Evidence for the weak ties theory using LinkedIn data
On September 15th, 2022, researchers from LinkedIn, Stanford, MIT, and Harvard published a 5 year longitudinal study on how the strength of ties impact job opportunity and mobility, or the likelihood that a user will change to a new job as a result of their network. This study involved over 20 million users, and some 2 billion new connections, and 600,000 accepted new jobs. The researchers were able to perform a randomly controlled survey by tuning and randomizing the “People You May Know” algorithm on LinkedIn, which recommends new connections to users. Some subjects were shown more possible connections to other users with strong ties (as measured by strength of network, or intensity/frequency of contact), while other subjects were shown more connections with weak ties. The number of job applications and job acceptances as detected by LinkedIn were then measured across these populations and compared. Multiple large-scale experiments such as this were performed, which helped identify causality of these social networks in job mobility.
The results of the study found causal evidence for the theory of weak ties we discussed in class, that acquaintances are likely to be more influential than closes friends in social networks. Given that this study is the largest of its kind by a significant margin, this alone would be an important result.
There are a number of interesting learnings from this paper that add nuance to what we learned in class. This paper concludes that while moderately weak ties by network strength and the weakest ties by communication intensity were most important for job transmission. This result was related to their finding that there are diminishing returns on tie weakness. They described the relationship between the weakness of ties and job transmission as an “inverted U-shape”, expressing that there was a limit to the quantity of weak ties before returns on those ties diminished.
Additionally, a topic of interest is the “paradox of weak ties”, or the fact that in the United States, while acquaintances are very likely to help a job seeker find a job, a strong tie is often more impactful in securing that job. While the “inverted U-shape” of tie weakness and job transmission may help resolve some of that paradox, another interesting finding was that job mobility was impacted by ties differently in different industries. The researchers found that in more digital industries, weak ties were more important in increasing job mobility, while in less digital industries, strong ties were more important. As a software engineer by trade, it looks like I should be spending more time on LinkedIn clicking on 3rd degree connections.
Given the prevalence of online social networks in important life decisions such as finding a job, and the vested interest these companies have in understanding how networks operate, it wouldn’t surprise me if there are more studies that explore networks with massive datasets in the pipeline. I would love to see more research done on how ideas propagate through social networks (the 2018 research project by NBC news studying how ideas were spread by Russian Twitter Trolls during the 2016 election comes to mind).
Links:
The power of weak ties in gaining new employment | MIT News | Massachusetts Institute of Technology
A causal test of the strength of weak ties | Science