This study has been getting picked apart on Twitter, and one thing (of many) that drives me crazy about X is having to analyze a scientific study in a 13 part thread dense with charts
Keven Hall starts his this morning, "A paper by ~AdrianSotoMota ~nicknorwitz ~davidludwigmd et al. reanalyzed data from our previous study, claiming to have invalidated our primary findings, & concluding that our data support the carbohydrate-insulin model (CIM) of obesity.
Our response: Despite generating lots of noise, their reanalysis never addressed our study’s primary outcome: differences in energy intake when the same people ate low carb (LC) vs low fat (LF) diets. Our result was valid & there was no significant effect of diet order.
Instead, they attempted to analyze whether the group of participants assigned to the LC followed by the LF diet had different outcomes compared to those assigned to the reverse order. That’s a separate but interesting topic. Unfortunately, their reanalysis had several fatal flaws….and then he lists them in excruciating detail, shredding the authors. ….
Overall, the reanalysis paper exhibited poor scholarship by failing to engage with our prior work, suggested their analysis was more novel than it was, ignored the within-participant study design, failed to disclose the possibility of bias, & committed many statistical errors.
……
In summary, the fatally flawed reanalysis by ~AdrianSotoMota ~nicknorwitz ~davidludwigmd et al attempted to resurrect the CIM by following the adage "if you torture data long enough, it will confess to anything". In this case, data torture led to a false confession of CIM support.
There's more, but you get his point.
I stopped believing in the CIM a few years ago so I ignored this study, but response is worth sharing.