Three Essays On The Strength Of Long-Range Communication Ties

Abstract

Granovetter's article on the strength of weak ties is one of the most widely cited in the social sciences in the past 40 years. Compared to a strong tie, a weak tie tends to span "long" network distances, thereby promoting access to otherwise unavailable information, greater social integration, and more rapid diffusion of innovations. However, the hypothesized length of weak ties has eluded empirical research, primarily due to the paucity of fine-grained network data at the population level. Using a bidirected phone call network (51.3M nodes) constructed from complete call logs during a one-month period in the entire United Kingdom, I confirm that the median tie strength, measured as call volume, initially declines as the length of the tie, measured as the second shortest path length, increases from 2 to 4 steps, as Granovetter predicts. However, I find that the opposite holds for ties of length greater than 4, such that ties of length 10, while relatively few in number, are nearly as strong as ties of length 3. Substantively similar patterns are found from the analysis of Twitter communication networks in eight countries that vary in national culture, suggesting that a common generative process may lie behind the increasing strength of long bridging ties. I examine three competing explanations: 1) nodes with few neighbors tend to invest heavily in their relations with one another but with a lower probability of having a neighbor in common; 2) the telephone is used both socially and instrumentally, such that the social use is consistent with Granovetter's thesis while the instrumental use is not; 3) social and spatial mobility causes social ties to be "stretched" across the network, with a probability of being broken that is greater for ties that are weak. I conclude that this selection effect is the explanation with the greatest empirical support.