ 𝒁 β‰  𝑬 𝑾 βˆ™ 𝑬 𝒁 𝐸 π‘Š 𝑍 β‰  𝐸 π‘Š 𝐸 𝑍 𝐸 𝑓 π‘Š β‰  𝑓 𝐸 π‘Š Expectations: Properties
  • 94. 𝐸 π‘Š βˆ™ 𝑍 β‰  𝐸 π‘Š βˆ™ 𝐸 𝑍 𝑬 𝑾 𝒁 β‰  𝑬 𝑾 𝑬 𝒁 𝐸 𝑓 π‘Š β‰  𝑓 𝐸 π‘Š Expectations: Properties
  • 95. 𝐸 π‘Š βˆ™ 𝑍 β‰  𝐸 π‘Š βˆ™ 𝐸 𝑍 𝐸 π‘Š 𝑍 β‰  𝐸 π‘Š 𝐸 𝑍 𝑬 𝒇 𝑾 β‰  𝒇 𝑬 𝑾 Expectations: Properties
  • 96. π‘Œπ‘– = 𝛽0 + 𝛽1 βˆ™ 𝑃𝑖 + πœ–π‘– 𝛽1 = 𝑖=1 𝑛 𝑃𝑖 βˆ’ 𝑃 βˆ™ π‘Œπ‘– 𝑖=1 𝑛 𝑃𝑖 βˆ’ 𝑃 2
  • 97. π‘Œπ‘– = 𝛽0 + 𝛽1 βˆ™ 𝑃𝑖 + πœ–π‘– 𝛽1 = 𝑖=1 𝑛 𝑃𝑖 βˆ’ 𝑃 βˆ™ π‘Œπ‘– 𝑖=1 𝑛 𝑃𝑖 βˆ’ 𝑃 2
  • 98. π‘Œπ‘– = 𝛽0 + 𝛽1 βˆ™ 𝑃𝑖 + πœ–π‘– 𝛽1 = 𝑖=1 𝑛 𝑃𝑖 βˆ’ 𝑃 βˆ™ π‘Œπ‘– 𝑖=1 𝑛 𝑃𝑖 βˆ’ 𝑃 2
  • 99. 𝐸 𝛽1 = 𝛽1
  • 100.
  • 101. 𝐸 𝛽1 = 𝛽1
  • 102.
  • 103. 𝛽1 = 𝑖=1 𝑛 𝑃𝑖 βˆ’ 𝑃 βˆ™ π‘Œπ‘– 𝑖=1 𝑛 𝑃𝑖 βˆ’ 𝑃 2
  • 104. 𝐸( 𝛽1) = 𝐸 𝑖=1 𝑛 𝑃𝑖 βˆ’ 𝑃 βˆ™ π‘Œπ‘– 𝑖=1 𝑛 𝑃𝑖 βˆ’ 𝑃 2
  • 105. π‘Œπ‘– = 𝛽0 + 𝛽1 βˆ™ 𝑃𝑖 + πœ–π‘–
  • 106. 𝐸( 𝛽1) = 𝐸 𝑖=1 𝑛 𝑃𝑖 βˆ’ 𝑃 βˆ™ π‘Œπ‘– 𝑖=1 𝑛 𝑃𝑖 βˆ’ 𝑃 2
  • 107. 𝐸 𝛽1 = 𝐸 𝑖=1 𝑛 𝑃𝑖 βˆ’ 𝑃 βˆ™ 𝛽0 + 𝛽1 βˆ™ 𝑃𝑖 + πœ–π‘– 𝑖=1 𝑛 𝑃𝑖 βˆ’ 𝑃 2
  • 108. 𝐸 𝛽1 = 𝐸 𝑖=1 𝑛 𝛽0 βˆ™ 𝑃𝑖 βˆ’ 𝑃 𝑖=1 𝑛 𝑃𝑖 βˆ’ 𝑃 2 + 𝑖=1 𝑛 𝛽1 βˆ™ 𝑃𝑖 βˆ™ 𝑃𝑖 βˆ’ 𝑃 𝑖=1 𝑛 𝑃𝑖 βˆ’ 𝑃 2
  • 109. 𝐸 π‘Š + 𝑍 = 𝐸 π‘Š + 𝐸 𝑍 𝐸 π‘Š + 𝑍 + 𝑄 = 𝐸 π‘Š + 𝐸 𝑍 + 𝐸 𝑄
  • 110. 𝐸 π‘Š + 𝑍 = 𝐸 π‘Š + 𝐸 𝑍 𝐸 π‘Š + 𝑍 + 𝑄 = 𝐸 π‘Š + 𝐸 𝑍 + 𝐸 𝑄
  • 111. 𝐸 𝛽1 = 𝐸 𝑖=1 𝑛 𝛽0 βˆ™ 𝑃𝑖 βˆ’ 𝑃 𝑖=1 𝑛 𝑃𝑖 βˆ’ 𝑃 2 + 𝑖=1 𝑛 𝛽1 βˆ™ 𝑃𝑖 βˆ™ 𝑃𝑖 βˆ’ 𝑃 𝑖=1 𝑛 𝑃𝑖 βˆ’ 𝑃 2
  • 112. 𝐸 𝛽1 = 𝐸 𝑖=1 𝑛 𝛽0 βˆ™ 𝑃𝑖 βˆ’ 𝑃 𝑖=1 𝑛 𝑃𝑖 βˆ’ 𝑃 2 + 𝑖=1 𝑛 𝛽1 βˆ™ 𝑃𝑖 βˆ™ 𝑃𝑖 βˆ’ 𝑃 𝑖=1 𝑛 𝑃𝑖 βˆ’ 𝑃 2
  • 113. 𝐸 𝛽1 = 𝐸 𝑖=1 𝑛 𝛽0 βˆ™ 𝑃𝑖 βˆ’ 𝑃 𝑖=1 𝑛 𝑃𝑖 βˆ’ 𝑃 2 + 𝑖=1 𝑛 𝛽1 βˆ™ 𝑃𝑖 βˆ™ 𝑃𝑖 βˆ’ 𝑃 𝑖=1 𝑛 𝑃𝑖 βˆ’ 𝑃 2
  • 114. 𝐸 𝛽1 = 𝐸 𝑖=1 𝑛 𝛽0 βˆ™ 𝑃𝑖 βˆ’ 𝑃 𝑖=1 𝑛 𝑃𝑖 βˆ’ 𝑃 2 + 𝑖=1 𝑛 𝛽1 βˆ™ 𝑃𝑖 βˆ™ 𝑃𝑖 βˆ’ 𝑃 𝑖=1 𝑛 𝑃𝑖 βˆ’ 𝑃 2
  • 115. 𝐸 𝛽1 = 𝐸 𝑖=1 𝑛 𝛽0 βˆ™ 𝑃𝑖 βˆ’ 𝑃 𝑖=1 𝑛 𝑃𝑖 βˆ’ 𝑃 2 + 𝑖=1 𝑛 𝛽1 βˆ™ 𝑃𝑖 βˆ™ 𝑃𝑖 βˆ’ 𝑃 𝑖=1 𝑛 𝑃𝑖 βˆ’ 𝑃 2
  • 116. 𝐸 𝛽1 = 𝐸 𝑖=1 𝑛 𝛽0 βˆ™ 𝑃𝑖 βˆ’ 𝑃 𝑖=1 𝑛 𝑃𝑖 βˆ’ 𝑃 2 + 𝐸 𝑖=1 𝑛 𝛽1 βˆ™ 𝑃𝑖 βˆ™ 𝑃𝑖 βˆ’ 𝑃 𝑖=1 𝑛 𝑃𝑖 βˆ’ 𝑃 2 + 𝐸 𝑖=1 𝑛 πœ–π‘– βˆ™ 𝑃𝑖 βˆ’ 𝑃 𝑖=1 𝑛 𝑃𝑖 βˆ’ 𝑃 2
  • 117. 𝐸 𝛽1 = 𝐸 𝑖=1 𝑛 𝛽0 βˆ™ 𝑃𝑖 βˆ’ 𝑃 𝑖=1 𝑛 𝑃𝑖 βˆ’ 𝑃 2 + 𝐸 𝑖=1 𝑛 𝛽1 βˆ™ 𝑃𝑖 βˆ™ 𝑃𝑖 βˆ’ 𝑃 𝑖=1 𝑛 𝑃𝑖 βˆ’ 𝑃 2 + 𝐸 𝑖=1 𝑛 πœ–π‘– βˆ™ 𝑃𝑖 βˆ’ 𝑃 𝑖=1 𝑛 𝑃𝑖 βˆ’ 𝑃 2
  • 118. 𝐸 𝑐 βˆ™ π‘Š = 𝑐 βˆ™ 𝐸 π‘Š
  • 119. 𝐸 𝛽1 = 𝐸 𝑖=1 𝑛 𝛽0 βˆ™ 𝑃𝑖 βˆ’ 𝑃 𝑖=1 𝑛 𝑃𝑖 βˆ’ 𝑃 2 + 𝐸 𝑖=1 𝑛 𝛽1 βˆ™ 𝑃𝑖 βˆ™ 𝑃𝑖 βˆ’ 𝑃 𝑖=1 𝑛 𝑃𝑖 βˆ’ 𝑃 2 + 𝐸 𝑖=1 𝑛 πœ–π‘– βˆ™ 𝑃𝑖 βˆ’ 𝑃 𝑖=1 𝑛 𝑃𝑖 βˆ’ 𝑃 2
  • 120. 𝐸 𝛽1 = 𝛽0 βˆ™ 𝐸 𝑖=1 𝑛 𝑃𝑖 βˆ’ 𝑃 𝑖=1 𝑛 𝑃𝑖 βˆ’ 𝑃 2 + 𝐸 𝑖=1 𝑛 𝛽1 βˆ™ 𝑃𝑖 βˆ™ 𝑃𝑖 βˆ’ 𝑃 𝑖=1 𝑛 𝑃𝑖 βˆ’ 𝑃 2 + 𝐸 𝑖=1 𝑛 πœ–π‘– βˆ™ 𝑃𝑖 βˆ’ 𝑃 𝑖=1 𝑛 𝑃𝑖 βˆ’ 𝑃 2
  • 121. 𝐸 𝛽1 = 𝛽0 βˆ™ 𝐸 𝑖=1 𝑛 𝑃𝑖 βˆ’ 𝑃 𝑖=1 𝑛 𝑃𝑖 βˆ’ 𝑃 2 + 𝐸 𝑖=1 𝑛 𝛽1 βˆ™ 𝑃𝑖 βˆ™ 𝑃𝑖 βˆ’ 𝑃 𝑖=1 𝑛 𝑃𝑖 βˆ’ 𝑃 2 + 𝐸 𝑖=1 𝑛 πœ–π‘– βˆ™ 𝑃𝑖 βˆ’ 𝑃 𝑖=1 𝑛 𝑃𝑖 βˆ’ 𝑃 2
  • 122. 𝐸 𝛽1 = 𝛽0 βˆ— 𝐸 0 𝑖=1 𝑛 𝑃𝑖 βˆ’ 𝑃 2 + 𝐸 𝑖=1 𝑛 𝛽1 βˆ™ 𝑃𝑖 βˆ™ 𝑃𝑖 βˆ’ 𝑃 𝑖=1 𝑛 𝑃𝑖 βˆ’ 𝑃 2 + 𝐸 𝑖=1 𝑛 πœ–π‘– βˆ™ 𝑃𝑖 βˆ’ 𝑃 𝑖=1 𝑛 𝑃𝑖 βˆ’ 𝑃 2
  • 123. 𝐸 𝛽1 = 𝛽0 βˆ— 0 +𝐸 𝑖=1 𝑛 𝛽1 βˆ™ 𝑃𝑖 βˆ™ 𝑃𝑖 βˆ’ 𝑃 𝑖=1 𝑛 𝑃𝑖 βˆ’ 𝑃 2 + 𝐸 𝑖=1 𝑛 πœ–π‘– βˆ™ 𝑃𝑖 βˆ’ 𝑃 𝑖=1 𝑛 𝑃𝑖 βˆ’ 𝑃 2
  • 124. 𝐸 𝛽1 = 𝐸 𝑖=1 𝑛 𝛽1 βˆ™ 𝑃𝑖 βˆ™ 𝑃𝑖 βˆ’ 𝑃 𝑖=1 𝑛 𝑃𝑖 βˆ’ 𝑃 2 + 𝐸 𝑖=1 𝑛 πœ–π‘– βˆ™ 𝑃𝑖 βˆ’ 𝑃 𝑖=1 𝑛 𝑃𝑖 βˆ’ 𝑃 2
  • 125. 𝐸 𝛽1 = 𝐸 𝑖=1 𝑛 𝛽1 βˆ™ 𝑃𝑖 βˆ™ 𝑃𝑖 βˆ’ 𝑃 𝑖=1 𝑛 𝑃𝑖 βˆ’ 𝑃 2 + 𝐸 𝑖=1 𝑛 πœ–π‘– βˆ™ 𝑃𝑖 βˆ’ 𝑃 𝑖=1 𝑛 𝑃𝑖 βˆ’ 𝑃 2
  • 126. 𝐸 𝛽1 = 𝛽1 βˆ™ 𝐸 𝑖=1 𝑛 𝑃𝑖 βˆ™ 𝑃𝑖 βˆ’ 𝑃 𝑖=1 𝑛 𝑃𝑖 βˆ’ 𝑃 2 + 𝐸 𝑖=1 𝑛 πœ–π‘– βˆ™ 𝑃𝑖 βˆ’ 𝑃 𝑖=1 𝑛 𝑃𝑖 βˆ’ 𝑃 2
  • 127. 𝐸 𝛽1 = 𝛽1 βˆ™ 𝐸 𝑖=1 𝑛 𝑃𝑖 βˆ™ 𝑃𝑖 βˆ’ 𝑃 𝑖=1 𝑛 𝑃𝑖 βˆ’ 𝑃 2 + 𝐸 𝑖=1 𝑛 πœ–π‘– βˆ™ 𝑃𝑖 βˆ’ 𝑃 𝑖=1 𝑛 𝑃𝑖 βˆ’ 𝑃 2
  • 128. 𝑖=1 𝑛 𝑃𝑖 βˆ™ 𝑃𝑖 βˆ’ 𝑃 = 𝑖=1 𝑛 𝑃𝑖 βˆ’ 𝑃 2
  • 129. 𝐸 𝛽1 = 𝛽1 βˆ— 𝐸 𝑖=1 𝑛 𝑃𝑖 βˆ™ 𝑃𝑖 βˆ’ 𝑃 𝑖=1 𝑛 𝑃𝑖 βˆ’ 𝑃 2 + 𝐸 𝑖=1 𝑛 πœ–π‘– βˆ™ 𝑃𝑖 βˆ’ 𝑃 𝑖=1 𝑛 𝑃𝑖 βˆ’ 𝑃 2
  • 130. 𝐸 𝛽1 = 𝛽1 βˆ™ 𝐸 𝑖=1 𝑛 𝑃𝑖 βˆ’ 𝑃 2 𝑖=1 𝑛 𝑃𝑖 βˆ’ 𝑃 2 + 𝐸 𝑖=1 𝑛 πœ–π‘– βˆ™ 𝑃𝑖 βˆ’ 𝑃 𝑖=1 𝑛 𝑃𝑖 βˆ’ 𝑃 2
  • 131. 𝐸 𝛽1 = 𝛽1 βˆ™ 𝐸 1 +𝐸 𝑖=1 𝑛 πœ–π‘– βˆ™ 𝑃𝑖 βˆ’ 𝑃 𝑖=1 𝑛 𝑃𝑖 βˆ’ 𝑃 2
  • 132. 𝐸 𝛽1 = 𝛽1 βˆ™ 1 +𝐸 𝑖=1 𝑛 πœ–π‘– βˆ™ 𝑃𝑖 βˆ’ 𝑃 𝑖=1 𝑛 𝑃𝑖 βˆ’ 𝑃 2
  • 133. 𝐸 𝛽1 = 𝛽1 +𝐸 𝑖=1 𝑛 πœ–π‘– βˆ™ 𝑃𝑖 βˆ’ 𝑃 𝑖=1 𝑛 𝑃𝑖 βˆ’ 𝑃 2
  • 134. 𝐸 𝛽1 = 𝛽1 +𝐸 𝑖=1 𝑛 πœ–π‘– βˆ™ 𝑃𝑖 βˆ’ 𝑃 𝑖=1 𝑛 𝑃𝑖 βˆ’ 𝑃 2
  • 137. 𝐸 𝛽1 = 𝛽1 +𝐸 𝑖=1 𝑛 πœ–π‘– βˆ™ 𝑃𝑖 βˆ’ 𝑃 𝑖=1 𝑛 𝑃𝑖 βˆ’ 𝑃 2
  • 138. 𝐸 π‘Š βˆ™ 𝑍 β‰  𝐸 π‘Š βˆ™ 𝐸 𝑍 Expectations: Properties
  • 139. 𝐸 𝛽1 = 𝛽1 +𝐸 𝑖=1 𝑛 πœ–π‘– βˆ™ 𝑃𝑖 βˆ’ 𝑃 𝑖=1 𝑛 𝑃𝑖 βˆ’ 𝑃 2
  • 140. 𝐸 𝛽1 = 𝛽1 + 𝑖=1 𝑛 𝐸 πœ–π‘– βˆ™ 𝑃𝑖 βˆ’ 𝑃 𝑖=1 𝑛 𝑃𝑖 βˆ’ 𝑃 2
  • 141. 𝐸 𝛽1 = 𝛽1 + 𝑖=1 𝑛 𝐸 𝑖=1 𝑛 𝑃𝑖 βˆ’ 𝑃 𝑖=1 𝑛 𝑃𝑖 βˆ’ 𝑃 2 βˆ™ 𝐸 πœ–π‘–
  • 142. 𝐸 𝛽1 = 𝛽1 + 𝑖=1 𝑛 𝐸 𝑖=1 𝑛 𝑃𝑖 βˆ’ 𝑃 𝑖=1 𝑛 𝑃𝑖 βˆ’ 𝑃 2 βˆ™ 0
  • 145. 𝐸 𝛽1 = 𝛽1 π‘Œπ‘– = 𝛽0 + 𝛽1 βˆ™ 𝑃𝑖 + πœ–π‘–
  • 146. 𝐸 𝛽1 = 𝛽1 π‘Œπ‘– = 𝛽0 + 𝛽1 βˆ™ 𝑃𝑖 + πœ–π‘–
  • 147. 𝐸 𝛽1 = 𝛽1 π‘Œπ‘– = 𝛽0 + 𝛽1 βˆ™ 𝑃𝑖 + πœ–π‘–
  • 148. Peter M. Lance, PhD MEASURE Evaluation University of North Carolina at Chapel Hill September 13 and 15, 2016 Selection on Observables: Part Deux
  • 149. π‘Œ0 = 𝛽0 + πœ– π‘Œ1 = 𝛽0 + 𝛽1 + πœ– π‘Œ = 𝑃 βˆ™ π‘Œ1 + 1 βˆ’ 𝑃 βˆ™ π‘Œ0 π‘Œ = 𝛽0 + 𝛽1 βˆ™ 𝑃 + πœ–
  • 150. π‘Œ0 = 𝛽0 + πœ– π‘Œ1 = 𝛽0 + 𝛽1 + πœ– π‘Œ = 𝑃 βˆ™ π‘Œ1 + 1 βˆ’ 𝑃 βˆ™ π‘Œ0 π‘Œ = 𝛽0 + 𝛽1 βˆ™ 𝑃 + πœ–
  • 151. π‘Œ0 = 𝛽0 + πœ– π‘Œ1 = 𝛽0 + 𝛽1 + πœ– π‘Œ = 𝑃 βˆ™ π‘Œ1 + 1 βˆ’ 𝑃 βˆ™ π‘Œ0 π‘Œ = 𝛽0 + 𝛽1 βˆ™ 𝑃 + πœ–
  • 152. 𝐸 𝛽1 = 𝛽1 +𝐸 𝑖=1 𝑛 πœ–π‘– βˆ™ 𝑃𝑖 βˆ’ 𝑃 𝑖=1 𝑛 𝑃𝑖 βˆ’ 𝑃 2
  • 153. 𝐸 𝛽1 = 𝛽1 +𝐸 𝑖=1 𝑛 πœ–π‘– βˆ™ 𝑃𝑖 βˆ’ 𝑃 𝑖=1 𝑛 𝑃𝑖 βˆ’ 𝑃 2
  • 154. 𝐸 𝛽1 = 𝛽1 π‘Œπ‘– = 𝛽0 + 𝛽1 βˆ™ 𝑃𝑖 + πœ–π‘–
  • 155. 𝐸 𝛽1 = 𝛽1 π‘Œπ‘– = 𝛽0 + 𝛽1 βˆ™ 𝑃𝑖 + πœ–π‘–
  • 156. 𝐸 𝛽1 = 𝛽1 π‘Œπ‘– = 𝛽0 + 𝛽1 βˆ™ 𝑃𝑖 + πœ–π‘–
  • 157. π‘Œ0 = 𝛽0 + πœ€ π‘Œ1 = 𝛽0 + 𝛽1 + πœ€
  • 158. π‘Œ0 = 𝛽0 + 𝛽2 βˆ™ π‘₯ + πœ€ π‘Œ1 = 𝛽0 + 𝛽1 + 𝛽2 βˆ™ π‘₯ + πœ€
  • 159. π‘Œ1 βˆ’ π‘Œ0 = 𝛽0 + 𝛽1 + πœ– βˆ’ 𝛽0 + πœ– = 𝛽0 + 𝛽1 + πœ– βˆ’ 𝛽0 βˆ’ πœ– = 𝛽1
  • 160. π‘Œ1 βˆ’ π‘Œ0 = 𝛽0 + 𝛽1 + 𝛽2 βˆ— π‘₯ + πœ€ βˆ’ 𝛽0 + 𝛽2 βˆ— π‘₯ + πœ€ = 𝛽0 + 𝛽1 + πœ– βˆ’ 𝛽0 βˆ’ πœ– = 𝛽1
  • 162. π‘Œ = 𝑃 βˆ™ π‘Œ1 + 1 βˆ’ 𝑃 βˆ™ π‘Œ0 = 𝑃 βˆ— 𝛽0 + 𝛽1 + πœ– + 1 βˆ’ 𝑃 βˆ— 𝛽0 + πœ– = 𝑃 βˆ— 𝛽0 + 𝑃 βˆ— 𝛽1 + 𝑃 βˆ— πœ– +𝛽0 + πœ– βˆ’ 𝑃 βˆ— 𝛽0 βˆ’ 𝑃 βˆ— πœ– = 𝛽0 + 𝑃 βˆ— 𝛽1 + πœ–
  • 163. π‘Œ = 𝑃 βˆ™ π‘Œ1 + 1 βˆ’ 𝑃 βˆ™ π‘Œ0 = 𝑃 βˆ™ 𝛽0 + 𝛽1 + 𝛽2 βˆ™ π‘₯ + πœ€ + 1 βˆ’ 𝑃 βˆ™ 𝛽0 + 𝛽2 βˆ™ π‘₯ + πœ€ = 𝑃 βˆ— 𝛽0 + 𝑃 βˆ— 𝛽1 + 𝑃 βˆ— πœ– +𝛽0 + πœ– βˆ’ 𝑃 βˆ— 𝛽0 βˆ’ 𝑃 βˆ— πœ–
  • 164. π‘Œ = 𝛽0 + 𝛽1 βˆ™ 𝑃 + 𝛽2 βˆ™ π‘₯ + πœ€
  • 165. Cost of participation 𝐢 = 𝜌0 + 𝜌1 βˆ™ π‘₯
  • 166. Cost of participation 𝐢 = 𝜌0 + 𝜌1 βˆ™ π‘₯
  • 167. Benefit-Cost>0 π‘Œ1 βˆ’ π‘Œ0 βˆ’ C > 0 𝛽1 βˆ’ 𝐢 > 0 𝛽1 βˆ’ 𝛾0 + 𝛾1 βˆ— π‘₯ > 0
  • 168. Benefit-Cost>0 π‘Œ1 βˆ’ π‘Œ0 βˆ’ C > 0 𝛽1 βˆ’ 𝐢 > 0 𝛽1 βˆ’ 𝛾0 + 𝛾1 βˆ— π‘₯ > 0
  • 169. Benefit-Cost>0 π‘Œ1 βˆ’ π‘Œ0 βˆ’ C > 0 𝛽1 βˆ’ 𝐢 > 0 𝛽1 βˆ’ 𝛾0 + 𝛾1 βˆ— π‘₯ > 0
  • 170. Benefit-Cost>0 π‘Œ1 βˆ’ π‘Œ0 βˆ’ C > 0 𝛽1 βˆ’ 𝐢 > 0 𝛽1 βˆ’ 𝜌0 + 𝜌1 βˆ™ π‘₯ > 0
  • 172. X Y P
  • 173. Benefit-Cost>0 𝛽1 βˆ’ 𝜌0 + 𝜌1 βˆ™ π‘₯ > 0 πœ€
  • 174. Benefit-Cost>0 𝛽1 βˆ’ 𝜌0 + 𝜌1 βˆ™ π‘₯ > 0 πœ€
  • 175. Benefit-Cost>0 𝛽1 βˆ’ 𝜌0 + 𝜌1 βˆ™ π‘₯ > 0 πœ€
  • 176. P and 𝜺 are independent
  • 177. True model: π‘Œ = 𝛽0 + 𝛽1 βˆ™ 𝑃 + 𝛽2 βˆ™ π‘₯ + πœ€ We actually attempt to estimate: π‘Œ = 𝜏0 + 𝜏1 βˆ™ 𝑃 + πœ– Error term now contains: 𝛽2 βˆ™ π‘₯
  • 178. True model: π‘Œ = 𝛽0 + 𝛽1 βˆ™ 𝑃 + 𝛽2 βˆ™ π‘₯ + πœ€ We actually attempt to estimate: π‘Œ = 𝜏0 + 𝜏1 βˆ™ 𝑃 + πœ– Error term now contains: 𝛽2 βˆ™ π‘₯
  • 179. True model: π‘Œ = 𝛽0 + 𝛽1 βˆ™ 𝑃 + 𝛽2 βˆ™ π‘₯ + πœ€ We actually attempt to estimate: π‘Œ = 𝜏0 + 𝜏1 βˆ™ 𝑃 + πœ– Error term now contains: 𝛽2 βˆ™ π‘₯
  • 180. 𝐸( 𝜏1) = 𝐸 𝑖=1 𝑛 𝑃𝑖 βˆ’ 𝑃 βˆ™ π‘Œπ‘– 𝑖=1 𝑛 𝑃𝑖 βˆ’ 𝑃 2
  • 181. 𝐸 𝜏1 = 𝐸 𝑖=1 𝑛 𝑃𝑖 βˆ’ 𝑃 βˆ™ 𝛽0 + 𝛽1 βˆ™ 𝑃𝑖 + 𝛽2 βˆ™ π‘₯𝑖 + πœ€π‘– 𝑖=1 𝑛 𝑃𝑖 βˆ’ 𝑃 2
  • 182. 𝐸 𝜏1 = 𝐸 𝑖=1 𝑛 𝛽0 βˆ™ 𝑃𝑖 βˆ’ 𝑃 𝑖=1 𝑛 𝑃𝑖 βˆ’ 𝑃 2 + 𝑖=1 𝑛 𝛽1 βˆ™ 𝑃𝑖 βˆ™ 𝑃𝑖 βˆ’ 𝑃 𝑖=1 𝑛 𝑃𝑖 βˆ’ 𝑃 2 + 𝑖=1 𝑛 𝛽2 βˆ™ π‘₯𝑖 βˆ™ 𝑃𝑖 βˆ’ 𝑃 𝑖=1 𝑛 𝑃𝑖 βˆ’ 𝑃 2 + 𝑖=1 𝑛 πœ€π‘– βˆ™ 𝑃𝑖 βˆ’ 𝑃 𝑖=1 𝑛 𝑃𝑖 βˆ’ 𝑃 2
  • 183. 𝐸 𝜏1 = 𝐸 𝑖=1 𝑛 𝛽0 βˆ™ 𝑃𝑖 βˆ’ 𝑃 𝑖=1 𝑛 𝑃𝑖 βˆ’ 𝑃 2 + 𝐸 𝑖=1 𝑛 𝛽1 βˆ™ 𝑃𝑖 βˆ™ 𝑃𝑖 βˆ’ 𝑃 𝑖=1 𝑛 𝑃𝑖 βˆ’ 𝑃 2 +𝐸 𝑖=1 𝑛 𝛽2 βˆ™ π‘₯𝑖 βˆ™ 𝑃𝑖 βˆ’ 𝑃 𝑖=1 𝑛 𝑃𝑖 βˆ’ 𝑃 2 +𝐸 𝑖=1 𝑛 πœ€π‘– βˆ™ 𝑃𝑖 βˆ’ 𝑃 𝑖=1 𝑛 𝑃𝑖 βˆ’ 𝑃 2
  • 184. 𝐸 𝜏1 = 𝐸 𝑖=1 𝑛 𝛽0 βˆ™ 𝑃𝑖 βˆ’ 𝑃 𝑖=1 𝑛 𝑃𝑖 βˆ’ 𝑃 2 + 𝐸 𝑖=1 𝑛 𝛽1 βˆ™ 𝑃𝑖 βˆ™ 𝑃𝑖 βˆ’ 𝑃 𝑖=1 𝑛 𝑃𝑖 βˆ’ 𝑃 2 +𝐸 𝑖=1 𝑛 𝛽2 βˆ™ π‘₯𝑖 βˆ™ 𝑃𝑖 βˆ’ 𝑃 𝑖=1 𝑛 𝑃𝑖 βˆ’ 𝑃 2 +𝐸 𝑖=1 𝑛 πœ€π‘– βˆ™ 𝑃𝑖 βˆ’ 𝑃 𝑖=1 𝑛 𝑃𝑖 βˆ’ 𝑃 2
  • 185. 𝐸 𝜏1 = 0 +𝐸 𝑖=1 𝑛 𝛽1 βˆ™ 𝑃𝑖 βˆ™ 𝑃𝑖 βˆ’ 𝑃 𝑖=1 𝑛 𝑃𝑖 βˆ’ 𝑃 2 +𝐸 𝑖=1 𝑛 𝛽2 βˆ™ π‘₯𝑖 βˆ™ 𝑃𝑖 βˆ’ 𝑃 𝑖=1 𝑛 𝑃𝑖 βˆ’ 𝑃 2 +𝐸 𝑖=1 𝑛 πœ€π‘– βˆ™ 𝑃𝑖 βˆ’ 𝑃 𝑖=1 𝑛 𝑃𝑖 βˆ’ 𝑃 2
  • 186. 𝐸 𝜏1 = 𝐸 𝑖=1 𝑛 𝛽1 βˆ™ 𝑃𝑖 βˆ™ 𝑃𝑖 βˆ’ 𝑃 𝑖=1 𝑛 𝑃𝑖 βˆ’ 𝑃 2 +𝐸 𝑖=1 𝑛 𝛽2 βˆ™ π‘₯𝑖 βˆ™ 𝑃𝑖 βˆ’ 𝑃 𝑖=1 𝑛 𝑃𝑖 βˆ’ 𝑃 2 +𝐸 𝑖=1 𝑛 πœ€π‘– βˆ™ 𝑃𝑖 βˆ’ 𝑃 𝑖=1 𝑛 𝑃𝑖 βˆ’ 𝑃 2
  • 187. 𝐸 𝜏1 = 𝐸 𝑖=1 𝑛 𝛽1 βˆ™ 𝑃𝑖 βˆ™ 𝑃𝑖 βˆ’ 𝑃 𝑖=1 𝑛 𝑃𝑖 βˆ’ 𝑃 2 +𝐸 𝑖=1 𝑛 𝛽2 βˆ™ π‘₯𝑖 βˆ™ 𝑃𝑖 βˆ’ 𝑃 𝑖=1 𝑛 𝑃𝑖 βˆ’ 𝑃 2 +𝐸 𝑖=1 𝑛 πœ€π‘– βˆ™ 𝑃𝑖 βˆ’ 𝑃 𝑖=1 𝑛 𝑃𝑖 βˆ’ 𝑃 2
  • 188. 𝐸 𝜏1 = 𝛽1 +𝐸 𝑖=1 𝑛 𝛽2 βˆ™ π‘₯𝑖 βˆ™ 𝑃𝑖 βˆ’ 𝑃 𝑖=1 𝑛 𝑃𝑖 βˆ’ 𝑃 2 +𝐸 𝑖=1 𝑛 πœ€π‘– βˆ™ 𝑃𝑖 βˆ’ 𝑃 𝑖=1 𝑛 𝑃𝑖 βˆ’ 𝑃 2
  • 189. 𝐸 𝜏1 = 𝛽1 +𝐸 𝑖=1 𝑛 𝛽2 βˆ™ π‘₯𝑖 βˆ™ 𝑃𝑖 βˆ’ 𝑃 𝑖=1 𝑛 𝑃𝑖 βˆ’ 𝑃 2 +𝐸 𝑖=1 𝑛 πœ€π‘– βˆ™ 𝑃𝑖 βˆ’ 𝑃 𝑖=1 𝑛 𝑃𝑖 βˆ’ 𝑃 2
  • 190. 𝐸 𝜏1 = 𝛽1 +𝐸 𝑖=1 𝑛 𝛽2 βˆ™ π‘₯𝑖 βˆ™ 𝑃𝑖 βˆ’ 𝑃 𝑖=1 𝑛 𝑃𝑖 βˆ’ 𝑃 2 +0
  • 191. 𝐸 𝜏1 = 𝛽1 +𝐸 𝑖=1 𝑛 𝛽2 βˆ™ π‘₯𝑖 βˆ™ 𝑃𝑖 βˆ’ 𝑃 𝑖=1 𝑛 𝑃𝑖 βˆ’ 𝑃 2
