We need both studies of both theoretical and applied GHG emissions from PB and omnivores

• modest meat reduction can have significant environmental benefits
• Going vegan is good for the environment
• A similar methodological issue is found here: Self report studies of food spend differences between veggies and omnis are probably useless

Is a plant based (vegetarian or vegan) diet better for the environment?

If you look at the academic literature on this topic, you'll find that there are two ways to answer this question:

1. We look at the real diets of vegetarians and vegans (henceforth we'll group them as "veg*ans"), and measure the greenhouse gas emissions of their diet. Then we do the same for omnis and see which has lower emission on average.

For example, Scarborough et al., 2023 find that across 55,000 people in 119 countries that the more animal products a person's diet contained, the higher on average their environmental impact was when it came to GHG emissions, river pollution, water and land usage. Another example is Conrad et al., 2023 found that in a sample of 8000 americans, plant based diets produced lower greenhouse gas emissions than any other diet: low carb, low fat, low grain or diets involving fasting.

1. We simulate an example veg*an diet (or several) and compare their hypothetical greenhouse gas emissions with the emissions from the general population of the omnivore diet.

Are both of these types of studies valid? Is one better than the other?

The answer is that both have important limitations, but by combining them, we can be more sure of the truth.

What's wrong with simulations?

We'll start with the obvious one: the limitation of simulations. Because the diets are created artificially by the researchers, they can be manipulated and cherry-picked. Researchers who want to push "the plant based agenda" might intentionally create a veg*an diet for their simulation that is really low in environmental impact. They could leave out pesky plant based foods that can be bad for the environment, like water guzzling almonds and forest-clearing palm oil.

What's good about simulations

But they do have advantages. The fact that researchers can design diets can be a strength as well as a limitation. It allows simulation studies to tell us about a variety of possible future scenarios and what would happen. A study might simulate the GHG emissions of a veg*an diet that also sourced most of it's food locally, or they might simulate an affordable veg*an diet, or any number of scenarios. Studies of real veg*an diets only tell us what would happen if the whole population started eating like today's veg*ans (more on this below), but if we're building a future of food, why settle for what we're doing today? Simulation studies allow us to figure out what the best diet of the future is, without being constrained by what we're currently doing.

What's wrong with using real diet data?

The limitations of "real diet" studies is a bit more subtle, but still quite a bit problem. In a nutshell: Using real diets may not actually answer the question that we want to answer!

We're not compare vegan diets and Omni diets just to see who is better for gloating reasons. We compare the diets because we want to make the argument that if an omnivore goes vegetarian or vegan they will reduce their greenhouse gas emissions.

But one issue here is that an omnivore going vegetarian or vegan does not necessarily mean they will eat the same diet as the current plant based eaters. Modern day veggies and vegans are mostly young, female, liberal, health conscious and all sorts of other peculiarities. They're not representative of the population. If all the old conservative men in the population went vegan, would we expect them to adopt the same diet as current vegans? This seems highly unlikely.

And if they don't adopt the same diet as current veg*ans, then why would their environmental impact be the same?

One example is that current vegans anecdotally don't tend to consume large amounts of meat alternatives like vegan burgers, chick'n nuggets and sausages. But it's reasonable to suspect that if we converted a large amount of the population to veganism, there would be lots of people who'd would want to eat these more or less daily. Why do I think this? Because there are many omnivores whose diet mostly consists of nuggets and burgers. They're probably not going to magically start eating broccoli or fruit as much as the average vegan. But this is what we're assuming when we used measured diets in research.

Another example: Clara mostly eats chicken nuggets, fish fingers, basically any breaded meat and chips. Clara decides that she'd prefer her beige diet to be a bit more green, so goes vegan. Is she going to start eating plenty of beans and vegetables like the average vegan? Probably not. She's more likely to just continue eating the same diet, but now with vegan chick'n nuggets, NotFish fingers etc. Heck, she may not even have to change brand: Birds Eye make great vegan fish fingers. Her diet would be very different from the average vegan's diet, and so might her GHG emissions.

How would we tackle this?

So simulated diet studies can be biased, but also so can measured diet studies. What do we do?

Like good scientists we should use the principle of converging evidence. Luckily for us, both simulation and measured diet studies suggest reducing animal product consumption will be better for the environment.