I have been thinking a great deal about cultural stability and cultural change in business, recently. Partly that’s the outgrowth of a report I am working on, and partly it’s the result of the Yahoo ‘no remote work’ brouhaha, which boils down to Marissa Mayer trying to change the culture of Yahoo. So, I thought I would advance a science-grounded discussion about cultural change, and risk an overly long piece. (I haven’t gotten any ‘TL;DR’ messages here, yet.)
First of all, let me submit to you that the tangible result of culture change is behavioral change. For example, if a company went through some series of activities intended to improve meeting hygiene — shorter, more effective meetings, with agendas and action plans, etc. — the proof would be that meetings were measurably shorter, participants hated them less, they led to higher degrees of follow-through, and so on. Yes, leading up to those measurable behavioral changes, cultural change has to start in the minds of the individuals: they have to change their value systems, learn new skills, and reject old ways of doing things. But the proof is in the pudding.
I recently wrote about ‘social deviants’ playing a key role in cultural change (see How ‘positive deviants’ help a culture change itself). Social deviants are not perverts: they are members of a community that already display some set of desired characteristics when most of the other members do not. In that earlier post, I recounted the story of how MRSA — the drug resistant strain of staph infection that plagues many hospitals — was stamped out at a Pittsburgh hospital. The technique was to have the community identify those positive deviants that were already displaying behaviors likely to decrease the spread of MRSA, and then put those deviants into a role of disseminating their practices, so that others could try to adopt them personally. In the Pittsburgh hospital, the MRSA infection rate fell by more than half in less than six months.
The premise behind positive deviant-based innovation is that you can find insiders approximating the behaviors needed for a cultural change, and the community can work within itself to spread those behaviors, and find new ones, and to make the change collectively. It doesn’t require outsiders, except to bring and spread the idea of positive deviancy.
But what is the social network analysis behind this sort of cultural transformation? What sorts of companies are most likely to be able to make cultural changes?
Damon Centola and his colleague Michael Macy wrote a foundational article about this issue, called Complex Contagions and the Weakness of Long Ties. Their work builds on the work of Mark Granovetter, who developed the distinction between the ‘strong ties’ between close friends or kin, and the ‘weak ties’ that exist between more casual acquaintances. Weak and strong are not only relational — referring to the strength of the tie, and the frequency of the individuals’ interactions — but also indicate a structural dimension. Weak ties connect strongly linked clusters — cliques of friends or tightly-knit families — and act as a mechanism for novel information to move from one cluster to another, and once that information reaches a cluster, it spreads to all the members. As a result, Grannoveter called this the ‘strength of weak ties’, and he credits them with being the most important means of information transfer. And information also includes disease, like passing around the newest flu bug, and other social phenomena, like happiness.
So the scenario for contagion is fairly intuitive: on Monday no one in Betty’s office has a head cold. Monday night, Betty attends a meeting of communications professionals, none of which are close friends, but she has a cocktail with a few casual acquaintances, and the new morning has a slight sniffle. She goes to the office Tuesday, but leaves early with a head cold. By Friday, 70% of the office has it.
And it doesn’t require very many weak ties in a city for the head cold to reach everyone. It’s a small world, as Granovetter famously put it. But not all contagion is simple, like a head cold, and so the primacy of weak ties — in other situations — may diminish. Centola and Macy call this the ‘weakness of long ties’.
In simple contagion, only one exposure to the rhinovirus is necessary to get the disease. But other sorts of information transmittal — especially around information that is controversial, advocates risky behavior, or is counterintuitive — has a greater threshold for being passed along. As the authors say,
A contagion is complex if its transmission requires an individual to have contact with two or more sources of activation. Depending on how contagious the disease, infection may require multiple exposures to carriers, but it does not require exposure to multiple carriers. The distinction between multiple exposures and exposure to multiple sources is subtle and easily overlooked, but it turns out to be decisively important for understanding the weakness of long ties. It may take multiple exposures to pass on a contagion whose probability of transmission in a given contact is less than one.
