Randomized single-case AB phase designs: Prospects and pitfalls.

Area of Science:

• Psychology
• Educational Psychology
• Clinical Psychology

Background:

• Single-case experimental designs (SCEDs) are crucial for evaluating interventions in individual participants.
• The AB phase design, a basic SCED, can be enhanced with randomization.
• Randomization involves randomly determining the intervention's start point.

Purpose of the Study:

• Introduce and review the randomized AB phase design.
• Discuss data analysis possibilities, pitfalls, and solutions using randomization tests.
• Investigate the impact of linear trends on randomization tests in this design.

Main Methods:

• Introduction and review of the randomized AB phase design.
• Exploration of data-analytical approaches and randomization tests.
• Simulation study to assess Type I error and power under linear trends.

Main Results:

• Randomization tests maintain Type I error control even with unexpected linear trends.
• The power of randomized AB phase designs is adequate only for substantial treatment effects and large sample sizes.
• Unexpected linear trends can impact the power of randomization tests.

Conclusions:

• Randomized AB phase designs are experimentally valid but have limited power for small effects.
• For small treatment effects or sample sizes, consider more complex designs like randomized ABAB or multiple-baseline designs.
• Effective analysis of randomized AB phase designs requires careful consideration of potential data pitfalls.