Follow TopicFollow Contributor Share Feedback
An Interview with Damon Centola and Betsy Levy Paluck

An Interview with Damon Centola and Betsy Levy Paluck

by Damon Centola, Betsy Levy Paluck

May 15, 2022


Q: Damon, in our report last year you described “complex contagions,” making the observation that, while most people think the key to driving change lies in identifying influencers who can advocate for an initiative, in fact, change is less so about finding the “right people” and more so about having the right “contagion infrastructure.” What does this mean and how is it relevant in a management context?

Centola: The rational choice model has come under heavy criticism, and the shift has been to say that, if people aren’t rational, there must be something more generally cognitive going on that we can define as “behavioristic” and address through “nudges.” There are unconscious biases and heuristics we can seek to understand which will then tell us how people will act.

To me, that approach is as narrow-minded as the rational choice perspective. Because the assumption is that a person is an isolated actor: give them the right stimuli, they’ll make the right choice. In the rational choice model, the important stimulus is usually taken to be information. In the behavioral model, it’s triggering parts of the brain conditioned by evolution to respond in certain ways. But that’s still methodologically individualist and overlooks the fact that a lot of the decisions people make are based on the people around them.

The orientation that Betsy and I bring to this discussion does not see people as so fixed. There’s tremendous plasticity to behavior, conditioned on social factors that can be studied, understood, measured, and treated. They can be intervened upon. People don’t like the term “social engineering.” They don’t like the idea of being manipulated. But it happens all the time. It’s what governments do to produce desired social outcomes and, when they fail to do it successfully, everyone throws up their hands and says, “what’s the point of government?” But these failures usually reflect bad policy. The idea is to figure out how to correct our policies based on our best science, and our best science has moved significantly beyond the individual.

For a long time now, a common intuition has been that, if we just give people access to the right information, they’ll make smart decisions. And other than broadcast media, one of the ways of disseminating that information is through social networks. For three-quarters of a century, research has been instructing us that information-spreading is the key to a better-informed population and better decision-making. More recently, we’ve taken this idea and applied it in terms of “influence.” Even if people don’t have all the information, if you give them certain information, then that will influence their decision-making.

But this thinking basically views social networks merely as pipes: conduits of transmission. This is consistent with the way we think of a disease spreading: I come into contact with you; you transmit the disease to me; and then I transmit it to somebody else. Hence the idea of a ‘super-spreader.’ If we get someone who’s highly connected, they’ll come into contact with lots of people and they’ll have a very high probability of transmitting the disease to all of them. Translated into the informational context, we think of someone like Beyoncé. With all her social followers, we believe that if she says something to those followers, or does something that’s seen by them, then that information will spread out to lots and lots of people and alter their behavior.

The important change in thinking here is that, when it comes to behavior change, and belief change, our social networks serve as prisms.

The important change in thinking here is that, when Betsy or I look at social networks, they’re not seen as pipes. When it comes to behavior change, and belief change, our social networks serve as prisms. They refract and shape the information we get.

Many people exhibit behaviors that aren’t part of any information spreading process, but those behaviors nevertheless shape how we interpret information. The people around us affect whether we are ready to believe new information or reject it, or whether we think it’s ultimately irrelevant to us. These networks of influence around us are not even noticed if we’re only looking at pure information-diffusion. But it’s those networks that do the majority of the work influencing our behavior, determining how information affects us once it does get to us, and influencing whether that information reinforces our existing belief in some idea, or helps to accelerate a change in opinion.

If we’re going to think about organizational culture, we have to think about how these networks govern our coordination in ways that are largely invisible to us

The complex contagion perspective centers on the structure of those social influences around us that really shape, and control our receptivity to information — and invisibly governing our behavior. In organizations, of course, this happens more powerfully than in everyday life. In everyday life, we’ve got stronger influences and weaker influences all around us, people we interact with more, like our family. [See also The Academy Article Trust and the Scale of Management] But within an organizational context, we are principally coordinating with small groups of people. That process of coordination governs many of our decisions, and it happens invisibly. Let me explain.

In an organizational context, it’s common to describe the big challenge in terms of a cooperation problem. But the cooperation problem is something we’re aware of. It’s something we think about. We can see when there’s a tension between our personal interests and the collective interest of the group.

