Data quality, concepts and methodology: Methodology
Archived Content
Information identified as archived is provided for reference, research or recordkeeping purposes. It is not subject to the Government of Canada Web Standards and has not been altered or updated since it was archived. Please "contact us" to request a format other than those available.
Notes
1.
Geometric means are used in tables 58-65, instead of arithmetic
means, which are used in tables 1-57. A geometric mean differs from an arithmetic mean as
follows:
• To calculate an arithmetic mean, all observations are added together, and the sum is divided by the number of observations.
• To calculate a geometric mean, all observations are multiplied together, and the nth root of the product is taken, where n is the number of observations.
The substances measured in tables 58-65 are environmental contaminants, which are normally found in very small quantities in the human body. This produces distributions with a high number of very small values, and a small number of comparatively large values. Because the distributions are skewed in this way, taking a geometric mean of such a distribution produces an estimate which is much closer to the true center of the distribution than would an arithmetic mean, which would be more influenced by the small number of very large values in the distribution.
• To calculate an arithmetic mean, all observations are added together, and the sum is divided by the number of observations.
• To calculate a geometric mean, all observations are multiplied together, and the nth root of the product is taken, where n is the number of observations.
The substances measured in tables 58-65 are environmental contaminants, which are normally found in very small quantities in the human body. This produces distributions with a high number of very small values, and a small number of comparatively large values. Because the distributions are skewed in this way, taking a geometric mean of such a distribution produces an estimate which is much closer to the true center of the distribution than would an arithmetic mean, which would be more influenced by the small number of very large values in the distribution.
2.
Certain tables were produced using different
statistical programming methods. Tables 1-57 were produced using BOOTVAR, a program developed by
Statistics Canada for use within the statistical software package SAS to estimate
variance for surveys with a complex survey design, such as CHMS. However,
since geometric means cannot be calculated within BOOTVAR, tables 58-65 were produced using
the software package SUDAAN.
Because SUDAAN and BOOTVAR use different methods in calculating variance, using SUDAAN to replicate tables produced using BOOTVAR (and vice versa) may yield slightly different variance estimations.
Because SUDAAN and BOOTVAR use different methods in calculating variance, using SUDAAN to replicate tables produced using BOOTVAR (and vice versa) may yield slightly different variance estimations.
- Date modified: