Network Analysis Maps Organizational Culture as a Web of Interconnected Elements

For more than four decades, organizational culture has been measured the same way: boil down a company’s shared values, behaviors, and norms into a handful of abstract dimensions—clan, adhocracy, market, hierarchy—and score them. Now a team of Belgian researchers argues that this approach has been hiding the most interesting part of the story. In a study published in Frontiers in Organizational Psychology, they introduce culture networks, a method borrowed from psychometric network theory that treats organizational culture not as a set of separate dimensions but as a living web of interconnected elements, each one pulling on the others in organization-specific ways.

The research, led by Senne Letouche, Sofie Dupré, Maarten De Schryver, and Bart Wille of Ghent University, responds to a problem that has long frustrated culture scholars. Seminal definitions of organizational culture, including Edgar Schein’s influential account, describe culture as a pattern or system of basic assumptions, values, and beliefs that interconnect and shape one another. Yet the dominant measurement tradition—the Competing Values Framework, or CVF—lumps concrete culture elements into higher-order dimension scores, discarding the very interactions that make a culture what it is. Scholars have described the resulting tangle of mutually reinforcing values, behaviors, and norms as a can of worms: pull on one element and the whole messy pack comes with it.

Psychometric network analysis, which gained popularity in clinical psychology as a way to study symptoms and their interactions rather than higher-order disorders, offers a way out. Instead of assuming that an invisible latent construct causes the observed items—a reflective measurement model—network models abandon latent constructs altogether. Each culture element becomes a node in a graph, and the statistical relationships between them, estimated as regularized partial correlations, become edges. Blue edges indicate positive associations, red ones negative, and the thickness of each line corresponds to the strength of the connection. Because organizational culture is inherently collective, the edges are computed from between-person covariation in employees’ responses within a focal organization and interpreted as a shared culture structure at the organizational level.

To demonstrate the approach, the researchers analyzed culture survey data from two Belgian organizations. The first dataset comprised 4,270 responses from a large public service organization, with a response rate of 98 percent because the survey was part of a mandatory strategic exercise. The second came from a health care organization, where 338 of 800 invited employees completed the survey voluntarily, a response rate of 42 percent. In both samples, culture was measured with a Dutch translation of the 16-item CVF scale, capturing the four classic dimensions—clan, adhocracy, market, and hierarchy—each through four concrete elements covering leadership behaviors, orientations, values, and emphases. Participants rated their agreement on a 7-point Likert scale, and the networks were estimated in R using the bootnet, networktools, and qgraph packages, with a graphical lasso regularization procedure to improve interpretability.

The resulting culture maps revealed connections that would go unnoticed at the dimension level. In both organizations, the element describing rule-enforcing leaders was positively related to the perception of a stable, permanence-emphasizing environment, with partial correlations of 0.59 in the public service organization and 0.16 in the health care organization. More striking were the cross-dimension and cross-organization differences. Risk-taking leader behaviors were positively associated with perceptions of a dynamic, entrepreneurial environment in the health care organization, at 0.34, but the relationship turned slightly negative in the public service organization, at −0.04. Warm and caring leader behaviors went hand in hand with perceptions of an achievement-oriented environment in the public service organization but with reduced competition in the health care organization. Traditional dimension scores would average these patterns away entirely.

The second demonstration tackled a question at the heart of every culture change effort: where are the roots of a culture, and where can interventions do the most good? In network terms, this means identifying the most central nodes—elements with more and stronger connections to the rest of the system. The researchers used the expected influence index, which accounts for the sign of edge values and is therefore appropriate for networks containing both positive and negative relationships. Two leadership elements ranked among the three most influential nodes in the public service network, with market leader—the degree to which managers formulate challenging objectives—emerging as the single most central element, at a standardized expected influence of 1.12. Hierarchy emphasis and hierarchy leader followed closely. The finding aligns with decades of theory holding that leaders establish norms for social conduct, clarify which behaviors are valued, and set priorities by allocating resources, making leadership behaviors plausible drivers of the wider culture.

The third demonstration addressed cultural inertia, the notorious resistance of organizational cultures to change. According to the network view, a culture is an interlocked set of values, practices, and assumptions that fit together as a mutually reinforcing system. Connectivity, or density, quantifies how tightly the elements are interlocked. When two strongly connected elements are both highly present—say, rule enforcement and environmental stability—an external force that reduces one creates an unstable equilibrium: the system must either undo the change or shift all the other connected elements to align with it. As connectivity increases, the latter becomes less likely, so dense networks tend to change in an all-or-none fashion. Drawing on attitude strength theory, the researchers hypothesized that employees with longer organizational tenure would report more strongly connected networks, because culture knowledge, memory accessibility, and person-culture fit all increase with time.

The results were more nuanced than expected. Using a network comparison test with 1,000 permutations, the researchers compared the culture networks of 1,002 low-tenure employees, who had worked in the organization for less than five years, with those of 1,349 high-tenure employees, who had worked there for more than twenty. Contrary to the hypothesis, overall connectivity did not differ significantly between the two groups. But the network structures were not identical: the maximum edge-weight difference was significant, indicating that at least one connection differed between the groups. The edge between hierarchy leader and hierarchy emphasis, for instance, was significantly stronger among high-tenure employees, while the connection between hierarchy leader and clan emphasis was stronger among low-tenure employees. Because age and tenure are naturally correlated, the researchers regressed age out of each culture item and replicated the same pattern.

The authors argue that centrality and connectivity are not opposing predictions but different properties of the same system. Centrality is a local, node-level property: it identifies which element would propagate furthest through the existing pattern of connections if an intervention takes hold. Connectivity is a global, system-level property that reflects how strongly the network pulls back toward its prior configuration once intervention pressure is removed. Dense connections are double-edged—the same conditional dependencies that allow a well-targeted change at a central node to reach many other elements are also what makes the system prone to reverting once the intervention is withdrawn. Centrality tells us where a change would spread; connectivity tells us how much sustained force is needed to make it take hold in the first place.

The implications reach beyond theory. For organizational development practitioners, culture networks generate visual maps that can be presented intuitively as culture maps, guiding change interventions by showing which elements are drivers and how tightly the system is interlocked. The approach also challenges the CVF’s own theoretical mapping: the substantial positive interconnections across elements suggest that the four supposedly competing value orientations co-occur in practice and are better interpreted as ideal-typical tensions rather than mutually exclusive opposites. The researchers acknowledge limitations—cross-sectional data support only putative causal conclusions, and both organizations were public service and health care firms rather than the commercial contexts in which the CVF was originally developed. Still, they argue the time is ripe for a conceptualization that returns to the roots of culture theory, treating cultures as emergent systems of nodes interconnected in specific, organization-dependent ways. Future work could extend the networks to include antecedents and outcomes such as innovation, turnover, or job satisfaction as nodes, and could tailor culture maps to each organization’s own set of elements rather than a universal item set.

Subject of Research: A psychometric network study reconceptualizes organizational culture as a system of interconnected elements and demonstrates the approach with survey data from two Belgian organizations.

Article Title: A psychometric network approach to organizational culture: opening the can of worms

Article References: Letouche, S., Dupré, S., De Schryver, M., & Wille, B. (2026). A psychometric network approach to organizational culture: opening the can of worms. Frontiers in Organizational Psychology, 4, Article 1819497. https://doi.org/10.3389/forgp.2026.1819497

Image Credits: AI Generated

DOI: 10.3389/forgp.2026.1819497

Keywords: organizational culture, psychometric network analysis, culture networks, Competing Values Framework, cultural inertia, centrality, connectivity, leadership, Ghent University, organizational psychology, psychometric, network