Author Correction: The paradox of second-order homophily in networks

Anna Evtushenko, Jon KleinbergScientific 2022 · DOI · OpenAlex

Introduction

In 2021, Anna Evtushenko and Jon Kleinberg published "The paradox of second-order homophily in networks" in Scientific Reports. This influential work explored how similarity between an individual and their friends' friends (second-order homophily) behaves in complex social structures. Recently, the authors issued a formal correction to address mathematical and logical inconsistencies found in the paper's Supplementary Information. While the core findings of the original research remain intact, these corrections ensure the analytical rigor required for replication and further theoretical development.

Technical Refinements in Supplementary Proofs

The correction focuses on four primary areas within the mathematical proofs provided in the Supplementary Information. These fixes are essential for maintaining the formal integrity of the network models described.

1. Indexing and Notation

A significant portion of the correction involves rectifying transcription errors in indices. Specifically, in the "Step-by-step proof of the list version," an index previously labeled as ‘R’ has been corrected to ‘B’. Such notation errors, while seemingly minor, are critical in formal proofs where indices represent distinct sets or properties within the network graph. Ensuring that these labels are consistent is vital for researchers attempting to step through the derivation.

2. Summation Logic and Symmetry

The authors identified an incomplete term in their interaction descriptions. A term originally written as $T_{ij}$ has been expanded to $T_{ij} + T_{ji}$. This change ensures that the summation properly accounts for the bidirectional or symmetric nature of the interactions being modeled. This is a fundamental requirement for the internal consistency of the "list version" of their network model, ensuring that all relevant interactions are captured in the final calculation.

3. Boundary Conditions and Constraints

To prevent mathematical singularities, such as division by zero, the authors have explicitly added a necessary condition for certain denominators. They note that "all homophily values are between 0 and 1." By formalizing these physical and probabilistic constraints, the proofs now explicitly exclude undefined states, reinforcing the robustness of the equations and ensuring they remain valid across the entire parameter space of the model.

Simplifying the "Red Gap" Argument

One of the more conceptual corrections involves the "red gap," a metric used to measure diversity or representation disparity within the network. In these models, groups are often color-coded (e.g., "red" and "blue") to track how connectivity differs between populations. The original logic explaining why this gap remains positive in specific network configurations was found to be unnecessarily complex.

The authors have streamlined this argument, making the phenomenon of homophily-driven diversity gaps more interpretable. This simplification clarifies how structural homophily leads to persistent disparities in node representation, providing a more direct path for researchers to verify the existence of these gaps in both theoretical models and empirical datasets.

Conclusion and Impact

This correction notice serves as a vital update to the scientific record. The primary result is the issuance of a corrected Supplementary Information file that is now mathematically consistent and logically robust. By refining the mathematical notation, correcting summation logic, and simplifying key arguments, Evtushenko and Kleinberg have ensured that the "paradox of second-order homophily" rests on a sound analytical foundation. For the network science community, these updates provide the clarity needed to build upon the original study's insights into social dynamics and network topology.

Summary updated

BibTeX
@inproceedings{evtushenko2022author,
  author = {Anna Evtushenko and Jon Kleinberg},
  title = {{Author Correction: The paradox of second-order homophily in networks}},
  booktitle = {Scientific},
  year = {2022},
  doi = {10.1038/s41598-022-23028-9},
}
Figure