You Are What You Eat DEBUNKED by a Statistics Teacher

Netflix’s “You Are What You Eat” makes a sweeping case that going vegan improves health markers—but the statistics teacher argues the documentary’s strongest claims don’t survive close scrutiny. The core issue isn’t that the vegan diet produced changes; it’s that the analysis and storytelling lean on selective reporting, while key confounders—especially weight loss and calorie intake—make it hard to attribute effects to “veganism” itself.

The study at the center of the debate randomized 44 twin pairs to vegan versus omnivore diets for four weeks, with meals delivered during that phase and then a further four weeks where participants ate within their assigned patterns. The design is praised for controlling genetics via twins and for standardizing meals during delivery. Blood tests, questionnaires, and microbiome samples were collected, and the primary outcome was LDL cholesterol (reported as the median change from baseline). Vegan participants showed a statistically significant LDL shift versus omnivores, while omnivores changed little.

But the critique sharpens when multiple statistical comparisons and presentation choices enter the picture. The teacher highlights that the documentary and underlying analysis treat results as “significant” using a conventional p<0.05 threshold without correcting for multiple comparisons—raising the risk that some “positive” findings could be chance. More importantly, the documentary allegedly cherry-picks which results to emphasize and omits context that would temper interpretation.

Several examples are offered. Vitamin B intake was lower in vegans, yet blood vitamin B levels weren’t statistically different at eight weeks; the documentary reportedly doesn’t foreground this practical detail, even though long-term vegans are typically advised to supplement vitamin B. For inflammation (using a marker labeled TMAO), the teacher says outliers were removed to create a clearer group difference, but the documentary presents the contrast without that methodological caveat. The most striking presentation concern involves body composition: the teacher claims the documentary shows dexa-scan results only for the eight participants featured in the program rather than all twins, and it selectively displays individual twin outcomes rather than full charts.

Beyond reporting, the teacher argues the biggest causal problem is that vegans ate fewer calories. Even if meals were delivered with matched calories, the vegan group reportedly consumed less—about 150–200 fewer kcal per day across phases—leading to greater weight loss. That matters because downstream signals, including “biological age” measures based on epigenetic clocks, may track weight change rather than diet composition. The biological-age analysis is described as a pre-print with limited system-level significance (only 5 of 11 systems), and the authors reportedly caution that epigenetic differences could be driven largely by weight loss and caloric restriction.

The teacher also points to limitations that reduce generalizability: a small sample size (44 participants), a female-heavy cohort (34 women), and concerns about satisfaction and adherence (vegans reported lower satisfaction and were less likely to commit to fully following recommendations afterward). The takeaway is less “veganism is false” and more “the benefits likely come from eating more plants and losing weight,” with the documentary’s one-sided framing overstating what can be concluded from the data.