7
Of course, to have this confidence requires that the design is
implemented well. Evaluators must ensure that: 1) assignment
to the treatment and control groups is actually random (e.g., de-
ciding to put all high-risk offenders in the treatment group, but
randomly putting others in the treatment and control groups is
not random assignment), 2) the control group does not receive
the intervention, and 3) the evaluation examines outcomes for
everyone, even those in the treatment group who don’t finish
the intervention. A good evaluation report on
an experimental design should discuss these
issues.
Quasi-Experimental Designs: Common in
Criminal Justice
The major difference between experimental
designs and quasi-experimental designs
relates to the use of a control group. Quasi-
experimental designs may or may not use a
control group (usually called comparison
groups in quasi-experiments). If a quasi-experi-
ment uses a control group, individuals/cases
will not be placed randomly into the groups.
The comparison group simply consists of a
group of individuals/cases that are considered
similar to those who received the treatment.
The evaluator may attempt to ensure compa-
rability between the two groups by matching the individuals in
the groups on factors that are considered relevant, such as age,
gender, or prior history with the criminal justice system. For ex-
ample, if gender and prior criminal justice history are considered
relevant factors, one way to match is to ensure that the treatment
and comparison groups have similar proportions of males and
subjects without previous arrests. If one group had more indi-
viduals without previous arrests, then its members may be less
likely to commit crime than the members of the other group.
The more confident we can be that the two groups are similar in
all key characteristics other than program participation, the
Given the program or policy
under study, it may be diffi-
cult to identify an appropri-
ate comparison or control
group. In criminal justice
this is particularly the case
with programs/policies that
affect an entire community
or jurisdiction. Keep this in
mind when considering the
strength of the design. Always
look for an explanation of
why the evaluator chose a
particular research design.