Some studies have shown teetotalers to be less intelligent than moderate drinkers are. Data from the GSS agrees with this. Those who sit on the barstools every night, and those who would never place their derrieres there, are less intelligent than those who go out every now and then.
Turning alcohol consumption into a dichotomous question, teetotalers are, on average, of more modest intelligence than drinkers are. Converting Wordsum results based on the white mean to IQ scores with an assumed standard deviation of 15 points yields an average IQ of 92.8 for non-drinkers and 99.6 for drinkers (N = 7,204). Fewer than three in ten respondents never drink. This total abstinence is partly religious in nature, as 77% of the non-drinkers are certain of God's existence, compared to 57% among drinkers. Religiosity and intelligence trend in opposite directions. Without a specific prohibition on it, even those who do not like to drink and rarely do are still going to show up in the "yes" column. If the frequency and quantity of consumption were inquired about, I suspect that moderate drinkers would be more intelligent than heavy drinkers.
But that assumption could be mistaken. In the United Health Foundation's annual report of health rankings by state, one of the factors tracked is the percentage of a population that binge drinks. The UHF uses the standard 5/4 definition of the behavior (five drinks on one occasion for a man, four for a woman) in arriving at its figures. They correlate inversely with estimated average IQ at .48 (p=0). Heavy drinking is more popular in smarter states than it is in duller states*.
The UHF does not provide a nationwide visualization of the percentage of each state's population that engages in binge drinking, so I've created one here.
As smarter states have more binge drinkers than duller states do, it is not surprising that the behavior is more common in irreligious states than it is in religious states. As part of its "state of the states" project, Gallup polled residents and inquired whether or not religion comprised an important part of their daily lives. The percentage of respondents who said it does and the percentage that binge drink inversely correlate at .54 (p=0).
The Southern Baptist Convention, concentrated in the nation's generally unhealthy states, takes a strong stance against the distribution and consumption of alcohol. As the visualization shows, the pious South does well in limiting binge drinking, and Utah stands out conspicuously in the West as a place of moderation. This looks to be a clear example of religiosity having beneficial consequences for the individual and the larger community, specifically for those on the left side of the bell curve.
Alternatively, many northern states--where the winters are longer, the sun shines less, and the population is more heavily northern European--are more prone to heavy bouts of drinking. As northern Europeans have had a shorter history with alcohol, they are presumably at higher risk for falling into alcoholism than other European groups are. Ancestry could conceivably offer a better explanation than religiosity does.
Perhaps I've finally found a pathological behavior that correlates inversely with intelligence. Anecdotally, I know several sharp teenagers who are regular heavy drinkers (and also into various forms of recreational drug use). To a tee, they are all leftists. I get a sense that it is a form of destructive entertainment for those who have high levels of openness to experience but who do not find constructive things to engage this tendency in because there are not external forces--peer, parental, transcendent, or otherwise--pushing them to do so.
GSS variables used: DRINK, WORDSUM, GOD
* It is also more popular in blue states. The correlation between voting for Obama and binge drinking is .47, even though "blue state" and "smart state" are not synonymous. Estimated average IQ and the percentage of a state's electorate who voted for Obama correlate at a weak and statistically insignificant .16 (p=.27)--if DC is included, the correlation becomes slightly inverted but still statistically meaningless.