SSRD
Xorian
5313_Module3_SSRD.pptxSingle System Research Designs
Holli Slater, PhD
School of Social Work
1
The Research Process
Step 1
Step 4
Step 7
2
Formulate the problem
Review the Literature
Design the Study
Collect the Data
Analyze the Data
Interpret Findings
Write Research Report
Experimental vs Single System
Experimental designs: Consists of a sample of multiple cases
Experimental
Pretest-posttest control group design
Posttest-only control group design
Solomon four-group design
Quasi-Experimental
Time-Series
Cross-Sectional
Longitudinal
Non-equivalent comparison groups
Pre-Experimental
Pilot studies
One-shot case study
One-group pretest-posttest design
Post-test only design with nonequivalent groups
Single System: Consists of studying a single case
3
SSRD Sample Size and Unit of Analysis
Sample size of one (N = 1)
Unit of analysis
Individual
Incident, phenomenon
Group, family, community
4
SSRD Design Considerations
Identify the appropriate target problem
Operationally Define it
Triangulation
Determine appropriate intervention
5
Step 1: Identify Target Client, Problem, and Intervention
Identify your subject/client
Must be looking for a change in something
Identify your target problem
Be sure you and the client agree on target problem
How are you going to measure it?
Will you use more than one measure?
Determine appropriate intervention
Look at the literature to guide your selection
6
SSRD Designs
Terminology in SSRD
Phases
Notation
Graphs
Questions in SSRD
Evaluation - Did the client systems improve during the intervention?
Experimentation - Did the client system improve BECAUSE of the intervention?
7
Sample SSRD Design: Generic
8
Days
Behavior
Sample AB Design: Anxiety Management with Play Therapy Intervention
9
Days
Level of Anxiety
Sample AB Design: Anxiety Management with Cognitive Behavioral Therapy Intervention
10
Days
Level of Anxiety
Baseline(A Phase)
Before the introduction or withdrawal of the intervention (independent variable)
Baseline phase behaviors compared to intervention phase behaviors
Extended until a stable trend in the data are evident
11
SSRD Baseline Phase
Observations gathered pre-intervention
Many observations points (5-10 ideally)
Stable pattern in desired direction (if anxiety is high, it remains high throughout the baseline)
12
Baselines
| Day of Week | # Times Participated in Class |
| Monday | |||| |
| Wednesday | | |
| Monday | ||| |
| Wednesday | || |
| Monday | |||| |
Purpose of Baselines
Establish base-rate of target behavior
Determine magnitude of problem in its “natural” state
Demonstrate whether intervention is making a difference
Without a baseline – no internal validity
Types of Baselines
Concurrent/prospective
Reconstructed/retrospective
13
How Long is Baseline Phase?
Must establish
Utility
Stability
How many data points?
Ideally 5-10 data points
At least 3 data points
Enough data points to establish trend
Not too many data points that target behavior begins to change naturally
14
Types of Single System Research Designs
Take a moment and take a deep breath. We are about to go into the different types of research designs.
15
Evaluation: The Baseline
Baseline Only – No Intervention
Assessing to determine if there is a problem or severity of the problem
16
Evaluation: The B Design
NO Baseline – Only Intervention
Assessment of Outcome takes place in one phase
Limitations
Cannot claim change was due to intervention
Cannot claim change was due to other alternative hypotheses
17
Evaluation: The AB Design
Includes Baseline Phase and Intervention Phase
A phase – Baseline - No attempt to effect change
B phase – Intervention
Assessment of outcome takes place in two phases
Limitations
Time to collect baseline data
Difficult to claim change was due to intervention
18
Experimental SSRD Designs
Purpose
To help maximize internal validity
Control for rival hypotheses
Types of Experimental Designs
Reversal/Withdrawal
Multiple Baselines
Changing Criterion
Multiple Treatments
19
Internal Validity - Causality
X caused Y
The confidence with which the researcher can assert that an observed change was caused by a prior event, phenomenon, intervention, etc.
