Key Concepts About Deciding Which NHANES III Variables to Include

Your analysis will require a subset of the variables available in NHANES III. To decide which variables are needed in your analysis, you need to review the survey documentation. The survey documentation for each component is slightly different. Survey documentation for:

These files, as well as the survey documentation for other components, are available on the NHANES III Data Files - Documentation/Codebook and SAS Code page for NHANES III Series 11. Within the data documentation files there are several main sections (refer to the documentation file Table of Contents).  Each documentation file contains the same introductory sections on survey description, guidelines, and data preparation and procedures.  The documentations files also contain a data file index, and data file item description, codes, counts and notes that are specific to that data file (e.g. Laboratory data). 

The survey documentation files are Adobe Acrobat PDF files. Use the search feature to speed up your search for variables relevant to your analysis. Read the documentation for each "hit" in your search results carefully though as not every result returned will be relevant to your analysis.

For example, assume you are preparing for an analysis using blood pressure variables, and search the examination survey documentation file for "blood pressure." Some blood pressure questions are used to assess dietary patterns.  There are multiple measures of blood pressure as well as an averaged measure. These are in the main collection of standardized blood pressure measurements conducted on all eligible participants aged 5 years and older in the Physician's Exam section of the survey. You will need to decide which blood pressure examination variables you would want to use in your analysis.



Blood pressure measurements were taken from survey participants during both the home interview and the MEC exam. However, for participants aged 2-11 months and those aged 20 years and over who could not travel to the MEC, they were administered a home examination at a later date. The MEC examination and home examination measures were both taken by a trained medical professional. The " average blood pressure" variables provided in the dataset are averaged over all six measurements from both MEC and home exams. 

In general, most analysts will use the three measures taken in the MEC, and average those measures themselves. However, if the focus of your analysis is the elderly population, you might want to include the home exam participants, because the majority of home examination blood pressure measurements are among elderly participants. If these several hundred home examined participants are included in your analysis, you must be certain to use the survey weight associated with the home examination to get accurate estimates.


In another example, assume you are also including cholesterol variables in your analysis and you search the laboratory survey documentation for "cholesterol." Again, you will have many results, but only some of them may be applicable to your analysis. Notice that some of the results are in the General Biochemistry Tests and others are in the Biochemistry Profile. In general, most people will want to use the results from the General Biochemistry Tests because it uses the reference method for cholesterol, rather than a standard test battery for a panel of biochemical variables.  You should review both portions of the documentation so you know which variables are appropriate for your analysis.

You must read the documentation to know this information and identify the correct variables for your analysis.



Comparison between NHANES III data files and continuous NHANES data files

The data files for NHANES III are structured quite differently than the data files for continuous NHANES. The continuous NHANES were first organized by survey cycle years, then by separate components for each topic area, plus separate links for the documentation, codebook, and frequency tables for each of the topic specific data components.

NHANES III is structured around the survey mode of administration (i.e. data collection instrument). The files are related to the interview questionnaire (adult, and youth), the MEC examination, the laboratory results, and the Foods and dietary recall section. Therefore, there are several key differences to keep in mind when you search and locate variables from NHANES III data:

  1. There are no separate links for variable list in NHANES III as in continuous NHANES.  You should check the Table of Contents and the Data File Location table to determine which of the files (there may be more than one) are likely to contain your variables of interest. For example, there are blood pressure measurements in both the home interview for the subset of home-examined subsample persons, and the MEC exam files. Then you need to check the Data File Index which is a list of all the variables in that file in sequential order, plus a descriptive label next to each one.
  2. There are no separate links for codebook.  The " codebook" is embedded within the data file documentation and immediately follows the Data File Index. It is not called a codebook. It is simply called by its instrument title (e.g. Household Adult Questionnaire).
  3. There is also no separate link for frequencies. There are simple counts provided within the " codebook" (see above) next to the listing of each possible response to each question or measurement.
  4. It is important to read the notes. This section immediately follows the " codebook" . Variables which require further explanation or clarification will have the word or words see notes written to their far right in the " codebook" section. It is important to follow through and read the pertinent text note in the notes section. For example, these notes will tell you that a fourth blood pressure measurement was taken only when the health examination professional could not successfully acquire valid readings for the first three measures.


close window icon Close Window