No one can weigh a whole population's dinner. Instead we ask — and every way of asking bends the truth. The two bends that matter are random error (scatter that averages out) and systematic error (a lopsided pull that does not). Pick a method, add days of data, and watch where its estimate lands next to a person's true usual intake.
The estimate vs. the truth
−900true usual intake+900 kcal
Estimate
—
kcal/day
Random ±
—
kcal (~95%)
Bias
—
kcal shift
Collect more data
1 day
More independent days shrink the random scatter (it falls with the square root of the number of days) — but the systematic bias arrow never moves. More data cannot fix a lopsided instrument.
Three ways to measure a diettap one below to begin