  • 192. 𝐸 𝜏1 = 𝛽1 +𝐸 𝑖=1 𝑛 𝛽2 βˆ™ π‘₯𝑖 βˆ™ 𝑃𝑖 βˆ’ 𝑃 𝑖=1 𝑛 𝑃𝑖 βˆ’ 𝑃 2
  • 193. 𝐸 𝑐 βˆ™ π‘Š = 𝑐 βˆ™ 𝐸 π‘Š
  • 194. 𝐸 𝜏1 = 𝛽1 +𝛽2 βˆ™ 𝐸 𝑖=1 𝑛 π‘₯𝑖 βˆ™ 𝑃𝑖 βˆ’ 𝑃 𝑖=1 𝑛 𝑃𝑖 βˆ’ 𝑃 2
  • 195. 𝐸 𝜏1 = 𝛽1 +𝛽2 βˆ™ 𝐸 𝑖=1 𝑛 π‘₯𝑖 βˆ™ 𝑃𝑖 βˆ’ 𝑃 𝑖=1 𝑛 𝑃𝑖 βˆ’ 𝑃 2
  • 196. 𝐸 𝜏1 = 𝛽1 +𝛽2 βˆ™ 𝐸 𝑖=1 𝑛 π‘₯𝑖 βˆ™ 𝑃𝑖 βˆ’ 𝑃 𝑖=1 𝑛 𝑃𝑖 βˆ’ 𝑃 2 π‘₯𝑖= 𝛾0 + 𝛾1 βˆ™ 𝑃𝑖 + πœ—π‘–
  • 197. 𝐸 𝜏1 = 𝛽1 +𝛽2 βˆ™ 𝐸 𝑖=1 𝑛 π‘₯𝑖 βˆ™ 𝑃𝑖 βˆ’ 𝑃 𝑖=1 𝑛 𝑃𝑖 βˆ’ 𝑃 2 π‘₯𝑖= 𝛾0 + 𝛾1 βˆ™ 𝑃𝑖 + πœ—π‘–
  • 198. 𝐸 𝜏1 = 𝛽1 +𝛽2 βˆ™ 𝐸 𝑖=1 𝑛 π‘₯𝑖 βˆ™ 𝑃𝑖 βˆ’ 𝑃 𝑖=1 𝑛 𝑃𝑖 βˆ’ 𝑃 2 π‘₯𝑖= 𝛾0 + 𝛾1 βˆ™ 𝑃𝑖 + πœ—π‘–
  • 199. 𝐸 𝜏1 = 𝛽1 +𝛽2 βˆ™ 𝛾1 𝑖=1 𝑛 π‘₯ 𝑖=1 𝑛 π‘₯ π‘₯𝑖= 𝛾0 + 𝛾1 βˆ™ 𝑃𝑖 + πœ—π‘–
  • 200. True model: π‘Œ = 𝛽0 + 𝛽1 βˆ™ 𝑃 + 𝛽2 βˆ™ π‘₯ + πœ€ We actually attempt to estimate: π‘Œ = 𝜏0 + 𝜏1 βˆ™ 𝑃 + πœ– Error term now contains: 𝛽2 βˆ™ π‘₯
  • 201. 𝐸 𝜏1 = 𝛽1 + 𝛽2 βˆ™ 𝐸 𝑖=1 𝑛 π‘₯𝑖 βˆ™ 𝑃𝑖 βˆ’ 𝑃 𝑖=1 𝑛 𝑃𝑖 βˆ’ 𝑃 2
  • 202. 𝐸 𝜏1 = 𝛽1 + 𝛽2 βˆ™ 𝛾1
  • 203. 𝐸 𝜏1 = 𝛽1 + 𝛽2 βˆ™ 𝛾1 The Actual Causal Effect of P on y
  • 204. 𝐸 𝜏1 = 𝛽1 + 𝛽2 βˆ™ 𝛾1 The actual causal effect of P on y The actual causal effect of the omitted variable X on Y
  • 205. 𝐸 𝜏1 = 𝛽1 + 𝛽2 βˆ™ 𝛾1 The actual causal effect of P on y Th”Effect” of P on x: π‘₯𝑖= 𝛾0 + 𝛾1 βˆ™ 𝑃𝑖 + πœ—π‘– The actual causal effect of the omitted variable X on Y
  • 206. True model: π‘Œ = 𝛽0 + 𝛽1 βˆ™ 𝑃 + 𝛽2 βˆ™ π‘₯ + πœ€ We actually attempt to estimate: π‘Œ = 𝜏0 + 𝜏1 βˆ™ 𝑃 + πœ– Error term now contains: 𝛽2 βˆ™ π‘₯
  • 207. True model: π‘Œ = 𝛽0 + 𝛽1 βˆ™ 𝑃 + 𝛽2 βˆ™ π‘₯ + πœ€ We actually attempt to estimate: π‘Œ = 𝜏0 + 𝜏1 βˆ™ 𝑃 + πœ– Error term now contains: 𝛽2 βˆ™ π‘₯
  • 208. True model: π‘Œ = 𝛽0 + 𝛽1 βˆ™ 𝑃 + 𝛽2 βˆ™ π‘₯ + πœ€ We actually attempt to estimate: π‘Œ = 𝜏0 + 𝜏1 βˆ™ 𝑃 + πœ– Error term now contains: 𝛽2 βˆ™ π‘₯
  • 209. True model: π‘Œ = 𝛽0 + 𝛽1 βˆ™ 𝑃 + 𝛽2 βˆ™ π‘₯ + πœ€ We actually attempt to estimate: π‘Œ = 𝜏0 + 𝜏1 βˆ™ 𝑃 + πœ– Error term now contains: 𝛽2 βˆ™ π‘₯
  • 210. True model: π‘Œ = 𝛽0 + 𝛽1 βˆ™ 𝑃 + 𝛽2 βˆ™ π‘₯ + πœ€ We actually attempt to estimate: π‘Œ = 𝜏0 + 𝜏1 βˆ™ 𝑃 + πœ– 𝑬 𝝉 𝟏 β‰  𝜷
  • 211. 𝐸 𝜏1 = 𝛽1 + 𝛽2 βˆ™ 𝛾1 The actual causal effect of P on y Th”Effect” of P on x: π‘₯𝑖= 𝛾0 + 𝛾1 βˆ™ 𝑃𝑖 + πœ—π‘– The actual causal effect of the omitted variable X on Y
  • 212. 𝐸 𝜏1 = 𝛽1 + 𝛽2 βˆ™ 𝛾1 The actual causal effect of P on y Th”Effect” of P on x: π‘₯𝑖= 𝛾0 + 𝛾1 βˆ™ 𝑃𝑖 + πœ—π‘– The actual causal effect of the omitted variable X on Y
  • 213. π‘Œ0 = 𝛽0 + 𝛽2 βˆ™ π‘₯1 + 𝛽3 βˆ™ π‘₯2 + 𝛽4 βˆ™ π‘₯2 + πœ€ π‘Œ1 = 𝛽0 + 𝛽1 + 𝛽2 βˆ™ π‘₯1 + 𝛽3 βˆ™ π‘₯2 + 𝛽4 βˆ™ π‘₯2 + πœ€
  • 214. π‘Œ0 = 𝛽0 + 𝛽2 βˆ™ π‘₯1 + 𝛽3 βˆ™ π‘₯2 + 𝛽4 βˆ™ π‘₯2 + πœ€ π‘Œ1 = 𝛽0 + 𝛽1 + 𝛽2 βˆ™ π‘₯1 + 𝛽3 βˆ™ π‘₯2 + 𝛽4 βˆ™ π‘₯3 + πœ€
  • 215. π‘Œ1 βˆ’ π‘Œ0 = 𝛽0 + 𝛽1 + πœ– βˆ’ 𝛽0 + πœ– = 𝛽0 + 𝛽1 + πœ– βˆ’ 𝛽0 βˆ’ πœ– = 𝛽1
  • 216. π‘Œ1 βˆ’ π‘Œ0 = 𝛽0 + 𝛽1 + 𝛽2 βˆ™ π‘₯1 + 𝛽3 βˆ™ π‘₯2 + 𝛽4 βˆ™ π‘₯2 + πœ€ βˆ’ 𝛽0 + 𝛽2 βˆ™ π‘₯1 + 𝛽3 βˆ™ π‘₯2 + 𝛽4 βˆ™ π‘₯2 + πœ€ = 𝛽0 + 𝛽1 + πœ– βˆ’ 𝛽0 βˆ’ πœ– = 𝛽1
