Documentation for the survey, including the exact wording of the questions: https://www.cdc.gov/brfss/index.html https://www.cdc.gov/brfss/index.html For the

Documentation for the survey, including the exact wording of the questions: https://www.cdc.gov/brfss/index.html https://www.cdc.gov/brfss/index.html For the alcohol consumption questions, they ask about the times people drank in the previous 30 days. >You go to a 4th of July party and now you're an excessive drinker for the whole year? This survey isn't used to morally judge populations, and the (very commonly used) summary statistics for heavy drinking don't mean every single respondents meeting those criteria is a heavy drinker. On a personal level, it depends on body mass, overall health, yadda yadda. But it's a good measure, because for the average man, 5 drinks in one day is definitely unhealthy. And, for everyone who gets swept up in that aggregate because St. Patrick's Day or the 4th of July was recent, there are people doing dry January or called in September. It averages out. And, even if you argue all that, the data is still useful because it's relative to everyone else. Different definitions should yield roughly the same proportions of heavy drinkers between counties. >cultural aspects don't vary that much just because of an imaginary administrative line Those lines aren't arbitrary. They were usually drawn along physical barriers. Those barriers also sorted people into towns and heavily influenced how they intermingled in the past. I work with county data, and stark differences between neighbors is very common across all data types. For instance, Philadelphia is abysmal for almost all health measures. But Chester county, just west of it, does amazingly. Why? Because Philadelphians have more pollution, lower average income, and a host of other problems. Chester is home to a lot of rich people who commute to Philadelphia.