Note that that data is unreliable. IIRC, the methodology is to start with a survey, multiply by the population size, find that it doesn't account for nearly hal

Note that that data is unreliable. IIRC, the methodology is to start with a survey, multiply by the population size, find that it doesn't account for nearly half of the drinks sold, and then multiply the survey results by a constant factor to account for the difference. So, for example with population = 2, one reported 35 drinks per week and the other reported 1 drink per week, but we sold 72 drinks per week so we assume that the person who claimed 35 actually drank 70.