I think the coordination problem is more interesting because it appears trivial. It’s less obvious, because we’re coordinating all the time, without even noticing it. Things like ‘implicit biases’ are not typically part of cooperation problems, because we cogitate over those things. But, with problems of coordination, it’s like the water a fish swims in; we don’t even notice it. Think about the things that make us feel comfortable, the things that make our peers feel comfortable. That’s what governs our behavior in implicit ways.

So, if we’re going to think about organizational culture, we have to think about how these networks govern our coordination in ways that are largely invisible to us. That’s what complex contagions are: forces of social influence that can shift the entire culture of an organization without anyone really knowing that they’re changing their behavior. And — precisely because we’re not aware of it — we need to refocus our attention towards it purposefully.

A scientific awareness about how that process works is quite powerful. It means you can change the culture of an organization without needing lots of motivational speeches and seminars to get people excited, nor does it require compliance and monitoring processes to police subsequent behavior. Instead, you can put people in a context specifically designed to shift their behavior in a desired direction. It’s hard to understand that such interventions are happening all the time. It’s hard to point to invisible structures and processes. But those structures and processes are there and, as scientists, we can measure them.

Q: Betsy, in this report, Amazon Web Services' Michael Arena describes a “natural interplay” between organizational culture and organizational social networks—one which isn’t typically factored into culture change initiatives. Your research suggests change to culture and behavior can be advanced through such networks. Would you please explain this? What led to your research? What are some of the most important or surprising learnings to date?

Paluck: Well, first of all, let’s take it back a step and recognize the importance of culture. When those who have to answer for the effects of some particular organizational culture think about trying to change that culture, they most often think about top-down processes. They think about institutional arrangements and rules — and I certainly think of that, too.

There's a lot of research showing the importance of mapping those organizational networks if we want to understand how culture is produced, reproduced, and shifted over time.

Maybe one reason people don’t think about networks quite as much is that the networks that people like me highlight don’t involve people doing top-down work.

It’s hard to imagine how people who aren’t at the top of the organizational ladder can have much to do with culture. But there’s a lot of research showing the importance of mapping those organizational networks if we want to understand how culture is produced, reproduced, and shifted over time — by people within the organization who are not associated traditionally with top-down cultural power.

My research is focused on networks of attention. Within every organization and community, there’s a certain amount of attention that’s going up towards the top. We look towards traditional leaders — those with the offices, the roles, and the titles. But we also pay a lot of attention to our peers, horizontally, and even to people who may be ranked below us. Why?

Well, we look to those people because we enjoy them, but we also look to those people for information. We have a sense of their place in the hierarchy, their role and status. Depending on what we observe them doing, we get a lot of information about the culture of the place. Many of those at the top of the hierarchy find that their behavior is pretty constrained in many ways. So, if you want to know what kind of humor is really allowable at an institution, if you want to know what kinds of attitudes are representative of the place, then you might not look toward the top, where the language is very constrained. Instead, you look to your peers and the people below you in the organizational structure, who are seen as maybe more genuine. 

These networks of attention are really important. They tell us about the kinds of information that people are taking-in most regularly, about the culture of a place, the perceptions of what’s typical and what’s desirable in terms of behavior. In my own research, that’s what we look for. People who get a lot of attention have an outsized influence on our ideas about what’s expected in any given organization. And, to me, that’s what culture is really about.

For anyone interested in culture change, it’s important to know what networks of trust are at work in the organization.

A leader can tell you what’s typical or desirable, but if you look around and that’s not how you see people acting, then it doesn’t matter what people are saying in a top-down direction. What matters is what’s happening in that attention network. It’s important to recognize that the people to whom you’re like to refer, when seeking cultural information, are usually those found in identity-based groups. They’re the people you trust, because you share an identity, or a role.

Trust is really important in those instances, because you are trying to find out how to fit in, to learn how to excel, how not to violate the norms of a place. You want to be able to trust that the people to whom you’re looking to for advice, or those whose behavior you want to observe and mimic, are in fact people who can give you good information. For anyone interested in culture change, it’s important to know what networks of trust are at work in the organization. What are the networks of attention within the organization? You need to know that because that’s where you want to target your efforts to drive change. You need to shift what people see within those networks if you want to change their impressions regarding the norms of a place.

