Strategies in social network formation

We run a computerised experiment of network formation where all connections are beneficial and only direct links are costly. Players simultaneously submit link proposals; a connection is made only when both players involved agree. We use both simulated and experimentally generated data to test the determinants of individual behaviour in network formation. We find that approximately 40% of the network formation strategies adopted by the experimental subjects can be accounted for as best responses. We test whether subjects follow alternative patterns of behaviour and in particular if they: propose links to those from whom they have received link proposals in the previous round; propose links to those who have the largest number of direct connections. We find that together with best response behaviour, these strategies explain approximately 75% of the observed choices. We estimate individual propensities to adopt each of these strategies, controlling for group effects. Finally we estimate a mixture model to highlight the proportion of each type of decision maker in the population -- network formation ; experiments ; mixture models