If you look at the percentage breakdown it’s much more modest, so it appears to be an indication of how much the participants were already spending on alcohol. Also if you filter by “high income” beverage spending appears to switch to marijuana, and there the % breakdown remains high (no pun intended)