A comparison of mean phase difference and generalized least squares for analyzing single-case data

Area of Science:

• Psychology
• Behavioral Science
• Research Methodology

Background:

• Single-case data analysis is crucial for evaluating interventions.
• Quantifying treatment effects in single-case designs requires robust statistical methods.
• Existing methods may have limitations in handling complex data patterns.

Purpose of the Study:

• To compare two distinct procedures for analyzing single-case data: generalized least square regression and a novel non-regression technique.
• To evaluate the performance of these methods across diverse data conditions, including serial dependency and trend variations.
• To provide guidance for applied researchers in selecting appropriate single-case data analysis techniques.

Main Methods:

• Generated data simulating various single-case measurement patterns, including independent and serially related data.
• Inclusion of heterogeneity in autocorrelation, data variability, trends, and slope/level changes.
• Statistical comparison of a generalized least square regression analysis with a proposed non-regression method.

Main Results:

• Both the regression-based and non-regression techniques demonstrated adequate performance across a wide range of simulated conditions.
• The regression procedure yielded more statistically efficient estimates.
• The non-regression procedure showed greater sensitivity to detecting intervention effects.

Conclusions:

• Both analyzed single-case data analysis techniques are reliable tools for researchers.
• The choice between methods depends on whether statistical efficiency or sensitivity to intervention effects is prioritized.
• Recommendations are provided to aid applied researchers in selecting the most suitable analysis technique for their specific single-case study.