Fig. 4.

Simulation of the “true” usual intake distribution. Legend: The “simulated BNFCS2004” consisting of two 24-h recalls per individual was simulated separately in the three age strata: the adolescents (15–18 years), the adults (19–59 years) and the elderly (60–74 years). The grey-shaded steps are performed at the transformed scale. In box A, the needed modeling parameters were extracted by modeling the original BNFCS2004. In box B, these parameters were used to simulate one thousand 24 HRs for every consumer. The median intake over these one thousand simulated days results in the “true usual intake distribution for consumers only”. In box C, the correct proportion of never-consumers gets assigned one thousand 24 HRs with an intake amount equal to zero, which corresponds to a usual intake of zero. Finally, the usual intake distributions of the consumers and the never-consumers are combined to obtain the “simulated true usual intake distribution”