Evaluation Plan Guidance Page 61
ADVERTISEMENT
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68EVALUATION PLAN GUIDANCE
SOCIAL INNOVATION FUND
General Design and Description
Matching Type
Description
Relation to Validity
less need assigned to the comparison group. The
difference (or discontinuity) in the average outcome
between those in the intervention (program participant)
group just below the cutoff score (e.g. those with lower
test scores) and those in the comparison group just above
the cutoff score (those with higher test scores) can be
attributed to the program, albeit with reservations.
Propensity Score
Propensity scores are based on multiple characteristics of
Propensity score matching controls for selection bias
many individuals, and allow for a more accurate
and other biases related to group assignment (although
matching between program participants and control
not perfectly). Provided both groups are drawn from
group members. The scores represent how likely any
the target population and are generally representative
individual is to be in the treated (program participant
of it, this form of matching may lead to generalizable
group) based on known characteristics of individuals in
results.
each group. Scores are calculated for the program
participation group and for the comparison group, and
the composition of the comparison group is generated to
best resemble the program participation group.
Control Group
Random
Random assignment specifies that control groups are
Provided study participants are drawn from the target
Assignment
created when individuals, groups, or sites are randomly
population and are generally representative of it, this
The non-program participation
group is selected randomly along
assigned to either the treatment (program participation)
form of matching leads to generalizable results.
with the program participant group.
group or to a control group (that does not receive
However, program participation must not distort the
Several techniques for randomly
treatment). Random assignment can take several forms,
situation under examination greatly. Additionally,
assigning people (or groups or sites)
ranging from simple random assignment in which each
external validity is boosted when multiple sites are
exist.
person (or group or site) has an equal probability of being
used in the study.
selected for program participation to more complicated
schemes involving stratification in which characteristics
of the population are taken into account.
C.12
ADVERTISEMENT
0 votes
Related Articles
Related forms
Related Categories
Parent category: Education