Skip to content
The Balance of Health
Evidence

Calculator assumptions and factors

These assumptions come directly from the calculator implementation. They describe model choices, not independently reviewed scientific findings. The calculator uses global factors; its estimates are illustrative.

Download assumptions and factors as JSON · Licence · Full methodology

55 factors are in use. Each states the source it came from and how it was arrived at: sourced straight from that publication, derived by our own arithmetic on it, or a proxy standing in for a food no source covers. They are calculator inputs, not reviewed findings; CC BY 4.0 covers our compilation of them.

Servings

A serving is the typical portion defined for each food in our comparison tool: 150 g of beef, lamb, pork or chicken; 140 g of fish; 250 ml of milk; 30 g of cheese; two eggs (100 g). Your real portions may be bigger or smaller.

Fish

All fish and seafood use farmed-fish environmental factors and salmon nutrition and yield as proxies. Wild fish, prawns and tinned tuna have different footprints, some lower and some higher.

Global factors

The calculator uses global average food footprints. Your location, preference cookie and any country parameter on an old shared link do not change its factors. Adjust the starting servings to describe your own diet.

Production systems

The calculator uses the global all-systems average for each food. Dairy-herd beef can be explored in the comparison tool but is not selectable here.

Replacement food

Removed animal servings are assumed to be replaced by an even mix of lentils, tofu and chickpeas supplying the same protein. This is a simplification, not a nutritionally complete diet plan. Digestibility is not adjusted. Pulse environmental factors use retail dry weight while protein uses cooked food; without cooking-yield conversion, replacement footprints are approximate.

Uncertainty

Where a source reports a range for a factor, that range is used. Where it does not, we apply ±40% (0.6× to 1.4×), as an illustrative sensitivity band informed by the variation in Poore & Nemecek (2018) and Clune et al. (2017). It is not a confidence interval or a full uncertainty model. Ranges are added, not combined statistically.

Savings range

The saving is the difference between the middle estimates, with the baseline's relative uncertainty applied to it. This avoids the artefact of a large baseline range implying a change could increase impact.

Animals

'Animals avoided' counts animals slaughtered per kilogram of meat or fish. Dairy and eggs count as zero because no animal is slaughtered per unit sold, although both systems involve animals. The figure is a statistical share, not a promise that a particular animal is spared.

Water

Plain freshwater use is shown but is a poor guide to harm. Scarcity-weighted water use, where available, weights each litre by how stressed the source region is. The global factors blend producing regions; consumer location is not a multiplier on them.

What is not counted

Cooking, food waste, packaging differences between products, transport from shop to home, and health effects are not included. Dietary change also has effects on land that are not captured by area alone, such as whether freed land is restored.

Precision

All outputs are rounded and prefixed 'about'. They describe an average person eating average products; your actual impact could differ by a factor of two or more.

Global food factors

All values are per kg, retaining the food record’s retail-weight and cooking assumptions.