All laboratory measurements could not be conducted for all survey respondents, therefore, a randomly selected subsample was selected for particular environmental components. The subsamples selected for these components are chosen at random with a specified sampling fraction (for example, 1/3 of the total examined group) according to the protocol for that component. Each subsample involves another stage of selection and separate sample weights that account for the additional stage of selection and additional non-response. For analysis of subsample data, appropriate subsample weights must be used, and they are included on any data file where relevant.
Users are encouraged to combine subsample data for a given environmental chemical across survey cycles to improve the statistical reliability of the estimated measured concentrations of environmental chemicals.
It is possible to conduct analyses of several environmental chemicals in the same group of participants if they were measured in the same randomly selected subsample (e.g., urinary mercury NHANES 2007-2008 and urinary barium from NHANES 2007-2008). It is NOT possible to analyze environmental chemicals in the same group of participants if they were measured in different randomly selected subsamples (e.g., urinary Mono(carboxyoctyl) phthalate NHANES 2005-2006 and urinary barium from NHANES 2005-2006).
Analyses combining data obtained with different subsamples (either within or between survey cycles) will be problematic. An example of this is chemicals within the 1/3 subsample for phthalates and the 1/2 subsample for volatile organic compounds. While some survey participants may have measured values for both chemicals of interest, most will not have measured values for both chemicals. In addition, the sample weights available for those with both values cannot be directly combined in an analysis. If an attempt is made to combine two or more subsamples for your analyses, then appropriate weights would need to be recalculated. However, details on how to recalculate weights when combining subsamples go beyond the scope of this tutorial. Therefore, it is strongly advised that you do not attempt to combine subsamples in any analysis.