When you're talking about the infrastructure of an organization, the key question to ask is, “what is the connectedness between groups that are functionally very different?”

Such a “reference group” could include the people with whom you sit within a particular wing of an institution — like the office or a school. But it’s more likely that you’ll skip over some of that geography. There are people to whom you are connected in ways besides proximity. They’re the ones you have a drink with after work, those you email or text regularly throughout the day. It is those within these networks of trust from whom you’ll seek information about how to behave, about what kinds of behavior will be rewarded in your day-to-day life at that place.

Q: Damon, in How Behavior Spreads, you write about the importance of trusted relationships and argue that managers can make use of organizational social dynamics, “to design organizational networks that improve institutional capacities for innovation and adaptation.” Can you offer concrete examples and explain how trust is relevant here?

Centola: When you’re talking about the infrastructure of an organization, the key question to ask is, “what is the connectedness between groups that are functionally very different?”

How much discourse is there between the marketing and the engineering groups, for instance. How much information flow? The classic idea of a network “bridge” is that there’s a person, let’s say in engineering, who talks to a person, let’s say in marketing, and so we expect there to be information flow between them. But that’s a very narrow bridge! It means that, if the engineers are trying to get the marketing people on board with a new program or new idea, one person has to convince a lot of other people to change their way of working.

This is probably not going to be very effective. But now imagine that there are more connections between those groups, that several people in engineering go to lunch with several people in marketing and talk informally with them and get to know them. There’s opportunity for more trusted relationships to develop there. So, when the engineers talk about this innovative idea that they’re trying to get marketing excited about, there are several people in marketing who can independently evaluate what the engineers are saying, and then go back and talk to each other and coordinate within the marketing group to try to get the other marketing people excited about it. By increasing the “width” of the bridge between these different groups within the organization, you increase the degree of exchange and the depth and quality of knowledge that can be shared across these groups.

And that comes as a result of trust. If one person is trying to convince people in a different group to take something up, it looks like they have a hidden agenda. And, oftentimes, if that person is a network “broker,” they may well have an agenda, because the more ideas they can spread around the organization, the more credit they’ll get as the broker for those ideas, and the more attention they attract, the more likely they’ll be to get promotions. That broker also has an incentive to keep his or her social ties exclusive, because their power in the organization comes from the fact that they’re the only engineer who talks to people in marketing.

The question then becomes, how does it affect that person if you suddenly develop more connections between engineering and marketing? It basically reduces that person’s power in the organization, because it increases the organizational capacity for those two groups to talk to each other. This is where adaptation comes in. If someone who is a “narrow bridge” leaves an organization, the question then becomes, how does the structure of the organization change? How stable is it? If a broker who has status in an organization can attract offers from other organizations, part of their power is, if they leave, they also take away a ton of infrastructure with them.

This is where the redundancy of social ties, and the role of “wide bridges” becomes really interesting. With multiple redundant ties, and greater bridge width across different groups, individuals may come and go but the social infrastructure stays consistent. So, you can still have a flow of ideas and coordination and communication across these different groups, which means that as there’s employee turnover, there’s nevertheless a stable infrastructure that allows people to share knowledge, and this allows the organization as a whole to be much more adaptive than an organization with “narrow” bridges.

Q: Betsy, you wrote a recent paper on engineering social change by using social norms1, calling them ongoing group processes rather than a set static beliefs. By developing an understanding of these processes, you suggest, we can devise interventions to promote (or discourage) specific norms with ‘predictive’ reliability. That sounds difficult to apply in a risk or compliance context. Is it?

Paluck: Any intervention that is going to have predictable effects on people’s behavior are not going to be the lightest-touch interventions. 

Many people have tried to run with this idea of “social proof.”2 One light touch way in which some have tried to really run with this idea is to ‘message’ people, rather than trying to provide them with real social proof. We’ve all been messaged this way. We’re told that 80% of the people in our neighborhood are going to go out to vote today (hoping that this will prompt us to do likewise), or that 90% of the people in your age group in your zip code have been vaccinated, (so why haven’t you done so yet?). What’s challenging about this approach is that it doesn’t always match up with your experience, what you see in your reference network, in your habitual environment. So, such messages can be safely disregarded, we conclude. So, these kinds of culture change initiatives have had very mixed effects.

What does it take to change people's behavior? Well, first we need to understand who people trust.

Sometimes you get a big bang for your buck, and I think that is especially the case when we’re talking about hidden behaviors. “Oh! I had no idea so many people in this area have gotten vaccinated. Wow!” Maybe that leaves you more open to getting the vaccine as well. But a lot of the behaviors that we want to change are not hidden. There, you need a different kind of social proof. You need to discover where in the network you should target your efforts so that a particular witnessed behavior provides a ‘right-in-your-face’ sort of social proof.

What does it take to change people’s behavior? Well, first we need to understand who people trust. Consider people who care a lot about the adoption of new technologies. One of the most powerful things they’ll do is to sit down with a friend who’s using the Big New Thing and ask, “What’s that?” And their friend says, “Oh yeah, let me show you this. This is what I love about it. This is so cool. This is how it’s changing my life right now.” That’s a kind of adoption that we don’t often think about when we’re trying to spread more ephemeral things, like responsibility, or anti-corruption practices. But we need that kind of social proof for the more ephemeral kinds of culture change as well. It’s not all going to be light touch.

We’re at the point in our science where we have an understanding of networks and of norms such that, if we’re going to invest in culture change, we know where to look. We have really good ideas about where and how to intervene to drive change. And that’s how we’ve used the word ‘predictive’ in my research.

For anyone trying to apply these ideas in a risk or compliance context, my counsel would be to take a bottom-up approach. Who are the people you want to see adopt the relevant behaviors?

In that paper we were looking at teachers, and one of the big lessons for us was that even those teachers who are observing the network closely every day don’t see the same networks that the students experience. What we observed was that teachers couldn’t pick out the same influential people that students could pick out. There was some overlap, but shockingly less than you might expect. And the people who teachers did point to as being influential were the kinds of people who appealed to teachers. They were influential in the eyes of teachers, though perhaps not so in the eyes of students. We all have our biases in terms of who we pick out as influential folks.

For anyone trying to apply these ideas in a risk or compliance context, my counsel would be to take a bottom-up approach in mapping out the network of folks who will be involved in this kind of exercise. Identify who among them are the most influential people, and then check out their behavior. As a social psychologist I would say that you know a new behavior has become rooted in the network when some nodes — those who’ve have adopted it — become more highly visible to everyone else. By metaphor, if the “teacher’s pet” is doing it, that certainly doesn’t mean that everyone else is going to adopt it. So, in your compliance context, who are the people you want to see adopt the relevant behaviors? I would take more unconventional methods in trying to identify where adoption of the new behaviors you want to see is most likely to take hold.

There are a lot of different behavioral traces that we drop all around in an organization that give us a clue about where our attention is going. Sometimes we use surveys. We look to see who people report spending their time with or asking questions of. And you can do an analysis of emails to see who people are paying attention to. Who do they consult regularly? Who are they calling, texting, or Slacking?

Q: Damon, in Change: How to Make Big Things Happen3, you make the provocative argument that we are more influenced by the percentage of people we know who opt into a particular behavioral norm, rather than the raw number of people seen doing so. Can you explain this, in the workplace context?

Centola: Think about the validity of an “urban legend.” If one person tells you an urban legend, you’re probably going to say it’s not true. But if you hear it from several people, you might start to think there’s some credibility to it. The people who aren’t repeating the story may not have heard it, so their silence doesn’t signal anything in terms of disbelief or resistance to the idea. Social confirmation has a strong, positive effect, but silence doesn’t have the same negative effect. Just a couple people can confirm the rumor and get you to repeat it. 

But it’s very different for any kind of visible change in behavior, because the people who are not seen to be engaging in it are implicitly against it. If some members of your office have started using a new technology, or are trying to advocate for a new project management strategy, and you see all the peers around you but for two are sticking with the old strategy, then those two adopters are not very convincing. The effort to initiate change is weighed against all the people who are seen still to be doing the same old thing. Essentially, the non-adopters are an implicitly countervailing influence against change. 