  • 218. π‘Œ = 𝑃 βˆ™ π‘Œ1 + 1 βˆ’ 𝑃 βˆ™ π‘Œ0 = 𝑃 βˆ— 𝛽0 + 𝛽1 + πœ– + 1 βˆ’ 𝑃 βˆ— 𝛽0 + πœ– = 𝑃 βˆ— 𝛽0 + 𝑃 βˆ— 𝛽1 + 𝑃 βˆ— πœ– +𝛽0 + πœ– βˆ’ 𝑃 βˆ— 𝛽0 βˆ’ 𝑃 βˆ— πœ– = 𝛽0 + 𝑃 βˆ— 𝛽1 + πœ–
  • 219. π‘Œ = 𝑃 βˆ™ π‘Œ1 + 1 βˆ’ 𝑃 βˆ™ π‘Œ0 = 𝑃 βˆ™ 𝛽0 + 𝛽1 + 𝛽2 βˆ™ π‘₯1 + 𝛽3 βˆ™ π‘₯2 + 𝛽4 βˆ™ π‘₯3 + πœ€ + 1 βˆ’ 𝑃 βˆ™ 𝛽0 + 𝛽2 βˆ™ π‘₯ + 𝛽2 βˆ™ π‘₯1 + 𝛽3 βˆ™ π‘₯2 + 𝛽4 βˆ™ π‘₯3 + πœ€
  • 220. π‘Œ = 𝛽0 + 𝛽1 βˆ™ 𝑃 + 𝛽2 βˆ™ π‘₯1 + 𝛽3 βˆ™ π‘₯2 + 𝛽4 βˆ™ π‘₯3 + πœ€
  • 221. Cost of participation 𝐢 = 𝜌0 + 𝜌1 βˆ™ π‘₯1 + 𝜌2 βˆ™ π‘₯2 + 𝜌3 βˆ™ π‘₯3
  • 222. True model: π‘Œ = 𝛽0 + 𝛽1 βˆ™ 𝑃 + 𝛽2 βˆ™ π‘₯1 + 𝛽3 βˆ™ π‘₯2 + 𝛽4 βˆ™ π‘₯3 + πœ€ We actually attempt to estimate: π‘Œ = 𝜏0 + 𝜏1 βˆ™ 𝑃 + πœ–
  • 223. 𝐸 𝜏1 = 𝛽1 + 𝛽2 βˆ™ 𝛾11 + 𝛽3 βˆ™ 𝛾21 + 𝛽4 βˆ™ 𝛾31
  • 224. 𝐸 𝜏1 = 𝛽1 + 𝛽2 βˆ™ 𝛾11 + 𝛽3 βˆ™ 𝛾21 + 𝛽4 βˆ™ 𝛾31 The actual causal effect of P on y
  • 225. 𝐸 𝜏1 = 𝛽1 + 𝛽2 βˆ™ 𝛾11 + 𝛽3 βˆ™ 𝛾21 + 𝛽4 βˆ™ 𝛾31 The actual causal effect of P on y Th”Effect” of P on x1: π‘₯1𝑖= 𝛾10 + 𝛾11 βˆ™ 𝑃𝑖 + πœ—1𝑖 The actual causal effect of the omitted variable X1 on Y
  • 226. 𝐸 𝜏1 = 𝛽1 + 𝛽2 βˆ™ 𝛾11 + 𝛽3 βˆ™ 𝛾21 + 𝛽4 βˆ™ 𝛾31 The actual causal effect of P on y ”Effect” of P on x1: π‘₯1𝑖= 𝛾10 + 𝛾11 βˆ™ 𝑃𝑖 + πœ—1𝑖 The actual causal effect of the omitted variable X1 on Y The actual causal effect of the omitted variable X2 on Y Th”Effect” of P on x2: π‘₯2𝑖= 𝛾20 + 𝛾21 βˆ™ 𝑃𝑖 + πœ—2𝑖
  • 227. 𝐸 𝜏1 = 𝛽1 + 𝛽2 βˆ™ 𝛾11 + 𝛽3 βˆ™ 𝛾21 + 𝛽4 βˆ™ 𝛾31 The actual causal effect of P on y ”Effect” of P on x1: π‘₯1𝑖= 𝛾10 + 𝛾11 βˆ™ 𝑃𝑖 + πœ—1𝑖 The actual causal effect of the omitted variable X1 on Y The actual causal effect of the omitted variable X2 on Y ”Effect” of P on x2: π‘₯2𝑖= 𝛾20 + 𝛾21 βˆ™ 𝑃𝑖 + πœ—2𝑖 Th”Effect” of P on x1: π‘₯3𝑖= 𝛾30 + 𝛾31 βˆ™ 𝑃𝑖 + πœ—3𝑖 The actual causal effect of the omitted variable X3 on Y
  • 228. 𝐸 𝜏1 = 𝛽1 + 𝛽2 βˆ™ 𝛾11 + 𝛽3 βˆ™ 𝛾21 + 𝛽4 βˆ™ 𝛾31 The actual causal effect of P on y ”Effect” of P on x1: π‘₯1𝑖= 𝛾10 + 𝛾11 βˆ™ 𝑃𝑖 + πœ—1𝑖 The actual causal effect of the omitted variable X1 on Y The actual causal effect of the omitted variable X2 on Y ”Effect” of P on x2: π‘₯2𝑖= 𝛾20 + 𝛾21 βˆ™ 𝑃𝑖 + πœ—2𝑖 Th”Effect” of P on x3: π‘₯3𝑖= 𝛾30 + 𝛾31 βˆ™ 𝑃𝑖 + πœ—3𝑖 The actual causal effect of the omitted variable X3 on Y
  • 229. 𝐸 𝜏1 = 𝛽1 + 𝛽2 βˆ™ 𝛾11 + 𝛽2 βˆ™ 𝛾21 + 𝛽3 βˆ™ 𝛾31
  • 230. Potential outcomes π‘Œ0 = 𝛽0 + 𝛽2 βˆ™ π‘₯1 + 𝛽3 βˆ™ π‘₯2 + πœ€ π‘Œ1 = 𝛽0 + 𝛽1 + 𝛽2 βˆ™ π‘₯1 + 𝛽3 βˆ™ π‘₯2 + πœ€ Costs of participation 𝐢 = 𝜌0 + 𝜌2 βˆ™ π‘₯2 + 𝜌3 βˆ™ π‘₯3
  • 232. 𝐸 𝜏1 = 𝛽1 + 𝛽2 βˆ™ 𝛾11 + 𝛽2 βˆ™ 𝛾21 + 𝛽3 βˆ™ 𝛾31
  • 233. 𝐸 𝜏1 = 𝛽1 + 𝛽2 βˆ™ 𝛾11 + 𝛽2 βˆ™ 𝛾21 + 𝛽3 βˆ™ 𝛾31
  • 234. 𝐸 𝜏1 = 𝛽1 + 𝛽2 βˆ™ 𝛾21 + 𝛽3 βˆ™ 𝛾31
  • 236. 𝐸 𝜏1 = 𝛽1 + 𝛽2 βˆ™ 𝛾21 + 𝛽3 βˆ™ 𝛾31
  • 237. 𝐸 𝜏1 = 𝛽1 + 𝛽2 βˆ™ 𝛾21
  • 239. 𝐸 𝜏1 = 𝛽1 + 𝛽2 βˆ™ 𝛾21
  • 241. π‘Œ = 𝛽0 + 𝛽1 βˆ™ 𝑃 + 𝛽2 βˆ™ π‘₯1 + 𝛽3 βˆ™ π‘₯2 + 𝛽4 βˆ™ π‘₯3 + πœ€
  • 242.
  • 243. π‘Œ = 𝛽0 + 𝛽1 βˆ™ 𝑃 + 𝛽2 βˆ™ π‘₯ + πœ€ 𝐸 𝛽1 = 𝛽1
