← Toutes les offres
J

Was I too optimistic about plant-based meat?

Jacob_Peacock

Global
Publié le 18 novembre 2026

Description du poste

Cross-posted from my new Substack, Alternative Perspective. Views my own. Part I of Do Beans Beat Alt Meat? People know me as a plant-based meat pessimist, mainly for my review “Price-, taste-, and convenience-competitive plant-based meat analogues would not currently replace the majority of meat consumption” (published & free version). I argued that even if products like the Beyond or Impossible burgers were as cheap, tasty and convenient as beef, most people would still choose beef. However, I ended the paper by claiming “This result does not imply that plant-based meat analogues have no role in environmental, public health or animal welfare policy.” Among other reasons for optimism, I thought that adding a competitive plant-based meat option to a menu might reduce meat consumption, if only by 5 or 10 percentage points. Alas, my own research would show me wrong and wrong again. So in the spirit of caring critique, I will begin with some self-care. You can see my optimism in predictions I made about the results of a 2023 study. Sagar Shah and I were about to reanalyze 35,048 hypothetical lunch choices by 1,348 university students in Germany. Each student had chosen a meal from a series of lunch menus with five items, two or three of which were meat-free. Some of those meat-free items were plant-based meats, although not the more modern options like Beyond or Impossible. The menus showed photos and descriptions of real university canteen dishes, but no prices. Before peeking at the data, we wrote out our premonitions. I guessed it was 65% likely that menus with a plant-based meat option resulted in more meat-free choices, and 60% likely that the association would be statistically significant. After conducting our analysis, we found the odds of picking a meat-free option were 1.18 times higher when a plant-based meat was available. So I was “right” on the first one. However, the 90% confidence interval, which accounts for the uncertainty of only asking so many p