It’s appealing, the notion that ideas can spread like viruses, but even those fond of the analogy acknowledge it’s not necessarily a perfect one. A recent experiment might help clarify the matter, though. Studying the spread of a lab test at Northwestern Memorial Hospital’s intensive care unit suggests that, in some contexts, it takes some persuasion on top of exposure for doctors to adopt a new idea.
Although interest in how ideas spread has exploded in recent years, tracking that process in the real world is generally pretty tough. Most of the time, researchers focus on tweets and other social media to study how relatively simple memes spread. But a collaboration led by Curtis Weiss and Julia Poncela-Casasnovas took an unusual approach. Weiss and another physician co-author knew about a new, faster test for bacteria in critically ill patients and asked the NMH lab to supply the new test—without actually telling anyone about it. On a randomly selected day the week after the team received approval for this experiment, the researchers told two other critical-care doctors about the pros and cons of the test, also mentioning that it was already available at the hospital.
In persuasion models, on the other hand, influence is a two-way street, and adoption isn’t an either/or condition. Instead, people have some belief in a new idea’s value, and those who believe in it more are in turn more likely to adopt it.
“This happened in one independent informal talk with each one. After that, we just sat back and collected the data of who was using which test, without any further interference,” Poncela-Casasnovas says in an email.
Over the next eight months, Weiss, Poncela-Casasnovas, and their team tracked the schedules of 36 doctors on the ICU—so they knew who had worked with whom—as well as lab orders for both the new and older tests. Data in hand, they tested two kinds of models: contagion models based on epidemiology and models of persuasion. In contagion models, influence goes one way, from one infected person to an uninfected one, or from someone who has adopted an idea to one who hasn’t. In persuasion models, on the other hand, influence is a two-way street, and adoption isn’t an either/or condition. Instead, people have some belief in a new idea’s value, and those who believe in it more are in turn more likely to adopt it.
Contagion models didn’t fit the empirical data very well, either in terms of the final number of doctors who adopted the new test or how that number had changed over time. “However, the persuasion model not only replicates very well the whole empirical dataset … but it is also good at predicting the future evolution of the process,” Poncela-Casasnovas says. That is, when the team fine-tuned the persuasion model with the first several months of data and kept it running, its predictions matched the real-world observations well.
While the results probably don’t apply everywhere—Poncela-Casasnovas says she’d expect different findings if doctors’ interactions had been random, as opposed to being structured by a schedule—they do suggest there’s more at work than infectious-disease models capture. The findings, she adds, may also help researchers design interventions aimed at boosting adoption of new medical tests or other ideas.