This is one of the key insights of complex contagion: it’s not just that we need multiple reinforcing signals to be convinced that something is a good idea, we also need to be convinced that the idea is ‘legitimate,’ and that involves looking at everyone, not just the adopters. To succeed in driving change, we need to think about getting enough social reinforcement so that the visible adopters outweigh the non-adopters— ultimately making the non-adopters feel like they’re the ‘illegitimate ones,’ the ones who need to catch up with the curve of change.

Q: Betsy, much of your past research is focused on prejudice4, how it shapes behavior and how this can be changed when bad outcomes result. You argue that perceptions of what is believed to be a desirable behavior within a group are more important than any actual knowledge about what behavior is in fact condoned. Perhaps that’s that ‘prism’ effect that Damon was just describing? Can you say a bit about this? And I’d be interested to know how you see this playing out in the context of DE&I initiatives.

Paluck: Damon just illustrated beautifully the landscape of how we would approach these kinds of problems. And notice what he didn’t say. He didn’t argue that we need to go in and try to convince each and every individual to change their beliefs or their behaviors. He didn’t describe working individual-to-individual to implement some sort of re-education programming. Because people exist within these social networks, they pick up and apprehend behavioral messages as a collective, rather than as individuals.

Because people exist within these social networks, they pick up and apprehend behavioral messages.

A critical bit of ‘connective tissue’ among us is our shared reality: our collective understanding of how we are to treat one another, our standards for acceptance, our shared practices for how to get work done and how to apportion responsibilities, how to strike a balance between rights and responsibilities within an organization. It’s all about these collective understandings.
And, to reiterate from Damon again, that is something that we can measure, something we know how to target. A lot of DE&I efforts have focused on the education and persuasion of individuals. The kinds of interventions I’ve been interested in focus instead on these shared understandings, the shared realities that characterize the networks within an organization. That represents an opportunity for a different kind of management intervention, one less so about workplace training and more so about shaping collective experiences and practices.

Q: Damon, you’ve also written that, as a network becomes more conducive of simple contagions, complex contagions are likely to become less common among that group. I can’t help but think that this is highly relevant in the context of new ‘hybrid’ work environments. Your thoughts?

Centola: The way to think about this is that there’s an economy of ideas. There’s only so much stuff we can ingest during the day. So, the stuff that’s easier to talk about and that therefore spreads more quickly is going to have an advantage in terms of affecting people’s beliefs and ways of doing things. If a change is initiated that requires social reinforcement, and if that faces countervailing influences, then the change initiative is competing for social network bandwidth. The more content that’s flowing around, and the easier and more familiar that content is, the less bandwidth we have for digesting new and complicated ideas.

Another way to think about this is to note the difference between information transfer and knowledge transfer. Information transfer is simple: you tell people things, and they repeat it. There’s lots of that flying around in any given the day.
But if we’re after knowledge transfer, where people will have to think differently about what they’re doing, adjust their strategies, or adopt new practices, then the effort required to spread that kind of knowledge is in competition with all the other informational detritus that’s floating around. The ‘economy’ of bandwidth preferences things that are easier to spread — that is, the things that are familiar, the ideas that reinforce our normal way of doing things.

So, if the communication channels in the hybrid workplace are narrowed by virtue of the fact that there aren’t as many informal opportunities for knowledge exchange, if communications aren’t as rich or detailed or personal as would be the case in a face-to-face environment, then that adds a further advantage for the simpler, easier things to propagate.

There are so many features of face-to-face communication that allow us to be more compelling in our articulations, that make people more emotionally receptive to things that may challenge their existing way of thinking. Narrowing down the communication channels in the way COVID has, limits bandwidth and gives further advantage to the things that are easy to talk about and easy to spread.

Q: Betsy, in another recent paper, about “network insiders and observers,”5 you explore who can best identify those within organizations who can truly influence peer conduct, arguing that this is critical to successful behavioral interventions. Would be important in the context of conduct risk management? How might senior leaders use these ideas to shape risk governance outcomes?

Paluck: We’ve mapped dozens and dozens of networks in schools. These are highly complex organizations, with well-established hierarchies. They do churn, to some extent, but we’ve been able to study these networks closely, looking at how students decide upon the cultures that prevail at their schools. Sometimes, those are cultures that their principals and teachers approve of. Sometimes, it’s the kind of culture that the higher-ups are working to change.

But it’s very hard for teachers or administrators to change a student culture. Because their patterns of behavior, their perceptions of the norms of the place, are created by the students themselves — and they’re very sticky. Students perceive what will make or break them at school, what will reinforce their reputation, or cause them to lose face. So, teachers and principals instituting new rules just doesn’t matter much — other than perhaps to identify who it is that’s going to be punished more often.

A goal of our research was to reduce cultures of abuse in these schools, cultures of violence, or bullying, or harassment. What we find in mapping these school social networks is that there are certain nodes in the network — particular students — who receive a great deal of peer attention. And sometimes, the most effective way to change the culture of a place is to work with these people, the “trouble-makers” who are often part of the problem.

It's very hard for teachers or administrators to change a student culture. Because their patterns of behavior, their perceptions of the norms of the place, are created by the students.

You would think that what you’d need would be someone in the network who can regulate those people. But, in fact, involving those troublemakers in culture change projects is what has proven most effective. It comes back to the fact that they are so highly observed by peers that everything they’re seen to do is ‘overweighted’ amongst their social connections, shaping again what would make or break your reputation at that school.

It always surprises teachers and principals when we come back to them and say, “All right, this is what our network findings tell us. We want to involve this student in our culture change efforts, with respect to conflict, violence, abuse at your school.” And they’d say, “No way! These are the people who get hauled into the office all the time!” And we’d say, “Yeah — precisely.”

And we we’re not targeting them because they need to be remade or reformed. Rather, they know intimately the rules of the place. If you can engage them in a way that would help them to adjust some of their public behavior, even in small ways, in ways that make the school an easier place to be, these people really do have an outsized effect. We’re able to measure and compare the effect of their involvement in our campaigns. Their involvement has a much bigger effect than the involvement of other students selected from the network who were perhaps more motivated to help change their school’s culture, but who didn’t have that outsized impact.

It is counter-intuitive: we’re usually eager to work with the people who seem motivated to drive positive culture change. It’s not that we shouldn’t involve those people, but sometimes it’s going to the people who have the eyes of their peers on them — even if they’re deeply involved in the problems themselves — we find that these people have a great influence.

Q: Damon, in How Behavior Spreads6, you write that by understanding the network structure of a population we can uncover novel ways of controlling the flow of behaviors across it. Bank regulators often emphasize setting an appropriate “tone from the top,” in this connection, but Betsy’s just questioned this approach to shaping organizational conduct. What does your own research suggest?

Centola: The intuition that, if a leader gives some advice, it will be followed by everyone, is not consistent with what we see when there is a conflict in the norms of an organization. If Unilever acquires Ben & Jerry’s, Unilever has a certain company culture, and Ben & Jerry’s has another. So, whoever has the highest authority can simply say, “we should all work together in the following way.” That’s a nice message, and everyone gets that it’s a nice message, and it sounds good to the board and shareholders. But it’s problematic when subsets of people within the merged companies have different ways of operating, of doing their jobs and feeling competent in their work. 

Simply saying, “we should all try to get along and cooperate on these sorts of tasks,” or “here’s the new ethics of the organization,” doesn’t at all change the way of doing business day-to-day. [See also the Peer Perspectives Article Reimagining Ethics, Risks & Compliance with Behavioral Science at Novartis] From the perspective of how we coordinate and get along, there’s a fundamental issue: we all want to feel like we know what we’re doing at work, we want to feel competent, we want to feel authoritative in whatever we’re talking about. But, that’s really very difficult when someone comes along and tells us that the things we’ve been doing, and the way we’ve been doing them, need to change abruptly.

It's hard to make real changes individually, without seeing that everyone else around you is doing their work differently as well.

While it may sound good to say that changes are important, it’s hard to make real changes individually, without seeing that everyone else around you is doing their work differently as well. Changing workplace routines is like learning a new language. You were fluent in the original language. You could communicate and move in nimble and effective ways. But now people are asking you to speak in another tongue. And you worry you’re going to look stupid or incompetent. You worry you’ll be seen as in effective at your job and be fired. So, there’s a very strong incentive to stick with the things you know how to do and the way you know how to do them, particularly when everyone else around you is doing likewise.

The top-down management approach fails in cases where people are being asked to innovate, or to shift norms in ways that are uncomfortable or different.

