diadanalyses

Diadanalyses is a field of data analysis focused on dyadic data—the study of relationships between pairs of units within a larger population. It asks how attributes of each unit and the interaction between the two shape outcomes, while explicitly modeling the dependence among dyads that share units.

Data in diadanalyses are typically organized as ratings or measurements for actor pairs, such as a matrix of ties in a social network, an edge list, or paired observations collected from collaborations or exchanges. Analysts distinguish directed versus undirected dyads and may consider whether the two actors are interchangeable or distinguishable.

Methodologically, diadanalyses encompasses regression models for dyadic data, multilevel or actor–partner interdependence models, and network-based approaches such as exponential random graph models and stochastic actor-oriented models. Analysts address non-independence, reciprocity, and potential confounding by leveraging random effects, permutation tests, or likelihood-based inference.

Applications appear across social science and beyond, including social networks, organizational behavior, international relations, epidemiology of contact patterns, and user interaction data in online platforms.

Limitations include model complexity, data requirements for stable estimation, and challenges in interpretation when dyad-level effects interact with higher-level context. Ongoing work seeks to improve identifiability, handle missing data, and integrate dyadic analyses with broader network or multilevel frameworks.

Diadanalyses draws on social network analysis, multilevel modeling, and econometrics. It shares goals with network science and relational statistics and continues to evolve with computational advances.

See also: social network analysis, dyadic data analysis, network science.