Diffusion models: two-step flow vs. network

More than once I’ve heard Everett RogersDiffusion of Innovations described as “Extension’s bible.”  Many Extension interventions apply diffusion theory to encourage adoption of target behaviors from horticultural practices to sustainable forest management.

But how do innovations move through a social network?  Two very different models are nicely summarized in an article I read recently (Watts & Dodds 2007; full citation below).  This post describes the two models, with some thoughts on applications to private forest management situations.

Two-step flow model


Source: Watts & Dodds 2007: 441.

Under the two-step flow model, a small number of early adopters receive information and pass on information from a central source to a much larger number of people.  These folks tend to access many media sources, filter information, and multiply certain messages through their networks to much larger audiences.

A common private forestry example is a master volunteer (e.g. Oregon State Master Woodland Manager or New York Master Forest Owner) who has received extensive training and subsequently shares her new knowledge with her neighbors.  She’s heard of every new idea, but has opinions that her less-involved neighbors have come to trust.

Because of the reputations they’ve earned, individuals like master volunteers also serve as opinion leaders rather than mere conduits of information.  Others look to them not only as sources of information, but as trusted filters or interpreters of that information.

Under the two-step flow model, opinion leaders play crucial roles–without their work to multiply and disseminate information, the information doesn’t reach other potential adopters.

Network model


Source: Watts & Dodds 2007: 441

Under the network model, information reaches all (or a much larger proportion of) the members of the community more or less equally.

Under the network model, influence occurs less through controlling the flow of information and more through filtering and interpreting it.  The individuals with more ties are those to whom others look for leadership.

A woodland owner example here might be downturns in stumpage prices.  Everyone might be aware of the market conditions, but would look to one another for help interpreting the information, speculating about future conditions, and deciding how to act.  Although all members have access to the same market information, some are clearly more influential than others.  For instance, in the figure at right, the individual at bottom center is consulted by many more of his neighbors than most others are.

The nature of an actor’s influence is not easy to quantify, and of course varies based on the community and situation.  One common decision rule is the threshold rule, which posits that a given individual will adopt a behavior when a certain percentage of her contacts has adopted.

Extension applications

These models, while sharing some common elements, are quite different.  Understanding which model applies, if either does, is obviously important to program design.  Classic master volunteer programs are based on the two-step flow model.  This model is well entrenched and demonstrated to be efficient and effective.

However, for some types of information and behaviors, new media may penetrate more deeply into a community, flattening the information hierarchy.  In these cases, landowners may be less dependent on others for information and more so for interpretation, discussion, and processing

Which of these models applies better to your situation?  Does thinking about diffusion in these ways suggest changes in the way you reach out to your key audiences?

Full citation: Watts, D.J. and P.S. Dodds. 2007. Influentials, Networks, and Public Opinion Formation. Journal of Consumer Research. 34(4): 441-458.


4 Responses

  1. Hi Eli,

    I think you raise some interesting questions, that could easily be converted into testable hypothesis both in terms of research and evaluation. I, for one, am not convinced that current programs radiate as far as we’d like to believe given some of the research we’ve done and are working on publishing. Social structure is a funny thing that isn’t near as malleable as we’d like it to be, and is more contextual than we would prefer.

    Specific to diffusion of innovations, the challenge, from my perspective, in application rests not such much in the social processes, but in the definition of the innovation. Rogers spends a fair amount of time defining the characteristics (e.g., trialability, observability, etc.) of an innovation that is likely to diffuse rapidly versus those that are not. In this way, there is a difference between an innovation (e.g., forest certification) and information (e.g., prices, good loggers, etc.).

    There are, of course, other theories, that might inform how extension/outreach/technical assistance might operate in a peer-to-peer mode (e.g., social learning, exchange theory, adaptive management, community, etc.), but relevance may depend largely on what knowledge and/or behavioral changes are desired (as well as other assumptions).

    Thanks for this blog and your willingness to foster discussion. I’d not the seen the paper you mention, so we’ll be taking a look.


  2. Mark, thanks for the comment. It’s nice to see you posting here–you are, after all, the first person I know of in this domain to have started a blog. I hope you’ll consider resurrecting it, or perhaps contributing to this one. We’d love to have you contribute new posts.

    You raise a good point about the nature of the innovation. I have another post brewing about that. More soon…

    As for other theories, the two I’ve been exploring are social learning and social influence. These two theories bear some resemblance to the belief and normative components (respectively) of the theory of reasoned action. I don’t plan to apply the TRA, but am interested in which, or to what degree, social learning and influence affect behavioral adoption in a private forestry context.

    Past TRA research (Young and Reichenbach 1987 comes to mind) suggests that the normative component is relatively minor factor and that social learning might be more important. Although this seems reasonable, I wonder if social influence might play a larger role than they found.

    Why? One problem with data on the normative component of the TRA is that it’s self-reported. Many woodland owners self-identify as independent decision makers, and hence might minimize (perhaps unconsciously) influence from others. I wonder if different results might be obtained by proxy measures like network density (ref. Kohler, Berhman, and Watkins 2001)?

    We’ll see where this all goes. Thanks again Mark for taking the time to comment.

  3. TRA?! We have a survey out now using Theory of Planned Behavior (TPB) to assess landowners’ intentions and behaviors vis-a-vis forest practices (see http://fwe.wisc.edu/facstaff/rickenbach/small-scale.html). We selected TPB over TRA as it includes “perceived behavioral control” which, given economies of scale, access to sources, etc., seems a reasonable constraint on landowners–particularly those with small landholdings. We should know more come spring.

    Interesting you should mention combining TRA/TPB with SNA as I have a student who is a presenting just such a model to our working group here later this week. I’ve not seen it, so I cannot say much at this point. However, your hypothesis about social influence is, I think, important and needs to be more fully vetted.

    As you can see, when it comes to blogging, I am better at reacting than posting.


  4. Interesting point about constraints on small-scale owners’ ability to adopt forestry behaviors. I’d been thinking of the perceived control component in different terms, but this sounds very interesting…

Leave a Reply

Fill in your details below or click an icon to log in:

WordPress.com Logo

You are commenting using your WordPress.com account. Log Out / Change )

Twitter picture

You are commenting using your Twitter account. Log Out / Change )

Facebook photo

You are commenting using your Facebook account. Log Out / Change )

Google+ photo

You are commenting using your Google+ account. Log Out / Change )

Connecting to %s

%d bloggers like this: