The model supports social network theory

A new Bayesian analysis of remote work data supports one of the oldest theories in social networks, with fresh implications for the future of work environments.

Weak ties describe the infrequent connections we maintain in our social networks with acquaintances, occasional colleagues, and occasional friends.1. Despite their ‘weakness’, these relationships often give rise to new ideas, opportunities, and advice in organizational settings.1,2,3,4. Although this finding has largely stood the test of time, much remains unknown about its causal mechanisms4. What dynamically generates these relationships? What keeps them? Can they withstand ‘external’ shocks from the outside, such as changes in location induced by the COVID-19 pandemic? This evolving landscape gives new urgency to the call for “insights into the particular.” [social] Processes at specific times through fresh data, methods, and analytical rigor5. Now, an article on Nature Computational Science answers this call, showing that new computational methods can investigate difficult questions about organizational networks. In their piece, Daniel Carmody and colleagues6 In the context of Covid-19 – use Bayesian time series analysis to provide evidence supporting an important, understudied theory in social networks called propinquity – which states that spatial proximity increases the chances of creating new connections and strengthening existing ones.

Weak tie formation often begins when family, community, or organizational activities bring people together. Being in physical proximity to someone increases the chance of serendipitous interactions. In turn, these interactions give people the opportunity to explore common qualities, interests, and behaviors with one another, and thus form relationships. These stages describe the social process of reciprocity—the closer we are physically to another person, the more likely we are to form a new tie or rekindle an existing one with them.7 (Figure 1a). Existing work has found that sharing socially important qualities can amplify the effects of proximity8 And that closeness extends to virtual closeness9,10. But the concept is often taken for granted despite its important implications for how we design organizations and social gatherings. This leaves a puzzling lack of evidence showing the process as such, and therefore a lack of knowledge about how the process can improve everything from technology diffusion to inequality.4.

Figure 1: Loss of physical proximity due to distant work led to atrophy of weak relationships with nearby researchers.

a, Propinquity relates the distance between two individuals (horizontal axis) to the probability that those individuals form a tie (vertical axis). Individuals who are physically close to each other are more likely to interact, and therefore form relationships with each other (curve shown). The referenced study provided empirical evidence to support this social science theory. bThe central finding from the referenced study, showing the change in the number of weak ties between researchers as a function of the distance between their labs from March 2020 to July 2021. Data were collected from the original study.6. Statistically significant increases in weak ties are shown in blue; Significant reductions are shown in orange; and non-significant changes appear in gray. Error bars represent 95% confidence intervals, and *** indicates a statistically significant finding. p <0.001 (all other bars were p > 0.1). The graph shows that researchers who once worked close to each other interacted less with each other throughout the COVID-19 pandemic, which led to the loss of the weak ties between those individuals. Meanwhile, researchers working in the same lab group (remotely) strengthened their existing ties with those individuals and formed weaker ties than they would have if they shared physical lab space.

Carmody et al. provided an important empirical demonstration that shows weak ties as they form and decline through proximity. Most social network studies compare a few snapshots of social networks over a period of time because collecting granular temporal network data often proves too difficult, both logistically and ethically. Ultimately, it prevents us from witnessing when and how many relationships form. The authors overcame this obstacle by estimating the number of weak ties among researchers at the Massachusetts Institute of Technology (MIT). Their e-mail dataset spans two dramatic changes in a year and a half in researchers’ work locations during the COVID-19 pandemic. The first transition occurred on March 23, 2020, when MIT halted most individual research activities. Researchers began working from home, preventing hypothetically weak relationships from forming through propriety. The second transition occurred on 15 July 2021, when researchers began returning to campus, hypothetically increasing weak tie formation through propinquity. The authors examined an e-mail network spanning these real-world location transitions through a synthetic counterfactual e-mail network absent these transitions to estimate how earlyness affects weak tie formation.

Methodologically, the authors constructed their synthetic counterfactual through a Bayesian structural time series (BSTS) approach that separates treatment effects (here, remote work) from treatment-unaffected properties (such as linear trends and cyclical variances). This enabled them to construct reliable intervals for the expected number of weak ties with and without remote work. Their analysis showed that remote work may have cost about 5,100 poor relationships during the remote work period—about 1.8 relationships per person—for important public health purposes, of course, due to the pandemic. In addition, researchers were more likely to have weak ties with people working in close labs than with those working in the same or distant labs (Figure 1b). Consequently, researchers were ‘stuck’ reinforcing their existing relationships. As a validation of this finding, they designed a generative network simulation to simulate several tie formation mechanisms (such as laboratory sharing, mutual friends, and co-location sharing). In doing so, they show qualitatively that a propinquity factor replicates the findings of their BSTS analysis.

Carmody and colleagues demonstrated the ability of modern computational techniques to support old social science theories and identify new phenomena. Their research may provide rigorous tools for validating the causal hypotheses of social networks and information voraciousness.4,5. Future studies should consider how other confounders may affect these results. For example, the authors note that insufficient data prior to the pandemic limited their ability to predict cyclical effects. This echoes the importance of appropriately constructing controls for quasi-experiments5. Building size and location, researcher demographics, university-required activities, and type of information sharing (professional vs. friendly11) may confound the results, but may also reveal unexplored questions. A word of caution, though: the benefits of computational power become moot without appropriate framing from qualitative and theoretical social science.5. Naive computational studies run the risk of drawing completely wrong conclusions. Interdisciplinary collaboration between scholars with computational and theoretical perspectives can begin to answer difficult and long-standing questions, though—with patience and curiosity on all sides—can benefit how we design opportunities for social interaction in the years to come.


  1. Granovetter, MS M. J. Social. 781360-1380 (1973).

    Article Google Scholar

  2. Constant, D., Sproull, L. & Kiesler, S . organ Science 7119-135 (1996).

    Article Google Scholar

  3. Hansen, MT the manager Science Q. 4482–111 (1999).

    Article Google Scholar

  4. Aral, S. M. J. Social. 1211931–1939 (2016).

    Article Google Scholar

  5. Davis, GF organ Res. meth. 13690–709 (2010).

    Article Google Scholar

  6. Carmody, D. et al. Nat. Compute. Science (2022).

    Article Google Scholar

  7. McPherson, M., Smith-Lovin, L. & Cook, JM year Rev. Social. 27415–444 (2001).

    Article Google Scholar

  8. Regans, R. organ Science 22835–849 (2011).

    Article Google Scholar

  9. Walther, JB and Bazarova, NN normal Res. 35622–645 (2008).

    Article Google Scholar

  10. Lewis, K. and Kaufman, J. M. J. Social. 1231684–1742 (2018).

    Article Google Scholar

  11. Balakundi, P and Harrison, DA Acad. management J. 4949–68 (2006).

    Article Google Scholar

Download references

Author information

Authors and affiliations

Corresponding author

Correspondence of John Meluso.

Ethics declarations

Competing interests

The author declares no competing interests.

About this article

Verify currency and authenticity with CrossMark

Cite this article

Meluso, J. A model supports social network theory.
Net Comput Sci 2, 471–472 (2022).

Download the quote

Leave a Comment

Your email address will not be published.