Nodal Heterogeneity can Induce Ghost Triadic Effects in Relational Event Models.

Area of Science:

• Network Science
• Computational Social Science
• Statistical Modeling

Background:

• Temporal network data captures interactions over time, crucial for understanding dynamic systems.
• Existing relational event models struggle with unobserved actor heterogeneity, potentially leading to biased results.
• Nodal heterogeneity, or individual differences, can influence network dynamics but is challenging to fully capture.

Purpose of the Study:

• To investigate how unobserved sender and receiver effects in temporal networks can create spurious findings.
• To propose and evaluate a novel random-effect extension of the relational event model.
• To demonstrate the model's effectiveness in resolving issues related to nodal heterogeneity.

Main Methods:

• Development of a random-effect extension for relational event models.
• Comparison of the proposed model against traditional approaches like in-degree and out-degree statistics.
• Analysis of temporal network data to identify and correct for sender and receiver effects.

Main Results:

• Failure to account for sender and receiver effects can induce "ghost triadic effects."
• The proposed random-effect relational event model effectively addresses nodal heterogeneity.
• The new model demonstrates superior performance compared to traditional methods in resolving hierarchy principle violations.

Conclusions:

• Including random effects in relational event models is crucial for accurately analyzing temporal network data.
• The proposed method provides a robust solution for unobserved heterogeneity in network analysis.
• This approach enhances the reliability of findings in dynamic network studies.