Analyzing Dyadic Sequence Data-Research Questions and Implied Statistical Models.

Area of Science:

• Social Psychology
• Quantitative Psychology
• Relationship Science

Background:

• Observational data analysis is crucial for understanding dyadic systems, particularly romantic relationships.
• Statistical models for dyadic observational data are underutilized.
• Medium sample sizes (N=100 couples or less) present unique analytical challenges.

Purpose of the Study:

• To present and compare statistical models for analyzing dyadic sequence data.
• To provide guidance on model selection for researchers working with medium-sized dyadic datasets.
• To facilitate the proper application of statistical models in empirical relationship research.

Main Methods:

• Comparison of statistical models including Pearson Correlation, aggregated logit models, multi-level approaches, basic Markov models, hidden Markov models, mixture Markov models, and optimal matching.
• Focus on models applicable to medium sample sizes (N ≤ 100 couples).
• Hands-on approach linking models to specific research questions regarding dyadic interactions.

Main Results:

• The study evaluates the applicability of various statistical models to different research questions about dyadic behavior.
• It demonstrates how specific models address questions about behavioral associations, immediate reactions, underlying dyadic processes, and latent group patterns.
• Recommendations are provided for model selection and data handling in dyadic research.

Conclusions:

• The contribution aims to increase the accessibility and application of statistical models for dyadic observational data.
• It equips researchers with tools and recommendations for analyzing relationship dynamics in medium-sized samples.
• An accompanying R-package ('DySeq') is introduced to support empirical research.