Fathers and Daughters, Exercise, and Scientific Rigor

Can fathers have a significant effect on physical activity in the lives of their daughters? This is an important question. Because right now, girls entering secondary school often don’t have fundamental movement skills that predict lifelong physical activity. Though we have plenty of data to say that fathers more often participate in physical activity with their sons, we have little data on what to do about it. So Philip Morgan and colleagues devised an intervention for fathers and daughters. But even better, they tested it in a randomized, controlled study.

What they found was a significant increase in physical activity levels for girls participating in an eight-week program with their fathers. Also, they found that these improvements seemed to stick. Nine months after the program, these girls were still more active than the control group.

But What About Clustering?

Anytime you randomize groups of subjects instead of individuals in a study, you have a cluster randomization. And that can cause problems for interpreting the results you get. It cuts into the statistical power of the study.

Because this was a family-based study, it was really a randomization of family clusters. Not of individuals. But the original manuscript said “the analyses did not account for clustering at the family level.” So Carmen Tekwe and David Allison contacted the authors and wrote to the journal to identify this important gap in the published analyses. It’s important, because in some situations, it can invalidate a study’s results and even lead to a retraction.

A Good Resolution

In this case, the response was more straightforward:

We appreciate the opportunity to clarify that the analyses presented in our article were indeed adjusted for clustering and it was an error on our part to report this was not the case.

In other words, they admitted their mistake. They had the correct analysis in hand. And the results were robust enough to hold up in the more rigorous analysis.

This was a good resolution, but not merely because it didn’t invalidate the original finding. It’s a good resolution because colleagues could raise an important issue and the original researchers could deal with it head on. This is how scholarly dialogue, scientific rigor, and scientific integrity can work.

Click here for the original study, here for the comment from Tekwe and Allison, and here for the correction from Morgan et al.

Father & Daughter, photograph © Dean White / flickr

Subscribe by email to follow the accumulating evidence and observations that shape our view of health, obesity, and policy.

December 29, 2019