  • 244. π‘Œ = 𝛽0 + 𝛽1 βˆ™ 𝑃 + 𝛽2 βˆ™ π‘₯ + πœ€ Regress π‘Œ on 𝑃 and π‘₯ 𝐸 𝛽1 = 𝛽1
  • 245. π‘Œ = 𝛽0 + 𝛽1 βˆ™ 𝑃 + 𝛽2 βˆ™ π‘₯ + πœ€ Regress π‘Œ on 𝑃 and π‘₯ 𝐸 𝛽1 = 𝛽1
  • 246. π‘Œ = 𝛽0 + 𝛽1 βˆ™ 𝑃 + 𝛽2 βˆ™ π‘₯ + πœ€ Regress π‘Œ on 𝑃 and π‘₯ and 𝑍 𝐸 𝛽1 β‰  𝛽1
  • 247. π‘Œ = 𝛽0 + 𝛽1 βˆ™ 𝑃 + 𝛽2 βˆ™ π‘₯ + πœ€ Regress π‘Œ on 𝑃 and π‘₯ and 𝑍 𝐸 𝛽1 β‰  𝛽1
  • 251. π‘Œ = 𝛽0 + 𝛽1 βˆ™ 𝑃 + 𝛽𝑧 βˆ™ 𝑍 + πœ” 1.Is correlated with the regressor of interest 2.Is correlated with the error term
  • 252. π‘Œ = 𝛽0 + 𝛽1 βˆ™ 𝑃 + 𝛽𝑧 βˆ™ 𝑍 + πœ” 1.Is correlated with the regressor of interest 2.Is correlated with the error term
  • 253. π‘Œ = 𝛽0 + 𝛽1 βˆ™ 𝑃 + 𝛽𝑧 βˆ™ 𝑍 + πœ” 1.Is correlated with the regressor of interest 2.Is correlated with the error term
  • 254. π‘Œ = 𝛽0 + 𝛽1 βˆ™ 𝑃 + 𝛽𝑧 βˆ™ 𝑍 + πœ” 1.Is correlated with the regressor of interest 2.Is correlated with the error term
  • 256. X Y P
  • 257. 1. 𝐸 π‘Œ1 |𝑋 = 𝑐 β‰  𝐸 π‘Œ1 |𝑋 = π‘˜ 𝐸 π‘Œ0 |𝑋 = 𝑐 β‰  𝐸 π‘Œ0 |𝑋 = π‘˜ 2. π‘ƒπ‘Ÿ 𝑃 = 1|𝑋 = 𝑐 β‰  π‘ƒπ‘Ÿ 𝑃 = 1|𝑋 = π‘˜ 𝐸 𝑋|𝑃 = 1 β‰  𝐸 𝑋|𝑃 = 0
  • 258. 𝑋 = 0 if male 1 if female
  • 259. 𝐸 π‘Œ1 |𝑋 𝐸 π‘Œ0 |𝑋 𝐸 π‘Œ1 |𝑋 βˆ’ 𝐸 π‘Œ 0 |𝑋 𝑋 = 0 6 4 2 𝑋 = 1 5 1 4
  • 260. 𝐸 π‘Œ1 |𝑋 𝐸 π‘Œ0 |𝑋 𝐸 π‘Œ1 |𝑋 βˆ’ 𝐸 π‘Œ 0 |𝑋 𝑋 = 0 6 4 2 𝑋 = 1 5 1 4 𝐸 π‘Œ1 βˆ’ π‘Œ0 |𝑋 = 0 βˆ™ π‘ƒπ‘Ÿ 𝑋 = 0 +𝐸 π‘Œ1 βˆ’ π‘Œ0 |𝑋 = 1 βˆ™ π‘ƒπ‘Ÿ 𝑋 = 1 = 2 βˆ™ .5 + 4 βˆ™ .5 = 3
  • 261. 𝐸 π‘Œ1 |𝑋 𝐸 π‘Œ0 |𝑋 𝐸 π‘Œ1 |𝑋 βˆ’ 𝐸 π‘Œ 0 |𝑋 𝑋 = 0 6 4 2 𝑋 = 1 5 1 4 𝐸 π‘Œ1 |𝑋 = 0 βˆ™ π‘ƒπ‘Ÿ 𝑋 = 0 +𝐸 π‘Œ1 |𝑋 = 1 βˆ™ π‘ƒπ‘Ÿ 𝑋 = 1 = 6 βˆ™ .5 + 5 βˆ™ .5 = 5.5
  • 262. 𝐸 π‘Œ1 |𝑋 𝐸 π‘Œ0 |𝑋 𝐸 π‘Œ1 |𝑋 βˆ’ 𝐸 π‘Œ 0 |𝑋 𝑋 = 0 6 4 2 𝑋 = 1 5 1 4 𝐸 π‘Œ0 |𝑋 = 0 βˆ™ π‘ƒπ‘Ÿ 𝑋 = 0 +𝐸 π‘Œ0 |𝑋 = 1 βˆ™ π‘ƒπ‘Ÿ 𝑋 = 1 = 4 βˆ™ .5 + 1 βˆ™ .5 = 2.5
  • 263. 𝐸 𝑃 = 1|𝑋 𝐸 𝑃 = 0|𝑋 𝑋 = 0 .1 .9 𝑋 = 1 .5 .5 𝐸 𝑃 = 1|𝑋 = 0 βˆ™ π‘ƒπ‘Ÿ 𝑋 = 0 +𝐸 𝑃 = 1|𝑋 = 1 βˆ™ π‘ƒπ‘Ÿ 𝑋 = 1 = .1 βˆ™ .5 + .5 βˆ™ .5 = .3
  • 264. 𝐸 𝑃 = 1|𝑋 𝐸 𝑃 = 0|𝑋 𝑋 = 0 .1 .9 𝑋 = 1 .5 .5 𝐸 𝑃 = 1|𝑋 = 0 βˆ™ π‘ƒπ‘Ÿ 𝑋 = 0 +𝐸 𝑃 = 1|𝑋 = 1 βˆ™ π‘ƒπ‘Ÿ 𝑋 = 1 = .1 βˆ™ .5 + .5 βˆ™ .5 = .3
  • 265. 𝑛 = 1000 𝑖=1 𝑛 𝑋𝑖 𝑛 = 𝑖=1 1000 𝑋𝑖 1000 β‰ˆ 500 𝑖=1 𝑛 𝑃𝑖 𝑛 = 𝑖=1 1000 𝑃𝑖 1000 β‰ˆ 300
  • 266. 𝑛 = 1000 𝑖=1 𝑛 𝑋𝑖 𝑛 = 𝑖=1 1000 𝑋𝑖 1000 β‰ˆ .5 𝑖=1 𝑛 𝑃𝑖 𝑛 = 𝑖=1 1000 𝑃𝑖 1000 β‰ˆ 300
  • 267. 𝑛 = 1000 𝑖=1 𝑛 𝑋𝑖 𝑛 = 𝑖=1 1000 𝑋𝑖 1000 β‰ˆ .5 𝑖=1 𝑛 𝑃𝑖 𝑛 = 𝑖=1 1000 𝑃𝑖 1000 β‰ˆ .3
  • 268. 𝑗=1 300 π‘Œπ‘— 300 : 6 βˆ™ .16667 + 5 βˆ™ .8333 = 5.16652
  • 269. 𝑗=1 300 π‘Œπ‘— 300 : 6 βˆ™ .16667 + 5 βˆ™ .8333 = 5.16652 𝑗=1 700 π‘Œπ‘— 700 : 4 βˆ™ .6428 + 1 βˆ™ .3571 = 2.9286
  • 270. ?
  • 271. 𝑖 = 237 𝑋237 = 0 𝑃237 = 1 π‘Œ237 = π‘Œ237 1 = 5.3
  • 272. 𝑖 = 237 𝑋237 = 0 𝑃237 = 1 π‘Œ237 = π‘Œ237 1 = 5.3
  • 273. 𝑖 = 237 𝑋237 = 0 𝑃237 = 1 π‘Œ237 = π‘Œ237 1 = 5.3
  • 274. 𝑖 = 237 𝑋237 = 0 𝑃237 = 1 π‘Œ237 = π‘Œ237 1 = 5.3
  • 275. 𝑖 = 237 𝑋237 = 0 𝑃237 = 1 π‘Œ237 = π‘Œ237 1 = 6.3
  • 276. 𝐼237 = π‘Œ237 1 βˆ’ π‘Œ237 0 = 6.3βˆ’?
