Slides to accompany Weathington, Cunningham & Pittenger (2010), Chapter 15: Single-Participant Experiments, Longitudinal Studies, and Quasi-Experimental Designs 1
Objectives Single-N experiments Types of Single-N designs Longitudinal design Types of longitudinal designs Quasi-experiments Types of quasi-experiments 2
Single-N Experiments Consider how IVs influence a participant –Commonly focused on behavioral outcomes Fechner, Ebbinghaus, and Skinner all used these techniques –Experimental analysis of behavior Used when examining effects of systematic changes in environment 3
Single-N: Reliability & Validity Sample size does not determine the validity of a study External validity depends on the type of generalization –When variability is expected to be low, large N not required for external validity (e.g., Figure 15.1) –Internal validity is about variable relationships (cause and effect), which can be observed with few participants 4
When to Use Single-N When the researcher has: –Direct control over IV –Ability to regularly measure participant’s behavior –Ability to observe over a long period of time e.g., Studying effect of specific environmental change on behavior –Clinical interventions 5
When Not to Use Single-N Trying to define a population Examining differences among populations When IV is not fully controlled by the experimenter 6
Types of Single-N Designs Baseline study –Changes in ongoing behaviors –Requires reliable behavior measure –Baseline measurement = control Discrete trial study –Response of participant to specific test conditions 7
Single-N Cause and Effect 1.Establish baseline What is the typical behavior to change? Offers a sort of “control” condition 2.Examine effects of intervention ABAB reversals 3.Replicate Follow same procedure with multiple folks Inter- and intra-person replication (ABAB) 8
Multiple Baseline Design Alternative to reversal designs (ABAB) Ongoing measurement of behavior, systematic introduction of the IV at different times Multi-baseline across participants, behaviors, or situations See Figure
Figure
Longitudinal Design For studying how behavior changes over time Requires monitoring sample over time Good for isolating cause and effect relationships Expensive and challenging –Attrition is a problem –Cohort effect 11
Cross-Sectional Sequential Design Good for developmental transition studies Does not take as much time as a full longitudinal study Can study groups of people from different age ranges, over time See Figure
Figure
Survival Analysis Alternative to: –Correlated groups ANOVA (which may have overly restrictive assumptions for your data) –Longitudinal design (which carry a high attrition risk) Time between events is a DV, not IV –Goal is to determine how long it takes an event to occur Can correct for attrition and still provide results that can be validly interpreted 14
Quasi-Experiments Useful if true experiment is impractical or unethical Uses an IV and DV and a control group Lacks random assignment to groups –Cannot rule out all alternative explanations Several forms of designs can be considered quasi-experiments 15
Nonequivalent Control-Group Two, pre-existing groups Researcher determines which gets the IV and which is the control group Pre-/post- measure of DV for both groups Main threats to validity: –History –Regression to the mean –Instrumentation 16
Interrupted Time Series Repeated measures of behavior in a sample pre- /post- a critical event Cannot easily rule out alternative explanations –Better if a control group available (see Figure 15.7 for example) 17
Figure
What is Next? **instructor to provide details 19