Single Subject/Small-N Designs
Traditional Group Design Are widely used Can address a broad range of questions Requires a large number of homogenous participants that can be randomly assigned Compares two or more groups in terms of effect on target behaviors
Single-Subject/ Small-N Designs Permits experimental analysis with single or minimal subjects Involves the study of one or more individuals across repeated observations before, during, and following the introduction of an intervention Often look for trends and variability in graphed data
General Requirements for Small-N Baseline Assessment Continuous Assessment Repeated Measures Stability of Performance Trend Variability
Control Group Design One person/group gets the intervention while the other doesn’t Okay with large numbers but hard to say apples to apples with small numbers Ethical issues withholding treatment. Ensure you have time to correct either group as needed with proper treatment
AB Design Get baseline data without the intervention then implement the intervention and look for a change Non-experiment design – next to impossible to control for variables i.e. “How do you know the person wasn’t just going to get better anyway regardless of the intervention?” Example – pre/post test design
ABAB Design Get baseline data without the intervention then implement the intervention and look for a change. Take away the intervention to see if the subject(s) revert back to the baseline then add the intervention again to see if it has the same effect Most important feature is that it includes a direct replication effect NOTE = be careful of ethical issues taking away treatment
Criterion Design Behavior improves in increments to match a criterion for performance Several subphases are used in intervention After the baseline, a criterion is set for behavior, and when behavior matches the criterion (and stabilizes) the criterion is changed Common in behavior modification: sticker for hour, reward for day, reward for week Disadvantage = gradual improvement not clearly connected to shifts in the criterion Disadvantage = when there is a lack of withdrawl, it’s harder to determine the true role of the intervention
Multiple Treatment Design After a stable baseline is established, two treatments are administered, alternating with each other Treatments do not occur at the same time Can be done with single subject, but stronger conclusions can be made when multiple subjects are used in different orders
Multiple Treatment Design Jack intervention B Jill intervention A Jack intervention A Jill intervention B
Multiple Baseline Design AB Design but multiple subjects receive the intervention a increasingly delayed intervals
Reliability - Interobserver Agreement Multiple people independently watch the same person(s) and record their findings After the observation period, calculate the level of agreement between observers Types of Calculations Percentage of Agreement Cohen’s Kappa Intercorrelational coefficient Generalizability using a two way ANOVA