  • 282. π‘˜=0 𝐾 𝐸 𝐼|𝑋 = π‘˜ βˆ™ π‘ƒπ‘Ÿ 𝑋 = π‘˜ = π‘˜=0 𝐾 𝐸 π‘Œ1 βˆ’ π‘Œ0 |𝑋 = π‘˜ βˆ™ π‘ƒπ‘Ÿ 𝑋 = π‘˜
  • 283. π‘˜=0 𝐾 𝐸 𝐼|𝑋 = π‘˜, 𝑃 = 1 βˆ™ π‘ƒπ‘Ÿ 𝑋 = π‘˜|𝑃 = 1 = π‘˜=0 𝐾 𝐸 π‘Œ1 βˆ’ π‘Œ0 |𝑋 = π‘˜, 𝑃 = 1 βˆ™ π‘ƒπ‘Ÿ 𝑋 = π‘˜|𝑃 = 1
  • 284. 𝐼237 = π‘Œ237 1 βˆ’ π‘Œ237 0 = 6.3βˆ’?
  • 285. π‘˜=0 𝐾 𝐸 𝐼|𝑋 = π‘˜ βˆ™ π‘ƒπ‘Ÿ 𝑋 = π‘˜ = π‘˜=0 𝐾 𝐸 π‘Œ1 βˆ’ π‘Œ0 |𝑋 = π‘˜ βˆ™ π‘ƒπ‘Ÿ 𝑋 = π‘˜
  • 286. X Y P
  • 287.
  • 288. 2X Gender-male/female 5X SES-5 quintiles 4X Age-broken up into 4 categories 2X Sector-Urban/Rural 6X Religion-Christian, Muslim, Hindu, Buddhist, Jewish, Traditional. 6X Occupation-6 occupation types 2 Insurance-Insured/not insured ____ 5,760 potential β€œtypes”
  • 289. 2X Gender-male/female 5X SES-5 quintiles 4X Age-broken up into 4 categories 2X Sector-Urban/Rural 6X Religion-Christian, Muslim, Hindu, Buddhist, Jewish, Traditional. 6X Occupation-6 occupation types 2 Insurance-Insured/not insured ____ 2 potential β€œtypes”
  • 290. 2X Gender-male/female 5X SES-5 quintiles 4X Age-broken up into 4 categories 2X Sector-Urban/Rural 6X Religion-Christian, Muslim, Hindu, Buddhist, Jewish, Traditional. 6X Occupation-6 occupation types 2 Insurance-Insured/not insured ____ 10 potential β€œtypes”
  • 291. 2X Gender-male/female 5X SES-5 quintiles 4X Age-broken up into 4 categories 2X Sector-Urban/Rural 6X Religion-Christian, Muslim, Hindu, Buddhist, Jewish, Traditional. 6X Occupation-6 occupation types 2 Insurance-Insured/not insured ____ 40 potential β€œtypes”
  • 292. 2X Gender-male/female 5X SES-5 quintiles 4X Age-broken up into 4 categories 2X Sector-Urban/Rural 6X Religion-Christian, Muslim, Hindu, Buddhist, Jewish, Traditional. 6X Occupation-6 occupation types 2 Insurance-Insured/not insured ____ 80 potential β€œtypes”
  • 293. 2X Gender-male/female 5X SES-5 quintiles 4X Age-broken up into 4 categories 2X Sector-Urban/Rural 6X Religion-Christian, Muslim, Hindu, Buddhist, Jewish, Traditional. 6X Occupation-6 occupation types 2 Insurance-Insured/not insured ____ 480 potential β€œtypes”
  • 294. 2X Gender-male/female 5X SES-5 quintiles 4X Age-broken up into 4 categories 2X Sector-Urban/Rural 6X Religion-Christian, Muslim, Hindu, Buddhist, Jewish, Traditional. 6X Occupation-6 occupation types 2 Insurance-Insured/not insured ____ 2,880 potential β€œtypes”
  • 295. 2X Gender-male/female 5X SES-5 quintiles 4X Age-broken up into 4 categories 2X Sector-Urban/Rural 6X Religion-Christian, Muslim, Hindu, Buddhist, Jewish, Traditional. 6X Occupation-6 occupation types 2 Insurance-Insured/not insured ____ 5,760 potential β€œtypes”
  • 296. 2X Gender-male/female 5X SES-5 quintiles 4X Age-broken up into 4 categories 2X Sector-Urban/Rural 6X Religion-Christian, Muslim, Hindu, Buddhist, Jewish, Traditional. 6X Occupation-6 occupation types 2 Insurance-Insured/not insured ____ 5,760 potential β€œtypes”
  • 298. 𝐼237 = π‘Œ237 1 βˆ’ π‘Œ237 0 = 6.3βˆ’?