If you try to do something new that no one else is doing, because someone at the top told you to do it, you’re just going to look less competent than everybody else. The processes of innovation and culture change are all about people trying to learn a language together, to learn how to speak it without worrying that they’ll be called out individually for failing. If we’re all trying to do a new thing, it removes the risk that we’ll be punished for missteps or stumbles. Without such reinforcement from our peers on this new thing, there’s a tremendous amount of risk that we face. So, unless we’re all going through change together, individually none of us are going to accept it. 

The top-down management approach fails in cases where people are being asked to innovate, or to shift norms in ways that are uncomfortable or different. And fundamentally, that’s because it ignores the realities of day-to-day work in an office, where we are not only coordinating with peers, but we’re also entrenched in ways of doing things that have been successful. Asking us to abandon a successful strategy for something that’s unproven is a high-risk proposition, which requires social reinforcement to succeed.

Paluck: Can just offer some jargon I learned from my students the other day: you’ve got the “grassroots” activity, and you’ve got the “grasstops” activity. I like that, because I think when there is a helpful tone-from-the-top, people within the network can draw upon it to reinforce whatever they’re doing. They may be taking a risk with their reputations by deviating from the status quo, and those people who could be very influential, in terms of promoting the adoption of new practices, they may need just that little bit of a guarantee that their deviance isn’t going to hurt their reputations too much.

So, tone-from-the-top can work in concert with more peer-oriented change. Where you get the perverse outcomes is when there’s only a tone-from-the-top but no substance beyond that. Everyone can see that it’s just the latest slogan, it’s empty messaging. Having the right tone and seeing social proof around you is extremely powerful, but it’s hard to do one without the other.

It’s also a challenge to know how to sequence these things as well. Should you start from the bottom? There’s a lot of very provocative research saying that it might be really helpful to see adoption around you start to happen, even before you see that taken up by leadership. Because that feels more authentic. It feels like there’s a genuine interest, a growing movement towards change, that you’re involved in convincing leadership that you want such and such to be more a part of your workplace, or country, or community. That’s also counterintuitive for people. Leadership might think, “I need to give permission,” or, “I need to set such and such a tone.” But the psychology of it is it says something else.

  1. Deborah Prentice & Elizabeth Levy Paluck, “Engineering Social Change Using Social Norms: Lessons from the Study of Collective Action,” Current Opinion in Psychology, vol. 35, pp. 138-142, 2020.
  2. Shubhangi Roy, “Theory of Social Proof and Legal Compliance: A Socio-Cognitive Explanation for Regulatory (Non) Compliance,” German Law Journal, vol. 22, no. 2, pp. 238–255, 2021. -journal/article/theory-of-social-proof-and-legal-compliance-a-sociocognitive-explanation-for-regulatory-non-compliance /D682206238995D15C1EA5E061B572E9B
  3. Damon Centola, Change: How to Make Big Things Happen. Little, Brown Spark, 2021. -Make-Things-Happen/dp/1529373387/ref=sr_1_1?crid=2OINATB7F71HD&keywords=change+how+to+make+big+things+happen+by+damon+centola&qid=1651610214&s=books&sprefix=Change%3A+How+to+Make+Big+Things+Happen %2Cstripbooks%2C67&sr=1-1
  4. Elizabeth Levy Paluck & Donald P. Green, “Prejudice Reduction: What Works? A Review and Assessment of Research and Practice,” Annual Review of Psychology, vol. 60, no. 1, pp. 339-367, 2009.
  5. Robin Gomila, Hana Shepherd & Elizabeth Levy Paluck, “Network Insiders and Observers: Who Can Identify Influential People?”, Behavioural Public Policy, pp. 1–28, 2020. -insiders-and-observers-who-can-identify-influential-people/4D02E01FAB5EC1DD273A41769B45F2C7
  6. Damon Centola, How Behavior Spreads: The Science of Complex Contagions. Princeton: Princeton University Press, 2018. https://


Join The Discussion

See something that doesn't look quite right?

We strive to provide high quality and accurate content at all times. With that said, we realize that sometimes links break, new information becomes available, or there is something that you feel we may have missed.

If you see something that you think we should be aware of, we would love to hear from you. Feel free to drop us a note below and leave your name and contact info if you'd like to hear back from us.

Thank you for being a key part of the Starling Insights community!