Overview of NHANES Survey Design and Weights
NHANES uses a complex sampling design and constructs sample weights to produce nationally representative data. Learning about the features of the NHANES survey design and weighting will help ensure that the results of your analyses represent unbiased estimates with accurate statistical significance levels. Much of the material presented in this module can also be found in the Continuous NHANES tutorial.
Task 1: Explain NHANES Survey Design
NHANES data are not obtained using a simple random sample. Rather, a complex, multistage, probability sampling design is used to select participant’s representative of the civilian, non-institutionalized U.S. population. Oversampling of certain population subgroups is done to increase the reliability and precision of health status indicator estimates for these groups. This task will describe the general statistical sampling used to identify NHANES participants.
For NHANES datasets, the use of sampling weights and sample design variables is recommended for all analyses because the sample design is a clustered design and incorporates differential probabilities of selection. If you fail to account for the sampling parameters, you may obtain biased estimates and overstate significance levels.
Task 2: Explain NHANES Sample Weights
NHANES has constructed various sample weights for each two-year survey cycle to take into account survey non-response, over-sampling, post-stratification, and sampling error. This task describes how sample weights are constructed in NHANES. Due to the way NHANES participants are selected, sample weights must be used to produce an unbiased national estimate. This task will also provide advice on the use of the sample weights when analyzing the data.
- Key Concepts about the NHANES Sample Weights
- When and How to Construct Weights When Combining Survey Cycles
Task 3: Explain NHANES Environmental Chemical Data Subsample Weights
Some NHANES environmental analytes such as blood lead or blood mercury are obtained on a full sample of participants, therefore full sample MEC weights can be used for analysis. However, most environmental analytes are measured in 1/3 subsamples. For analysis of this subsample data, appropriate subsample weights must be used. They are included on any data file where relevant. This task will describe the special features of most environmental chemical sample weight.