Structural balance in signed networks : separating the probability to interact from the tendency to fight

Abstract

Structural balance theory implies hypothetical network effects such as “the enemy of an enemy is a friend” or “the friend of an enemy is an enemy.” To statistically test such hypotheses researchers often estimate whether, for instance, actors have an increased probability to collaborate with the enemies of their enemies and/or a decreased probability to fight the enemies of their enemies. Empirically it turns out that the support for balance theory from these tests is mixed at best. We argue that such results are not necessarily a contradiction to balance theory but that they could also be explained by other network effects that influence the probability to interact at all. We propose new and better interpretable models to assess structural balance in signed networks and illustrate their usefulness with networks of international alliances and conflicts. With the new operationalization the support for balance theory in international relations networks is much stronger than suggested by previous work.