Relationships Among PrePost Designs

1
Relationships Among Pre-Post Designs
2
Pre-Post Designs
O X O O O

• randomized experiment (R)
• nonequivalent group design (N)
• regression-discontinuity design (C)

3
The Main Issue

• selection threats
• are the groups similar before the treatment?
• related to assignment method in each design

4
Assignment Method
Probability of Assignment to Treatment Given the
Pretest
1.0
.5
0
Pretest
5
Assignment Method
Probability of Assignment to Treatment Given the
Pretest
1.0
RE
.5
0
Pretest
6
Assignment Method
Probability of Assignment to Treatment Given the
Pretest
1.0
RE
.5
RD
0
Pretest
7
Assignment Method
Probability of Assignment to Treatment Given the
Pretest
1.0
RE
.5
NEGD
RD
0
Pretest
8
Assignment Method
Probability of Assignment to Treatment Given the
Pretest
1.0
RE
.5
NEGD
RD
0
Pretest
9
Pretest Equivalence
Maximally Nonequivalent
Equivalent
10
Pretest Equivalence
RE
Maximally Nonequivalent
Equivalent
11
Pretest Equivalence
RD
RE
Maximally Nonequivalent
Equivalent
12
Pretest Equivalence
RD
NEGD
RE
Maximally Nonequivalent
Equivalent
13
Assignment Rule Known

Randomized Experiment
NonEquivalent Group Design
Regression-Discontinuity
14
(No Transcript)