Overview
This lecture completes part three of experimental design, covering multiple baseline designs, their variations, changing criterion designs, experimental analyses, and internal vs. external validity, with practical examples and review questions.
Multiple Baseline Designs
- Multiple baseline design tests one intervention across different behaviors, settings, or participants (dependent variables).
- Experimental control is demonstrated by replicating AB relationships across multiple dependent variables.
- Stacked graphs and stair-step phase change patterns are visual indicators of this design.
- Baselines are staggered to show prediction and verification; extending baseline in non-intervention variables demonstrates verification.
- Stronger experimental control is shown with more replications, but this design is weaker than reversal or alternating treatments because the intervention is not withdrawn.
- Best used for irreversible behaviors or when behavior cannot be unlearned.
- Not ideal for severe/dangerous behaviors due to ethical concerns with extended baselines.
Variations of Multiple Baseline Designs
- Multiple probe design: Probes (occasional tests) are used instead of continuous data collection, suitable for irreversible behaviors.
- Delayed multiple baseline: Baselines for some variables begin later; used when participants become available at different times.
- Non-concurrent multiple baseline: Baselines for variables or participants do not overlap in time and occur independently.
Changing Criterion Design
- Used to gradually change an existing behavior already in the individual's repertoire.
- Experimental control is shown when the behavior's level closely matches the changing goal line or criterion.
- More and varied criterion jumps, and bidirectional changes (returning to previous criteria), enhance control.
- Not recommended for severe/dangerous behaviors.
Complex Experimental Designs
- Combine features of two or more single-subject designs (e.g., multiple baseline with reversal).
- Enhance experimental control and allow for testing in more complex situations.
- Examples include alternating treatments with reversals and combining designs to test multiple variables or interventions.
Internal vs. External Validity
- Internal validity: How well the study controls variables and demonstrates a cause-effect relationship.
- Factors improving internal validity: appropriate design, relevant participants, short duration (controls maturation), strong interobserver agreement, high procedural integrity, and control of confounding variables.
- External validity: How widely results can generalize to other settings, populations, or similar procedures.
Experimental Analyses
- Comparative analysis: Compares two different interventions or an intervention to baseline to see which is more effective.
- Component analysis: Determines which parts of a treatment package are necessary by systematically removing components.
- Parametric analysis: Examines the effect of different values or dosages of a single intervention.
Key Terms & Definitions
- Multiple Baseline Design — A design testing one intervention across different DVs with staggered baselines.
- Probe — A brief test to check if a skill is present.
- Verification — Confirming a predicted outcome by extending baseline.
- Changing Criterion Design — Design that changes the goal/criterion to shape behavior incrementally.
- Component Analysis — Identifies which parts of a treatment package are necessary.
- Parametric Analysis — Tests the effects of varying levels/dosages of an intervention.
- Internal Validity — The degree to which an experiment demonstrates a causal relationship.
- External Validity — The extent to which findings generalize beyond the experiment.
Action Items / Next Steps
- Review parts one and two of experimental design for foundational concepts.
- Read the "Right to Effective Behavioral Treatment" article (Van Houten et al.) for ethical guidelines.
- Complete any assigned quizzes, practice questions, or error analyses for exam preparation.
- Sign up for the email list to receive study tools and updates.