Finally, the weights were calibrated separately in each province or territory using the benchmark variables given in Table C4. Attempts to include household size and education variables proved unsatisfactory and were abandoned. Variables that were used had been validated through matches of the collected survey data with available frame information. Small amounts of missing data for the calibration variables were imputed. Census counts for all calibration variables at the enumeration area level were inflated according to the growth measured between provincial age and gender totals from the Census and the corresponding official demographic counts as of June 21, 2003. This reference date represented an approximation of the midpoint of collection both in terms of calendar days, and in terms of completed response.

Table C4