RESEARCH LAB 1 (S3) Salem State College School of Social Work

Research in Social Work
Practice
Salem State University
School of Social Work
Class 6
Jeff Driskell, MSW, PhD
2
Class Agenda
• Check-in/Announcements
• Review
▫ Think…Pair…Share
• Lecture▫ Theory in Research
▫ Quantitative approaches to research
▫ Internal/external validity
Elements of Quantitative Design
• Variables
▫ IV and DV
• Relationships and
comparisons
▫ Correlation
▫ comparisons
• Theory Driven
• Testable Hypotheses
▫ Directional or nondirectional
• Methodology
▫ Sampling
▫ Data collection
• Design structure
• Cross-sectional, longitudinal
• Experimental, quasi
• Statistical analyses
Quantitative Research
Goal: An attempt to determine if there is a
relationship between identified variables.
▫ Correlation ( a link between two or more
variables)
▫ Comparison (comparing groups on a particular
variable(s)
▫ Causal (one variable causing change in another)
Criteria for Inferring Causality
1) Cause (IV) must precede the effect (DV) in
time (temporal order)
2) The two variables are empirically correlated
with one another (statistically
determined)
3) The correlation between the two variables
can not be due to the influence of a third
variable
Quantitative Research Design
Quantitative Research Design
▫ Experimental




True randomized trial
Experimental and control group
Establishes cause and effect
Most common- pre-test/post-test design
▫ Quasi-experimental
 Comparison/control group- may not be comparable
due to a lack of random assignment
▫ Non-experimental designs (Pre)
 No cause and effect
 Looks at relationships and comparisons
Breaking it Down…
• Non-experimental (pre-experimental)
▫ One-group post test only design
▫ One-group pre/post test design
• Quasi-experimental
▫ Time series design (no comparison group)
▫ Time series design with comparison group
▫ Post-test only with non-equivalent group design
• Experimental
▫ Pre-test/post test control group design
Mapping research designs
…not tic-tac-toe…
• R = Randomization
• O = Observation (survey, quiz)
• X = Intervention/Stimulus (counseling, lecture)
• ------ = more than one group in study
Experimental Designs
 Experimental designs are the strongest
designs allowing social work researchers and
evidence-based practitioners to have
increased confidence in making causal
inferences based of study findings
Random Assignment
 Controls for selection bias in experimental
designs
 Participants are divided into groups using
procedures based on probability theory
 Improves the likelihood that the control
group represents what the experimental
group would look like had it not been
exposed to the experimental stimulus
Diagram of a Basic Experimental Design
R
R
O1
O1
X
O2
O2
©2011, Brooks/ Cole Publishing, A Division of Cengage
Learning, Inc.
Decision Tree
Example- Project Enhance
Research Design Application
Research design 101
• Cross-sectional
• 01
• Example:
– Caregiver views on a
support group for parents
with children in DSS
custody
30%
50%
20%
Very helpful
Not helpful
Sort of helpful
Research design 101
70
60
50
Percent
• Multistage
crosssectional,
panels of
different
people
• 01-01-01-01
40
30
20
10
0
January
March
June
Has health insurance
September
• Pre/post test:
• 01-X1-02
• Example:
– Evaluation of a
treatment compliance
intervention
Percent
Research design 101
90
80
70
60
50
40
30
20
10
0
May '06
Dec. '06
Year
Compliant
Non-compliant
• Post-test only
• X1-01
• Example:
– Policy analysis of
employment among
single mothers after
leaving TANF
Percent
Research design 101
90
80
70
60
50
40
30
20
10
0
Post-TANF
Not working
Working
Research design 101
•
Evaluation of longterm
psychotherapeutic
treatment with same
group over time
60
50
Percent
• Time series,
longitudinal
• 01-X1-02-X203-X3-04
40
30
20
10
0
Year 1
Year 2
Year 3
General life satisfaction score
Year 4
One-Shot Case Study
X
0
Fails to control for any threats to internal validity
One Group Pretest-posttest Design
01
X
O2
Quasi-experimental Designs
 Designs that are distinguished from true experiments
primarily by the lack of random assignment of subjects
 Useful when it is not feasible to obtain a control group
 Three common quasi-experimental designs :
− Nonequivalent comparison group designs
− Simple time-series designs
− Multiple time-series designs
Nonequivalent Comparison
Groups Designs
 Two existing groups that appear to be similar are
identified or created (homogenous)
 The dependent variable is assessed before and
after an intervention is introduced to one of the
groups
 Comparison group (not to be confused with
control group) does not receive the intervention
O1
O1
X
O2
O2
Simple Time-Series Designs
 A simple interrupted time-series design attempts
to develop causal inferences based on a
comparison of trends over multiple measurements
before and after an intervention is introduced
and requires no comparison group
O1 O2 O3 O4 O5 X O6 O7 O8 O9 O10
 The more measurements, the stronger the design
Multiple Time-Series Designs
 Both an experimental group and a comparison
group are measured at multiple points in time
before and after an intervention is introduced
to the experimental group
O1 O2 O3 O4 O5 X O6 O7 O8 O9 O10
O1 O2 O3 O4 O5
O6 O7 O8 O9 O10
Exercise- Putting it all together
• Locate 2 quantitative based research articles
• Answer the following question:
▫
▫
▫
▫
▫
▫
Identify the research question(s)
What are the IV and DV’s?
Identify hypotheses
Identify type of design
Map out the research design (R, O, X).
Internal and external validity. What are the
concerns for both types of validity (think about the
terms you just learned about)?