Bayesian Time-Series Models in Single Case Experimental Designs: A Tutorial for Trauma Researchers.

Area of Science:

• Psychology
• Clinical Psychology
• Psychiatry

Background:

• Single-case experimental designs (SCEDs) are increasingly popular in trauma treatment research due to their efficiency and ability to provide causal evidence.
• Despite their growing use, sophisticated analytical techniques for SCED data are underutilized in the field.

Purpose of the Study:

• To discuss the utility of SCED data in trauma research.
• To provide recommendations for overcoming challenges associated with SCED methodologies.
• To introduce and demonstrate two Bayesian models (Bayesian interrupted time-series and Bayesian unknown change-point) for analyzing SCED data.

Main Methods:

• The study discusses the application of Bayesian interrupted time-series (BITS) and Bayesian unknown change-point (BUCP) models.
• Tutorials and software codes are provided for estimating these models.
• Analyses were conducted using a published dataset and a simulated trauma-specific dataset.

Main Results:

• The Bayesian models (BITS and BUCP) are suitable for analyzing small-sample, autocorrelated SCED data common in trauma research.
• The study illustrates the practical application and interpretation of these models using real and simulated data.
• The findings highlight the potential of these advanced analytical techniques for strengthening causal inference in SCED trauma studies.

Conclusions:

• Sophisticated Bayesian models like BITS and BUCP can effectively analyze SCED data in trauma research.
• The utilization of these methods can enhance the robustness of causal evidence derived from small-sample studies.
• Recommendations and practical tools are provided to facilitate the adoption of these advanced analytical techniques in trauma research.