Codes and Frequencies
HEIGHT reports the approximate height of adults, in inches. The values for this variable are based on self-reports or proxy reports/estimates by respondents for other household members, supplied in response to the question, "About how tall is [person] without shoes?" Such data are less accurate than, and thus not fully comparable to, data on height gleaned via direct measurement in other surveys, such as NHANES (the National Health and Nutrition Examination Survey fielded by the National Center for Health Statistics).
Along with changes in the variable universe, changes in top- and bottom-codes limit the comparability of this variable over time.
For 1976-1981, the recognized range was 36 inches to 84 inches; for 1982-1995, 36 inches to 98 inches; for 1996, 58 to 77 inches. In 1997 forward, the recognized range for height differed between men and women. For men, it ranged from 63 to 76 inches; for women, 59 to 70 inches.
The question for HEIGHT in 1977 was asked for a sub-sample of randomly selected persons age 20 and over; NHIS automatically classifies persons who were selected but did not complete the questions as responding "unknown-all causes" for this variable. Users may choose to use this variable in combination with SUBSRESP77, which reports the person's sub-sample status in 1977.
For years prior to 1997, persons who were taller than the top-coded value or shorter than the bottom-coded value were included in the highest and lowest categories, respectively. Thus, for example, in 1996, category 77 includes people who were 77 or more inches tall. By contrast, for 1997 forward, persons who were unusually tall or unusually short were grouped together in a separate "Exceptionally tall or short" category (code 96). As the Survey Description Document for 1997 explained, "Beginning in 1997, when a very large or very small value was reported for either height or weight, the data in both variables were changed to '96' or '996' ('Not available') on the public use data files. This was done to protect the confidentiality of NHIS respondents who might be identifiable by their unusual physical characteristics."
Because of changes in the recognized ranges and in the treatment of outlying values, analysts should exercise caution in comparing results across periods.
For example, average height calculated using 1981 data would necessarily be shorter than average height calculated using 1982 data, because the topcodes for the two years differ (84 inches versus 98 inches). For comparisons across periods, the best approach would be to limit the sample to persons whose height falls within the shortest recognized range for the period under study and exclude persons assigned the top or bottom codes.
To achieve comparability across all years, researchers should restrict their analysis to persons whose height fell within the identified range for 1997 forward, namely, from 59 to 76 inches.
- 1974: Sample persons age 17+.
- 1976: Persons age 18+.
- 1977: Sub-sampled persons age 20+.
- 1978-1981: Persons age 17+.
- 1982-1996: Persons age 18+.
- 1997-2017: Sample adults age 18+.
- 1974, 1976-2017