Small Portion Error

Small portion error is the disproportionate elevation in percentage estimation error for foods consumed in small quantities. The effect is partly arithmetic — a fixed absolute error on a small reference value is a larger percentage — and partly methodological, because measurement tools are generally specified for typical meal-scale quantities and lose relative accuracy at the low end.

The arithmetic

The MAPE formula divides absolute error by the absolute reference value. A 5-gram weighing error on a 200-gram chicken breast is a 2.5 per cent error; the same 5-gram error on a 25-gram pat of butter is 20 per cent. The scale's specification in grams does not change between measurements, but the methodological implication does. For accurate logging of small items, a higher-precision scale (0.1 g) becomes necessary; for large items, a 1 g scale suffices.

The methodological compound

Three effects compound the arithmetic problem:

• Scale resolution. A 2 g kitchen scale reporting "15 g" for an actual 14.5 g portion rounds; for small portions, the rounding band (±1 g) is a large fraction of the reading.
• Tare accuracy. Small portions are often weighed on or in a container. The container's own weight, taring out, has its own precision. If the taring is off by 1 g, a 20 g portion carries a 5 per cent error immediately.
• User behaviour. Small items — dressings, condiments, oils drizzled over finished dishes, garnishes, sugar in coffee — are often not logged at all. A 2016 observational study of MyFitnessPal users found that condiments and dressings were logged in roughly 40 per cent of restaurant meals where dietitian observation confirmed their presence. Non-logging is a 100 per cent error for that component.

Magnitude across logging modes

• Kitchen scale, 0.1 g precision: small-portion MAPE 0.5 to 2 per cent (instrument-limited).
• Kitchen scale, 1 g precision: 5 to 15 per cent for items under 20 g.
• Measuring spoons without a scale: 10 to 30 per cent, with systematic underestimate bias for viscous items (peanut butter, honey).
• Visual estimation: 30 to 100 per cent for small portions, approaching total miss for items under 10 g.
• Photo-based methods: variable by method; small items in images are often below the model's resolution threshold entirely.

The "forgotten condiment" problem

Over a day of moderately-attentive tracking, the accumulated small-portion error is not catastrophic in calorie terms — a typical diet carries 100 to 300 kcal of condiments, dressings, and small additions, and missing half of them produces a 50-to-150 kcal daily underestimate. But the proportional effect on specific nutrients is larger: sodium and added sugars are disproportionately concentrated in small items (a tablespoon of soy sauce is 12 per cent of the FDA sodium DV), so a tracker that under-logs small items will systematically under-report sodium intake while tracking calories reasonably well.