NHANES participants are asked to participate in a variety of survey components, some of which are completed with subsamples of examinees. Several laboratory assays, nutrition, environmental health assessments, and mental health interviews are completed on subamples of examinees. (Please see the respective survey protocol/documentation for more specific information. Each component subsample has its own designated sample weight, which accounts for the additional probability of selection into the subsample component, as well as the additional nonresponse.
Recall from the “Overview of NHANES Survey Design and Weights” module in the Survey Orientation course that the set of all individuals that have nonzero values for a particular version of sample weights comprise a nationally representative sample, so long as those sample weights are incorporated into statistical analyses. However, the construction of sample weights does not take into account item non-response. Item non-response is described in Module 9, Task 1: Key Concepts about Missing Data in NHANES.
The table below shows the versions of sample weights that have been constructed for individuals within different groups of dietary data files. To produce estimates appropriately adjusted for survey non-response, it is important to check all of the variables in your analysis and select the weight variable that applies to all members of the smallest analysis subpopulation. The following table lists the types of sample weights by decreasing sample size.
|Sample Size||Component||Data file(s) Containing Sample Weight Variable||Weight Variable|
|Largest||Questionnaire, supplement use||Demographic||WTINT2YR|
|▼||24-hour recall (Day 1)||Individual Foods; Total Nutrient Intakes||WTDRD1|
|▼||24-hour recall (Day 1 and 2)||Individual Foods; Total Nutrient Intakes||WTDRD2|
|Smallest||Special subsamples (e.g. fasting)||Varies||Varies, depending on the subsample|
For example, if you wish to combine data from the Day 1 24-hour recall data and the supplement use section (collected during the household interview), you must restrict your analysis to individuals with dietary recall data. To do this, you designate that persons with a non-zero dietary recall Day 1 sample weight value (variable WTDRD1 in the Individual Foods and Total Nutrient Intakes files) be included in the analysis. In doing so, you will be analyzing the supplement use data for individuals who completed the first 24-hour recall.
Other examples for selecting the correct sampling weights for your analysis are included below:
All of the variables in the analysis were collected in the in-home interview
You are performing analysis on NHANES 2003-2004 data to look at the association of race/ethnicity and supplement use. All of the variables were collected in the in-home interview.
For this analysis, you would use the interview weight (WTINT2YR).
Some of the variables in the analysis were collected in the MEC
You are performing an analysis on NHANES 2003-2004 data to look at the association of race/ethnicity, age, poverty and the prevalence of obesity. All three demographic variables were collected during the in-home interview. However, height and weight are collected during the MEC exam. MEC-examined sample persons are a subset of those interviewed in the survey.
For this analysis, you would use the MEC exam weight (WTMEC2YR).
Some of the variables in this analysis were collected in the MEC, but have special circumstances
You are performing an analysis on NHANES 2003-2004 data to estimate mean milk intake for subpopulations defined by race/ethnicity and body mass index. Race/ethnicity information is available from the in-home interview. Weight and height are available from the MEC sample. MEC-examined sample persons are a subset of those interviewed in the survey. Mean milk intake comes from the dietary recall data. Dietary recall data are collected during the MEC interview, however, these data have their own sample weight based on complete and reliable recalls.
For this analysis, you would use the dietary recall weight (WTDRD1).
Some of the variables in the analysis were collected as part of a subsample
You are performing an analysis on NHANES 2003-2004 data to look at the association of race/ethnicity, age, obesity, and fasting triglycerides on persons age 20 and over. Race/ethnicity and age are available from the in-home interview. Obesity information comes from the MEC exam. MEC-examined sample persons are a subset of those interviewed in the survey. Fasting triglycerides are collected from those sample persons who were sub-sampled to do the 9 hour AM fast and who actually fasted. This group is comprised of approximately half the sample of those who were MEC examined.
For this analysis, you would use the morning fasting subsample weight (WTSAF2YR).