The Structure of Negative Social Ties in Rural Village Networks

Alexander Isakov, James Fowler, Edoardo M. Airoldi, Nicholas A. ChristakisSociological 2019 · DOI

Abstract

Negative (antagonistic) connections have been of longstanding theoretical importance for social structure. In a population of 24,696 adults interacting face to face within 176 isolated villages in western Honduras, we measured all connections that were present, amounting to 105,175 positive and 16,448 negative ties. Here, we show that negative and positive ties exhibit many of the same structural characteristics. We then develop a complete taxonomy of all 138 possible triads of two-type relationships. Consistent with balance theory, we find that antagonists of friends and friends of antagonists tend to be antagonists; but, in an important empirical refutation of balance theory, we find that antagonists of antagonists also tend to be antagonists, not friends. Finally, villages with comparable levels of animosity tend to be geographically proximate. Similar processes, involving social contact, give rise to both positive and negative social ties in rural villages, and negative ties play an important role in social structure.

Introduction

Most social network research focuses on "positive" ties—who we like, work with, or seek advice from. Negative ties (animosity or "enemies") are much harder to study because people are often reluctant to admit them, and mapping them requires a "sociocentric" approach where you survey everyone in a closed ecosystem. This research addresses a long-standing debate in social psychology called Balance Theory. Proposed in the mid-20th century, this theory suggests that social systems gravitate toward "balanced" states (e.g., your friend’s friend is your friend; your friend’s enemy is your enemy). Specifically, it predicts that "the enemy of my enemy is my friend." This paper tests whether these theoretical "rules" actually hold up in large-scale, face-to-face human populations.

Methods

The researchers conducted a massive sociocentric study in Western Honduras using a custom tablet-based platform called Trellis. They conducted face-to-face interviews with 24,696 adults across 176 isolated villages, asking residents to nominate friends and people they "do not get along with."

Representative social network maps of six Honduran villages showing positive (grey) and negative (red) ties.

To analyze the data, they developed a taxonomy of all 138 possible "triads" (relationships between three people) involving positive, negative, and null ties. They then created 1,000 "random worlds" by reshuffling the observed ties while keeping the number of people and ties constant. This allowed them to see which social patterns occurred more or less often than would be expected by pure chance. Finally, they used logistic regression with village-level fixed effects to determine how personal traits influenced the likelihood of a tie.

Key Insights and Results

Structural Mirroring and Homophily

Surprisingly, negative ties exhibit the same structural properties as friendships, such as skewed degree distributions and homophily. We don't just like people like us; we tend to dislike people who share our traits (same sex, similar age). Furthermore, familiarity breeds contempt: the data showed that having friends in common actually increases the odds of two people being enemies, likely because they are forced into frequent contact.

The Refutation of Balance Theory

Traditional balance theory suggests a triangle of three enemies is "unstable" and should resolve into two people teaming up against a third. However, the researchers found the opposite: the enemy of my enemy is 1.6 times more likely to be my enemy than a friend. While some aspects of balance theory were validated (e.g., the friend of an enemy is an enemy), the core tenet regarding mutual enemies was empirically refuted.

Validation of Balance Theory and the empirical refutation of the 'enemy of my enemy' rule.

Social and Geographic Distribution

  • Prevalence and Reciprocity: Negative ties made up roughly 15.6% of all reported connections. Friendships are much more likely to be mutual (33.6% reciprocated) than animosities (only 5.1% reciprocated).
  • Community Fissures: Negative ties are 3 times more likely to exist between different social "communities" within a village than within the same community, acting as a "repulsive force" that defines group boundaries.
  • Geographic Clustering: Villages with high levels of negative ties tend to be geographically near other high-animosity villages, suggesting regional cultural or environmental factors influence social friction.

Discussion

This study demonstrates that negative ties are a fundamental, structurally organized part of our social fabric. However, the research is limited by its cross-sectional nature, which cannot show how a friend becomes an enemy over time. Future work should investigate how these negative ties affect a village's ability to cooperate on public goods. Does a certain level of animosity drive group cohesion through competition, or does it purely degrade social capital?

Summary updated

Abstract

Negative (antagonistic) connections have been of longstanding theoretical importance for social structure. In a population of 24,696 adults interacting face to face within 176 isolated villages in western Honduras, we measured all connections that were present, amounting to 105,175 positive and 16,448 negative ties. Here, we show that negative and positive ties exhibit many of the same structural characteristics. We then develop a complete taxonomy of all 138 possible triads of two-type relationships. Consistent with balance theory, we find that antagonists of friends and friends of antagonists tend to be antagonists; but, in an important empirical refutation of balance theory, we find that antagonists of antagonists also tend to be antagonists, not friends. Finally, villages with comparable levels of animosity tend to be geographically proximate. Similar processes, involving social contact, give rise to both positive and negative social ties in rural villages, and negative ties play an important role in social structure.
BibTeX
@inproceedings{isakov2019structure,
  author = {Alexander Isakov and James Fowler and Edoardo M. Airoldi and Nicholas A. Christakis},
  title = {{The Structure of Negative Social Ties in Rural Village Networks}},
  booktitle = {Sociological},
  year = {2019},
  doi = {10.15195/v6.a8},
}
Figure