Sampling Procedure
Data were collected in a nationally representative cross-sectional survey (the 2008–2009 Brazilian POF), which investigated a sample of 55 970 households that had been selected using a two-stage cluster sample design. In the first stage, census tracts, which were the primary sampling units, were randomly selected based on data from the 2000 Brazilian Demographic Census to obtain homogeneous socio-economic and geographic strata. In the second stage, households were selected within each tract by simple random sampling without replacement. The National Dietary Survey was conducted in about 24 percent of these households (n = 13 569) to obtain food consumption data for all family members, individuals aged 10 years and older.
Weighting
Each household in the POF sub-sample represents a certain number of permanent private households from the population (universe) from which the original sample was selected. As such, each household in the sub-sample is associated with a sample weight or expansion factor. When applied to the characteristics investigated by the POF, this factor allows for the estimation of quantities of interest for the entire survey population.
The expansion factors were initially calculated based on the sampling plan effectively used in the selection of the sub-sample, incorporating adjustments to compensate for non-response from the investigated units. Subsequently, these factors underwent calibration adjustments, a procedure that aimed to ensure that, for each Federation Unit (calibration domains), the estimated totals for the population in certain segments matched the respective totals obtained through the expansion of the original POF sample.
Three variables represent the expansion sample weights in the survey: ESTRATO_POF, COD_UPA, and WEIGHTING_FACTOR.
WEIGHTING_FACTOR: Final sampling weights (original survey variable "PESO_FINAL").
ESTRATO_POF: Stratification variable of the survey sampling design. This variable identifies the sampling strata used in the complex survey design.
COD_UPA: Primary Sampling Unit (PSU) identifier of the survey sampling design. This variable identifies the selected clusters (UPAs) and should be used in the analyses that account for the complex survey design.
WEIGHTING_FACTOR can be used alone to obtain weighted point estimates, but ESTRATO_POF and COD_UPA should also be used in analysis that account for the complex survey design, to allow for an accurate estimation of standard errors, confidence intervals, and statistical tests
