Interviewers began the section of the survey on alcohol by stating, "These next questions are about drinking alcoholic beverages. Alcohol includes beer, wine, wine coolers, liquors such as vodka, whiskey or rum, mixed drinks or cocktails with alcohol, and any other type of alcoholic drink." After asking whether the person has had at least one drink of any kind of alcohol in their entire life, they then asked, "During the past 12 months, how many days per week, per month or per year did you drink any type of alcoholic beverage?" Respondents could frame an answer using various time units (e.g., "5 times a week," "once a month," "10 times a year"), and the interviewer recorded the number stated and the time period that fit the answer (i.e., "Never/None," "Week," "Month," or "Year"). For the original NHIS public use files, the National Center for Health Statistics recoded these data in terms of days per week (for ALCDAYSWK), in terms of days per month (for ALCDAYSMO), and in terms of days per year (for ALCDAYSYR).

The effect of reporting in time periods shorter than a year (i.e., times per week or per month) is evident in digit clustering in the frequencies for ALCDAYSYR. There is strong clustering of responses in codes that are multiples of 12, corresponding to responses given in terms of times per month. Users should, therefore, be wary of treating ALCDAYSYR as a simple interval variable.

The Field Representative's Manual for 1997-2018 directed interviewers to "always probe for an exact number. If the sample adult reports a range or interval, assist the sample adult in making an estimate by probing. For example, you might ask, 'Could you give me a more exact number?'" To the same end, in 2000-2018, the survey instrument instructed interviewers, "If necessary, prompt with 'How many days per week, per month, or per year did you drink?'"