CRIM 220 Lecture Notes - Lecture 5: Dependent And Independent Variables, Repeated Measures Design, Participant Observation
Lecture 6, Week 7: Chapter 7 Tuesday 27th February
Crim 220: Quasi-Experimental and Non-Experimental Designs
• Variations on the classical experimental designs are especially useful for explanatory
research and in the evaluation studies
Quasi-Experimental Designs
• When random assignment is not possible or the IV is not manipulated
• There are more internal validity concerns
• Could be due to practical or administrative reasons, or legal or ethical reasons
Requirements:
1. comparison group
2. multiple measures
Can be grouped into two design categories:
1. non-equivalent group designs
2. time-series design
•No QE design is immune to alternative explanations for difference between
treatment and control groups
•True experimental designs are stronger
Non-Equivalent group designs
• Groups are constructed in some non-random procedure, however it cannot be
assumed that the groups are statistically equivalent
• Groups are selected in a way to make them as comparable as possible
Attempt to minimize differences through matching, comparison
cohorts, etc.
Groups should be comparable in terms of variables that are related to
the DV under study
Static Group Comparison design
•Experimental Group is measured after being exposed to the experimental
treatment
•Control group is measured without having been exposed to the experimental
treatment
•No pre-measure is taken
•No way of knowing if they are equivalent groups (can be for both IV and DV)
Individual Matching
•the two groups should be comparable in terms of variables likely to be related to the
DV
What might be those?
gender
31
Quasi-Experimental: research design that includes most, but not all, elements
of an experimental design.
racial structures
socio-economic status
age
•First three known to be correlated with juvenile and adult arrest rate
•Age = time at risk
Example:
•Children who have been abused, were more likely to have juvenile or adult arrest
records, than were children who had not been abused (presumed)
Natural Experiments
•Studies done with people who had something happen to them
the nature of this event creates the treatment and control groups
•Events are not created by researchers
Examples:
•Released NCRMD (NGRI) patients → later offending
•Car crash survivors → PTSD
Found Groups
Pretest-Posttest Non equivalent Control Group Design
• Assignment of the participants is not controlled by the investigator
• A treatment group and a comparison group are used
•Pretest and post-test measures are taken
Pretest-Posttest Multiple Non equivalent Control Group Design
•Various areas of comparison
•Areas similar on socio-demographic, geographical, crime patterns
•CCTV likely not the cause of reduced crime (it went down everywhere, as did fear,
likely as a result of less crime)
Aggregate Matching
•Researchers select a control group with the same general composition of relevant
characteristics as treatment group
•Different areas of the state
•Assumes areas are similar with respect to relevant variables ‘on the average’
One group Pretest-Posttest
•No comparison group at all
•Something happened or a treatment is given
•The before and after are measured
•Allows to show that scores on the DV changed after the treatment or event occurred
Cohort
• Group of subjects who enter or leave an institution at the same time
Example: a class of police officers who graduate from a training
academy at the same time
All persons who were sentenced to probation in May vs those who
enter in January
OR
32
Document Summary
Tuesday 27th february: variations on the classical experimental designs are especially useful for explanatory research and in the evaluation studies. Quasi-experimental designs: when random assignment is not possible or the iv is not manipulated, there are more internal validity concerns, could be due to practical or administrative reasons, or legal or ethical reasons. Can be grouped into two design categories: non-equivalent group designs. Quasi-experimental: research design that includes most, but not all, elements of an experimental design: no qe design is immune to alternative explanations for difference between treatment and control groups, true experimental designs are stronger. Non-equivalent group designs: groups are constructed in some non-random procedure, however it cannot be assumed that the groups are statistically equivalent, groups are selected in a way to make them as comparable as possible. Attempt to minimize differences through matching, comparison cohorts, etc. Groups should be comparable in terms of variables that are related to the dv under study.