You are here

Who You Know


Photo: Welton Doby III
Most agencies might be surprised at the ways they could use social network analysis to improve performance, says AHRQ's David Introcaso.

Here are a few realities: Improvements in clinical practice delivery occur at a slow pace; too low a percentage of Americans receive recommended health care; and the number of medical errors is unacceptable.

Because of these sobering realities, my employer, the Agency for Healthcare Research and Quality, whose mission it is to advance excellence in health care, is evolving its grant-funded publishing model of scientific research to include a network model of collaboration. The goal? Improve the quality of health care delivered in the United States.

This means that AHRQ, an agency within the Health and Human Services Department, has begun to focus more intentionally on improving its own network health by mapping and building its network web of relationships within the health care delivery community. To do this, we are using a social network analysis, or SNA, methodology.

Whether you recognize social networks or not, they are not just pervasive, they are the core or an essential feature in achieving organizational success—however that success is defined.

Popularized by the early 1990s play "Six Degrees of Separation" (just to note Martin Sheen and Charlton Heston, for example, are more networked or connected than Kevin Bacon) and later by Malcolm Gladwell's 2002 work The Tipping Point, networks are simply patterns of relations between individuals.

Fundamentally, networks are made up of who you know and who knows you. A more formal definition would be a set of self-organizing working relationships among actors, such that any relationship has the potential both to elicit action and to communicate information in an efficient manner.

Networks matter largely because more than 80 percent of people obtain their information from other people. Networks also matter because knowledge creation (as opposed to information sharing) is essentially an active, ongoing process of relating.

Understanding a network of relationships—figuring out who is obtaining information from whom and who is exchanging it with whom—or working to build an organization's network is therefore critical. Relationships between agents (not the agents themselves) are fundamental to understanding and explaining organizational innovation, performance and a host of other system dynamics. As the physicist John Wheeler has noted, nobody can be anybody without somebody being around.

Impact Gauge

Stats for Every Conceivable Relationship

How do you measure what's going on within any given network? Here are a half dozen basic measures common to social network analysis:


Brokerage: who brokers information to others
Centrality: the number of outgoing and incoming ties
Closeness: position in the network with respect to the total distance to all others
Cohesion: the average of the shortest paths between every pair of people
Density: how well-tied individuals are
Power: the level of closeness between people

In the case of AHRQ, network analysis is a tool being applied to both the planning and evaluation of the use of health-care research and its impact. We're doing this because government agencies increasingly play the role of integrator, or work to create or add maximum value by orchestrating a network of government and nongovernment assets or relationships.

Government organizations today, as a recent study by the Brookings Institution in Washington suggests, are about being networked or, more specifically, being "wired, joined up and pushed down." Beyond networks of individuals, network analysis can also play a vital role in government efforts to improve the use of information technology resources via the creation of cyberinfrastructures or by advancing computational collaboration, data acquisition and management services through high-performance cyberenvironments. After all, the enterprise architecture movement strikes right at the heart of how agents within a network—in some cases, groups of networks—interact.

Trying to understand and improve networks is not a new discipline. Over the past 70 years, SNA—which has evolved through the use of a wide variety of disciplinary fields—has become increasingly sophisticated in assessing and measuring patterns of connections. There are now numerous analytical methods designed to describe or quantify social networks. (Likely the most comprehensive network analysis text, Social Network Analysis: Methods and Applications by Stanley Wasserman and Katherine Faust, runs well over 500 pages.)

Beyond the ubiquity of networks, it is important to note that social, professional and organizational structure can enable or constrain networks; network function is related to, or controlled by, power, influence and position; network ties or relationships can be asymmetrically reciprocal, meaning information exchange can be unequal; and so-called weak ties can frequently be more important than strong ties since the former are vital to an individual's access to larger networks.

To unweave these factors from one another and understand their interrelationships takes research tools, and there are numerous software products that can graphically map networks, create socio-grams, gather data and report metrics.

Network analytics can inform both individual behavior and network use. SNA tools have generally been used to improve process and/or distribution (markets are, after all, nothing but directed networks), facilitate learning, spur innovation and, more generally, bring leaders together in teams to create trust and leverage collaboration.

Network tools can help create communities of practice by identifying key members and measuring connectivity. They can be used to measure and assess existing collaboration or the extent to which appropriate partnering cross-collaboration or intra-collaborations are occurring. They can measure information flow to align required expertise and be used for integration because, for example, large-scale and organizational system change is information- and knowledge-intensive and therefore substantially a matter of network integration.

Network analysis also is used to identify other desirable players or parties for product or program diffusion and dissemination. These players or links (again weak ties) can aid or build the network by providing diagnostic information in assessing further connections among network members and how information is entering and leaving or channeling in the network. Finally, SNA can be used to examine how network members are drawing upon and integrating the various expertise provided by those within their own organization and throughout the network.

Recognizing that improving the quality of health-care delivery is not a disembodied process of change but rather one that necessitates optimizing social interaction, AHRQ in this past year therefore has begun employing network analysis in a number of ways:
• to improve hospital health care through a collaboration with
the Healthcare Roundtable for Chief Quality Officers;

• to analyze a network of nine health plans working to reduce
disparities in care delivery;

• to identify a process by which AHRQ can collect external stakeholder input to inform future research decisions;

• to improve interactions with sister HHS agencies—for example,
the Centers for Medicare and Medicaid Services;

• to understand the qualities and capacities of our network interactions—for example, what are the levels of awareness, concordance, inclusion, intention, respect, trust and other qualities the agency is attempting to engender;

• from a more traditional evaluative perspective, to assess the past performance of a number of AHRQ programs.

Consider the Possibilities

So look around and ask yourself, how can I apply SNA to better create and understand the network dynamics at work for the program activities at my agency? I think you will be surprised to see that the metrics gained through network analysis will give you new insights to make improvements in ways you have yet to realize.

Dec 31 2009

Comments