  • 299. 𝐼237 = π‘Œ237 1 βˆ’ π‘Œ237 0 = 6.3βˆ’?
  • 302. π‘ƒπ‘Ÿ 𝑃 = 1 𝑋 = 𝑒π‘₯𝑝 𝛽0 + 𝛽1 βˆ™ 𝑋 1 + 𝑒π‘₯𝑝 𝛽0 + 𝛽1 βˆ™ 𝑋
  • 303. The basics of the method 1. Pool your sample of participants and non-participants and define various characteristics 𝑋. 2. Estimate the probability of program participation conditional on Pr 𝑃 = 1|𝑋 3. Compute the propensity score using the fitted binary participation model. π‘ƒπ‘Ÿ 𝑃𝑖 = 1|𝑋𝑖 4. Find a counterfactual outcome for each individual by identifying some individual who did experience the counterfactual conditions and had a similar propensity score
  • 304. Matching approaches 1. Nearest neighbor 2. Caliper 3. (Budget) Caliper 4. Weighting
  • 305. Matching approaches 1. Nearest neighbor 2. Caliper 3. (Budget) Caliper 4. Weighting
  • 309. Other applications of the propensity score 1.Weighting 2.Regression
  • 310. Other applications of the propensity score 1.Weighting 2.Regression
  • 311. 𝐸 𝑃 = 1|𝑋 𝐸 𝑃 = 0|𝑋 𝑋 = 0 .1 .9 𝑋 = 1 .5 .5 𝐸 𝑃 = 1|𝑋 = 0 βˆ™ π‘ƒπ‘Ÿ 𝑋 = 0 +𝐸 𝑃 = 1|𝑋 = 1 βˆ™ π‘ƒπ‘Ÿ 𝑋 = 1 = .1 βˆ™ .5 + .5 βˆ™ .5 = .3 1,000 observations
  • 312. 𝐸 𝑃 = 1|𝑋 𝐸 𝑃 = 0|𝑋 𝑋 = 0 .1 .9 𝑋 = 1 .5 .5 𝐸 𝑃 = 1|𝑋 = 0 βˆ™ π‘ƒπ‘Ÿ 𝑋 = 0 +𝐸 𝑃 = 1|𝑋 = 1 βˆ™ π‘ƒπ‘Ÿ 𝑋 = 1 = .1 βˆ™ .5 + .5 βˆ™ .5 = .3 1,000 observations 300 participants
  • 313. 𝐸 𝑃 = 1|𝑋 𝐸 𝑃 = 0|𝑋 𝑋 = 0 .1 .9 𝑋 = 1 .5 .5 𝐸 𝑃 = 1|𝑋 = 0 βˆ™ π‘ƒπ‘Ÿ 𝑋 = 0 +𝐸 𝑃 = 1|𝑋 = 1 βˆ™ π‘ƒπ‘Ÿ 𝑋 = 1 = .1 βˆ™ .5 + .5 βˆ™ .5 = .3 1,000 observations 300 participants 700 non-participants
  • 314. 𝐸 𝑃 = 1|𝑋 𝐸 𝑃 = 0|𝑋 𝑋 = 0 .1 .9 𝑋 = 1 .5 .5 𝐸 𝑃 = 1|𝑋 = 0 βˆ™ π‘ƒπ‘Ÿ 𝑋 = 0 +𝐸 𝑃 = 1|𝑋 = 1 βˆ™ π‘ƒπ‘Ÿ 𝑋 = 1 = .1 βˆ™ .5 + .5 βˆ™ .5 = .3 1,000 observations 300 participants 700 non-participants 50 men 250 women
  • 315. 𝐸 𝑃 = 1|𝑋 𝐸 𝑃 = 0|𝑋 𝑋 = 0 .1 .9 𝑋 = 1 .5 .5 𝐸 𝑃 = 1|𝑋 = 0 βˆ™ π‘ƒπ‘Ÿ 𝑋 = 0 +𝐸 𝑃 = 1|𝑋 = 1 βˆ™ π‘ƒπ‘Ÿ 𝑋 = 1 = .1 βˆ™ .5 + .5 βˆ™ .5 = .3 1,000 observations 300 participants 700 non-participants 50 men 250 women 450 men 250 women
  • 316. 𝐸 𝑃 = 1|𝑋 𝐸 𝑃 = 0|𝑋 𝑋 = 0 .1 .9 𝑋 = 1 .5 .5 𝐸 𝑃 = 1|𝑋 = 0 βˆ™ π‘ƒπ‘Ÿ 𝑋 = 0 +𝐸 𝑃 = 1|𝑋 = 1 βˆ™ π‘ƒπ‘Ÿ 𝑋 = 1 = .1 βˆ™ .5 + .5 βˆ™ .5 = .3 1,000 observations 300 participants 700 non-participants 500 men 500 women 450 men 250 women
  • 317. 𝐸 𝑃 = 1|𝑋 𝐸 𝑃 = 0|𝑋 𝑋 = 0 .1 .9 𝑋 = 1 .5 .5 𝐸 𝑃 = 1|𝑋 = 0 βˆ™ π‘ƒπ‘Ÿ 𝑋 = 0 +𝐸 𝑃 = 1|𝑋 = 1 βˆ™ π‘ƒπ‘Ÿ 𝑋 = 1 = .1 βˆ™ .5 + .5 βˆ™ .5 = .3 1,000 observations 300 participants 700 non-participants 500 men 500 women 500 men 500 women
  • 318. π‘Šπ‘Œπ‘– = 𝑃𝑖 βˆ™ π‘Œπ‘– π‘ƒπ‘Ÿ 𝑃𝑖 = 1|𝑋𝑖 βˆ’ 1 βˆ’ 𝑃𝑖 βˆ™ π‘Œπ‘– 1 βˆ’ π‘ƒπ‘Ÿ 𝑃𝑖 = 1|𝑋𝑖 𝐴𝑇𝐸 = 𝑖=1 𝑛 π‘Šπ‘Œπ‘– 𝑖=1 𝑛 𝑃𝑖 π‘ƒπ‘Ÿ 𝑃𝑖 = 1|𝑋𝑖 + 1 βˆ’ 𝑃𝑖 1 βˆ’ π‘ƒπ‘Ÿ 𝑃𝑖 = 1|𝑋𝑖
  • 319. π‘Šπ‘Œπ‘– = 𝑃𝑖 βˆ™ π‘Œπ‘– π‘ƒπ‘Ÿ 𝑃𝑖 = 1|𝑋𝑖 βˆ’ 1 βˆ’ 𝑃𝑖 βˆ™ π‘Œπ‘– 1 βˆ’ π‘ƒπ‘Ÿ 𝑃𝑖 = 1|𝑋𝑖 𝐴𝑇𝐸 = 𝑖=1 𝑛 π‘Šπ‘Œπ‘– π’Š=𝟏 𝒏 π‘·π’Š 𝑷𝒓 π‘·π’Š = 𝟏|π‘Ώπ’Š + 𝟏 βˆ’ π‘·π’Š 𝟏 βˆ’ 𝑷𝒓 π‘·π’Š = 𝟏|π‘Ώπ’Š
  • 320. π‘Šπ‘Œπ‘– = 𝑃𝑖 βˆ™ π‘Œπ‘– π‘ƒπ‘Ÿ 𝑃𝑖 = 1|𝑋𝑖 βˆ’ 1 βˆ’ 𝑃𝑖 βˆ™ π‘Œπ‘– 1 βˆ’ π‘ƒπ‘Ÿ 𝑃𝑖 = 1|𝑋𝑖 𝐴𝑇𝐸 = 𝑖=1 𝑛 π‘Šπ‘Œπ‘– 𝑖=1 𝑛 𝑃𝑖 π‘ƒπ‘Ÿ 𝑃𝑖 = 1|𝑋𝑖 + 1 βˆ’ 𝑃𝑖 1 βˆ’ π‘ƒπ‘Ÿ 𝑃𝑖 = 1|𝑋𝑖
  • 321. Other applications of the propensity score 1.Weighting 2.Regression
  • 322. Regress π‘Œπ‘– on 𝑃𝑖 and π‘ƒπ‘Ÿ 𝑃𝑖 = 1|𝑋𝑖
  • 323. Regress π‘Œπ‘– on 𝑃𝑖 and π‘ƒπ‘Ÿ 𝑃𝑖 = 1|𝑋𝑖
  • 325. Links: The manual: http://www.measureevaluation.org/resources/publications/ms- 14-87-en The webinar introducing the manual: http://www.measureevaluation.org/resources/webinars/metho ds-for-program-impact-evaluation My email: pmlance@email.unc.edu
  • 326. MEASURE Evaluation is funded by the U.S. Agency for International Development (USAID) under terms of Cooperative Agreement AID-OAA-L-14-00004 and implemented by the Carolina Population Center, University of North Carolina at Chapel Hill in partnership with ICF International, John Snow, Inc., Management Sciences for Health, Palladium Group, and Tulane University. The views expressed in this presentation do not necessarily reflect the views of USAID or the United States government. www.measureevaluation.org