By contrast, for complex contagions to spread, multiple sources of activation are required since contact with a single active neighbor is not enough to trigger adoption. There are abundant examples of behaviors for which individuals have thresholds greater than one. The credibility of a bizarre urban legend (Heath, Bell, and Sternberg 2001), the adoption of unproven new technologies (Coleman et al. 1966), the lure of educational attainment (Berg 1970), the willingness to participate in risky migrations (MacDonald and MacDonald 1974) or social movements (Marwell and Oliver 1993; Opp and Gern 1993; McAdam and Paulsen 1993), incentives to exit formal gatherings (Granovetter 1978; Schelling 1978), or the appeal of avant-garde fashion (Crane 1999; Grindereng 1967) all may depend on having contacts with multiple prior adopters.
The authors enumerate conditions of complex contagion:
- Strategic complementarity — Knowing about some new innovation is not enough to induce people to adopt it. We all evaluate the costs and benefits of an innovation, and wait until we feel it’s ‘worth it’.
- Credibility — When other people we know adopt innovations, we are more likely to do so too. Hearing the same story for different people makes it more believable, and we are them more likely to pass it along.
- Legitimacy — When we see others wearing a fauxhawk or jeggings, we are more inclined to do so ourselves. As the authors point out, ‘Innovators risk being shunned as deviants until there is a critical mass of early adopters’.
- Emotional contagion — Again, it has been shown that emotionality can be passed along in tight social groups, like cruelty and happiness.
Adopting innovative behaviors, like stand-up desks, or dropping older behaviors, like smoking, is more likely in settings where others are talking about or doing the same things. And since the greatest frequency of interaction takes place with close friends and family, people can be pushed over the threshold to adopt new behaviors in settings of higher social density.
Note that this doesn’t necessarily have to involve face-to-face interactions. Online friendships are as real as those IRL (in real life): the same brain chemicals surge when someone give you an ‘attaboy’ in the company chat, as when you are physically patted on the back.
But the big takeaway for me, with regard to cultural change in the business, is that new behaviors are hard to spread when the following conditions hold:
- Workers have few trusted and close company friends
- The new behaviors being advocated are unfamiliar, risky, or contrary to the current status quo
- Very few employees have adopted the new behaviors.
We can consider this the downside conditions for ‘complex cultural change’. The trick for turning it around is demonstrated in the research of the authors, as well. They demonstrated that new healthy behaviors are more likely to be adopted when clusters of people who are all friends or kin meet and discuss on the progress or difficulties they are experiencing in adopting the new behaviors. This is also congruent with positive deviancy.
So, when companies want cultural change, the first step is not to tell people how to change their behaviors. The first step in cultural change is to increase the density of social engagement, or, more simply, to try to help people make more friends at the company. The second step is to find positive deviants — people who are already demonstrating the behaviors that will define the new culture. Then, those deviants and the deviants’ closest contacts should work to share their efforts in amplifying the desired behaviors, and as those ideas become less controversial and more mainstream, others will begin to be more open to adoption as well. But for fast and lasting change, this must all grow from within. As Centola and Macy’s work shows, the weakness of long ties make it very difficult for Betty in the New York City office to be convinced to adopt new marketing approaches from people she doesn’t know and trust in the San Francisco office. She will have to hear about it many times, and with the words coming from trusted mouths.
And this suggests both a path for Mayer to take and why her challenge is so significant. She may think that barring remote work will lead to people making more close connections in the office, but she might have been better off framing her first step as increasing the social density of the distributed workforce across Yahoo, and then as a second act, finding the social deviants inside of Yahoo and let them figure out how to spread that cultural change, themselves. You have to leave cultural change up to the deviants, not to management.
In the Pittsburgh hospital, for example, the doctors — the most well-educated and highest paid on staff — were the worst offenders in spreading MRSA, and the single most effective positive deviant was a low-paid health care aide with only a high school diploma.
You can’t predict where the cultural change will start, but you can predict how it will spread: through strong ties.