OTHER SIDE PROJECTS

Strength of Weak Ties

One of the most influential and widely cited theories in all of social science is Granovetter's 1973 Strength of Weak Ties. We might expect that weak social ties like acquaintances would have limited value because they are weak. However, Granovetter hypothesized that, because bridges between different social worlds are more likely to be weak social ties than strong ties, weak ties can be especially useful as sources of novel information.

Despite its enormous influence, this theory has received limited empirical support. One reason is that the theory was derived from some assumptions about social networks that are often unmet. Not all bridges are weak ties, and sometimes strong ties can be bridges. In practice, this means that there is nothing particular special or "strong" about weak ties. Instead, what really matters is not whether a tie is weak or strong, but whether or not it is a bridge.

Neal, Zachary P. (2024). The not-so-forbidden triad: Evaluating the assumptions of the Strength of Weak Ties. Network Science, 12, 289-304. https://doi.org/10.1017/nws.2024.12

Neal, Zachary P. (2022). Sometimes weak ties are just weak. Science (eLetters), 377. https://doi.org/10.1126/science.abl4476#elettersSection (post-print)

Hear me talking about this work

for 60 seconds on the

NPR Academic Minute segment

Psychological Networks

The nodes in networks can often take many forms: people in social networks, places in transportation networks, or cities in urban networks. In "psychological networks" the nodes are variables (e.g., beliefs, symptoms, or traits), and the edges are the statistical associations (e.g., partial correlations) among them in a sample of people. This type of network has become popular to study in psychology, however this might not be a good idea:

There are already methods for studying statistical associations among a set of variables.
Metrics developed for social networks or flow networks may not make sense in this context.
The edges are estimated from non-network data, not directly observed.

But, one overarching problem for this kind of network is the boundary specification problem. A network's structure is known to be distorted if nodes are missing. Because there is no way for a psychological network to include all beliefs or all traits, it will always be distorted.

Neal, Z. P., and Neal, J. W. (2023). Out of bounds? The boundary specification problem for centrality in psychological networks. Psychological Methods, 18, 179-188. https://doi.org/10.1037/met0000426 (pre-print available at https://psyarxiv.com/nz6k3)

Neal, Z. P., Forbes, M. K., Neal, J. W., Brusco, M., Krueger, R., Markon, K. E., Steinley, D., Wasserman, S., and Wright, A. G. (2022). Critiques of network analysis of multivariate data in psychological science. Nature Reviews Method Primers, 90. https://doi.org/10.1038/s43586-022-00177-9 (preprint available at: https://psyarxiv.com/jqs3n/; annotated bibliography: https://psyarxiv.com/ke6qn/)