Not Bad, Just Wrong
Why theory and interventions have to be developed together
This weekend, two giants in the field of Behavior Change Research, professors Martin Hagger and Mike Weed had a debate at the 2018 Meeting of the International Society of Behavioral Nutrition and Physical Activity in Hong Kong. (Oh, you didn’t know? Did you miss the Super Bowl commercial?) The topic was pretty controversial:
Do Interventions Based on Behavioral Theory Work in the Real World?
This is not a stupid question. As Dunton (2018) points out;
Evidence is lacking on intervention strategies to maintain these health-protective behaviors over sustained periods of time. Intervention studies promoting physical activity and healthy eating typically focus on the initiation of these behaviors for periods shorter than 6 months among individuals who were not previously performing them. Even when successfully adopted, new patterns of behavior are not maintained over longer follow-up intervals and typically regress to baseline levels (Wood & Neal, 2016). There is limited evidence on how to help individuals avoid temporary lapses in behavior.
Large-scale intervention studies looking for big effects in large, heteronomous populations haven’t done well in the health domain. So it’s not a stupid question to ask, “do interventions work?”
It’s just the wrong question.
Let me ask you a similar question: “Does penicillin work?”
“Of course penicillin works,” you say. “It kills bacteria.”
Woah, slow up there, cowpoke. You’ve just told me a mechanism. A mechanism based on a theory called “The Germ Theory of Disease.” A theory that was originally proposed by Girolamo Fracastoro in 1546, but was openly mocked by the medical profession—who believed that disease was caused by “bad air”—until about 1890. Which meant 344 years of interventions like perfumed scarves and cholera belts.
Now imagine I ask you “does mould prevent fever?” in 1640, before the germ theory of disease is widely believed. You probably would laugh and say, “testing that would waste our valuable funding dollars. What does mould have to do with miasma?”
I might tell you that Polish pharmacists have been mixing wet bread with spider webs to treat wounds. And that Sri Lankan soldiers have carried mould-covered oil cakes with them on campaigns to treat their wounds since 150 BCE. Or that the Egyptians, Greeks, and Indians were using mould on wounds for centuries before that.
This somehow convinces you that it’s worth a try, so we take all we know and we start to develop a Randomized Control Trial.
But what do we test? And on whom? And compared to what? Without a theory to inform our choices about what to test on whom and in what doses, we’ll just be blindly guessing. For example, we could put wet, mould-y bread on 10,000 leg amputations and see if that works better to prevent fever than just keeping the wounds clean (sneak preview: that might kill more people because the the amount of Penicillium in mould-y bread is minuscule).
Would our RCT have “proven” that mould doesn’t prevent fever?
The (Sometimes Violent) Dance of Progress
The problem is that it took 382 years of co-development for new tools (like the microscope), new theories, and new interventions to lay the ground work for Alexander Fleming to isolate, discover, and test Penicillin in 1928. It took Agostino Bassi proving that a microorganism can cause disease in 1808. It took Ignaz Semmelweis proving that washing hands could prevent puerperal fever in 1847. It took Louis Pasteur seeing bacterium in a microscope 1864 and Robert Koch proving those bacterium caused diseases in 1884. All this plus thousands more discoveries, tests, hunches, and stories had to be in place for Alexander Fleming to “accidentally” discover that mould-juice from a fungus could prevent a bacteria from growing.
And that was not 382 years of linear progress. After eliminating deaths from puerperal fever in his obstetrics ward by forcing doctors to wash their hands between handling corpses and delivering babies, Ignaz Semmelweis was not given the Nobel Prize for Medicine—like Alexander Fleming would receive in 1945. No, Semmelweis was fired, mocked out of the profession, forced into a mental institution, tortured, and beaten to death by guards in 1865.
Theory and intervention-testing need to work together to inform answers to the important questions of science: what does what, on whom, under what conditions, in what doses, and with what side-effects? All of these questions are implied in “does it work” and it takes a developing, testable theory to work all those out. Without theory, we won’t even know what to look for. In summary:
Intervention-testing without theory is missing the point.
Theory without intervention-testing is pointless.
To put things in context, it took 382 years to develop the Germ Theory of Disease and “discover” penicillin. William Wundt, the father of modern experimental psychology, published his textbook Grundzüge der physiologischen Psychologie in 1874. That’s just 144 years ago.
So psychology is only 144 years old and there’s already 93 Behavior Change Techniques, 26 proposed mechanisms to explain how they might work, and 83 widely accepted theories of behavior change. That’s kinda nuts. We need intervention-testing to help us winnow these down and improve the theories we have. We need better theory to help develop more insightful experiments. Talking about things as, “either / or” is getting ahead of ourselves by about 238 years.
Steven M. Ledbetter is the CEO of Habitry, a behavior change consultancy that helps companies make more impactful products.