Four Paths from Links to Cause and Effect
Every day, research finds new links to obesity in behaviors, nutrition, the environment, and an array of other factors. And every day, headlines falsely suggest a cause and effect relationship in those links.
A recent example is a widely reported link between late bedtimes for preschoolers and obesity in teenagers. The headlines tell parents to tuck their little darlings into bed early or they’ll grow up to have obesity. And then buried near the end of the report, you’ll find this disclaimer:
The research hasn’t proved that later bedtimes directly cause obesity, only that there seems to be some connection between the two, the sleep scientists agree. Research on this point has only just begun.
This week, NHLBI and the UAB School of Public Health presented a crash course in ways to strengthen the understanding of cause and effect relationships in obesity research. Four tools play a key role in this process:
- Observational Studies serve mainly to identify factors with links to obesity. That study of bedtimes was an observational study. Such studies are a great place to start, but they don’t often provide the final word on cause and effect.
- Randomized Controlled Experiments (RCEs or RCTs) are the gold standard for proving cause and effect. Researchers randomly assign people into a group that gets an intervention or a group that doesn’t. Such experiments provide great proof, but they’re not so easy to do. Imagine randomly telling parents that they have to keep their preschoolers up late. And then there’s the problem of following up with those kids when they’re teenagers.
- Quasi Experiments provide an alternative to to RCEs, but the assignment to an intervention is not truly random. Nonetheless, a well-constructed quasi-experimental study can provide important information about cause and effect.
- Natural Experiments occur when forces outside the control of researchers randomly assign people to an intervention or a control group. Finding natural experiments requires creativity and careful attention to factors that might compromise the randomness of the experiment.
These tools are simply a starting place for solving the puzzle presented by links to obesity. The commitment to solve those puzzles is what’s essential. Serious errors come in both policy and practice when people stop asking questions after a link is found.
Click here for the presentations from the first short course on causality in obesity research at UAB. Presentations from this year’s short course will be available here, soon after the course is complete.
Subscribe by email to follow the accumulating evidence and observations that shape our view of health, obesity, and policy.
July 28, 2016