Exchange and cohesion in dyads and triads: A test of Simmel's hypothesis.

This paper uses social exchange theory to address a classic question posed by Simmel (1964) regarding dyads and triads. The question is whether exchanges in a triad will generate more cohesion at the group level than exchanges in an isolated dyad. The main hypotheses, integrating several ideas from Simmel and social exchange theories, are as follows. First, triads generate less variability of behavior than dyads; that is, there is more uniformity or convergence in triads. Second, in the context of repeated exchange, we predict higher levels of cohesion in triads than in dyads. Third, positive emotion or affect has a stronger impact on cohesion in dyads than in triads, whereas uncertainty reduction has a stronger impact on cohesion in triads. To test these hypotheses, an experiment compared isolated dyads to dyads nested in a triadic exchange network. Subjects engaged in exchanges across a series of distinct episodes, using standard experimental procedures from research on relational cohesion (Lawler and Yoon, 1996) and exchange networks (Molm and Cook, 1995; Willer, 1999). Consistent with the hypotheses, the results reveal more convergence of behavior and higher cohesion in triads than in dyads; moreover, uncertainty reduction is the primary basis for cohesion in the triad, whereas positive affect was the primary basis for cohesion in the dyad. These results are discussed in relation to Simmelian dyad-triad dynamics and the theory of relational cohesion.