20
Sample AB Design: Anxiety Management with Play Therapy Intervention with Downward Trend
21
Days
Level of Anxiety
The ABAB Design
How it works
Baseline (A)
Intervention (B)
Remove Intervention (A)
Reintroduce Intervention (B)
Purpose
To establish that a behavior change observed during the intervention phase occurs only when the intervention is present
Predict – Verify - Replicate
Limitations
Not willing to withdraw effective intervention
Improvements my not be reversible during removal of intervention phase
22
When To Use ABAB
When a clear relationship between the independent and dependent variables needs to be demonstrated
When the target behavior can be reversed
When withdrawal of treatment does not cause harm or conflict with ethics
23
The ABCD Design (Changing Conditions)
How it works
A= Baseline
B = Intervention 1 (I1)
C = Discontinue Intervention 1 (I1) and Introduce Intervention 2 (I2)
D = Discontinue Intervention 2 (I2) and Introduce Intervention 3 (I3)
Purpose
To make changes if original intervention is not working
To determine which, among a variety of interventions, is effective
24
The ABCD Design
Baseline
Intervention #1
Stop Intervention #1 and Start Intervention #2 OR Modify Intervention #1
Stop Intervention #2 and Start Intervention #3 OR Modify Intervention #2
Multiple Baseline Designs
How it works
Repeated AB designs stacked on top of one another
Across Clients: AB with multiple clients
Across Settings: AB with multiple settings
Across Problems: AB with multiple problems
Purpose
Allows research to replicate the effects of the IV without withdrawing the treatment
Limitations
Possible covariance
Functional relationships may not be clearly demonstrated
Time consuming and requires substantial resources
26
Multiple Baseline Design: Across Clients
Morgan, D. L., & Morgan, R. K. (2009). Multiple-baseline designs. In Single-case research methods for the behavioral and health sciences. (pp. 125-161). Thousand Oaks, CA: SAGE Publications, Inc.
27
Multiple Baseline Design: Across Settings
Morgan, D. L., & Morgan, R. K. (2009). Multiple-baseline designs. In Single-case research methods for the behavioral and health sciences. (pp. 125-161). Thousand Oaks, CA: SAGE Publications, Inc.
28
Multiple Baseline Design: Across Problems
Morgan, D. L., & Morgan, R. K. (2009). Multiple-baseline designs. In Single-case research methods for the behavioral and health sciences. (pp. 125-161). Thousand Oaks, CA: SAGE Publications, Inc.
29
Step 2: Identify Your Design
What type of SSRD will you have?
B Only or AB or ABA or ABCD or Multiple Baselines
Will you have a baseline phase?
What type of baseline will you have?
How long is your baseline phase going to last?
When will your baseline be collected?
How feasible is the design you are selecting?
Consider time constraints and resources.
What are the limitations?
30
Inferring Causality in SSRD
Timing of behavioral change
occurs after the intervention OR
co-occurs with intervention
Repeated co-occurrence of intervention and desired behavioral change (when intervention present, behavior change occurs)
Behavior change only occurs in presence of intervention
Consistent change over time
Relationship between behavioral change and intervention makes empirical and theoretical sense
31
Threats to Internal Validity: Alternative Explanations of Change – History & Maturation
History
Unaccounted for events that may affect the dependent variable during the course of the research
Maturation/Passage of Time
Naturally occurring mental or physical changes in participants over the course of the study
32
Threats to Internal Validity: Alternative Explanations of Change – Instrumentation & Testing
Instrumentation
Measurement bias that affects outcomes
Instrument fails to measure consistently
Testing
Testing experience affects outcome variable
Pre-test affecting post-test scores
33
Threats to Internal Validity: Alternative Explanations of Change – Statistical Regression and Experimental Mortality
Statistical Regression
Tendency of extreme scores to move toward the mean score upon retesting
Severe behaviors fall to average
Experimental Mortality
