There's a common assumption in social science: your environment shapes your behavior. Spend time with heavy drinkers, and your drinking drifts up. That part is probably true. What I find more interesting in a new systematic review is the other half. Your drinking habits also shape who you end up spending time with.

The review, published in Frontiers in Public Health, looked at 27 studies that applied whole network analysis (WNA) to four behaviors: smoking, alcohol consumption, physical activity, and diet. The aim was to trace how these behaviors spread and take root inside social systems.

What they did

Researchers screened six academic databases for empirical studies using WNA, following PRISMA 2020 guidelines. 27 studies made the cut.

The key filter was sociocentric data. Most behavior research is egocentric: you ask someone about their personal network ("who do you spend time with?"). WNA is different. It maps every relationship within a defined population. That's what sets WNA apart.

What they found

Four mechanisms showed up consistently across all four behaviors:

  • Peer influence. Connected people adopt each other's behaviors over time.
  • Homophily. People with similar behaviors cluster together. Drinkers end up in circles with drinkers.
  • Network cohesion. Tightly knit groups reinforce behaviors more strongly than loosely connected ones.
  • Centrality. A handful of structurally central individuals have disproportionate influence on what norms spread through a network.

WNA also turned out to be well-suited to identifying behavior clusters - subgroups where a behavior has concentrated - and to capturing co-evolutionary dynamics, where network structure and behavior shift together over time.

The clearest findings came from longitudinal studies using stochastic actor-oriented models (SAOMs): statistical models designed to untangle two effects that often get conflated. Selection (you chose friends who drink like you) and influence (your friends changed your drinking). Most studies found both operating at once.

What it means

The selection-influence split is the key insight. Most people treat their social environment as something that happened to them. WNA shows the other direction. Your habits helped construct that environment.

Centrality is probably the most actionable finding. In any group, a small number of people sit at structural hubs - many connections, high overlap across subgroups. Their behavior quietly sets the norm. If someone central shifts how they drink, the cluster tends to follow. That's not a prescription. It's just how diffusion works.

The co-evolutionary part is worth sitting with. Network structure and behavior evolve together. The social system you're in today is partly a product of the habits you've had over the last few years.

Caveats

27 studies is a modest base for this kind of review. The search was also English-only, which likely underrepresents non-Western research. The included studies vary widely in design, population, and methodological sophistication - some used full longitudinal SAOM approaches, others didn't. That variation makes clean generalization hard.

There's also a structural limit the authors flag: WNA requires complete data on an entire network. That's logistically demanding and rarely feasible in standard public health settings.

Source: Frontiers in Public Health, doi:10.3389/fpubh.2026.1834356