In 2008, a group of British epidemiologists went looking for evidence that clever children grow up to drink sensibly. They found the reverse. The authors said so in print. The result had run counter to what they expected.
That paper was Batty, Deary, Schoon, Emslie, Hunt and Gale, in the American Journal of Public Health, using the 1970 British Cohort Study. Mental ability measured at age ten was associated with higher alcohol intake at age thirty, more frequent drinking, and a greater prevalence of problem drinking. The association survived adjustment for markers of socioeconomic position across the life course. It was stronger in women than in men.
What follows is a reading of the research, not medical advice.
What the cohort studies actually measured
Childhood test scores are not intelligence in any complete sense. They are performance on a particular instrument, on a particular morning, at a particular age. Adult drinking, in these datasets, is self-reported, measured as units per week, frequency, and questionnaire responses about consequences. Self-report tends to undercount.
The British result is not isolated. Satoshi Kanazawa and Josephine Hellberg, writing in Review of General Psychology in 2010, examined the 1958 UK cohort and a large American longitudinal sample and reported that more intelligent children were more likely to grow up consuming more alcohol. Their own framing was cautious: the data, they wrote, partly supported their prediction. The evolutionary explanation they attached to it is a hypothesis they proposed, not a mechanism anyone has demonstrated.
The direction reverses in other countries
Sara Sjölund, Tomas Hemmingsson and Peter Allebeck examined 49,321 Swedish men born between 1949 and 1951, tested at military conscription. Their 2015 paper in Alcoholism: Clinical and Experimental Research found the opposite pattern: lower test results were associated with higher total intake and more binge drinking. Earlier work by the same group, following the conscripts for 37 years, linked lower adolescent scores to alcohol-related hospital admissions and deaths.
So the correlation is real in some cohorts and inverted in others, depending on the country, the birth year, and above all which outcome is being counted.
None of this makes drinking a marker of anything good
This is not evidence that alcohol is protective, and nothing in these papers turns drinking into a sign of ability. A population-level correlation of this size says nothing usable about any individual person, and the established health risks of alcohol sit entirely outside the question these studies were asking. An average across a birth cohort is not a fact about you or anyone you know.
How often and how much are separate measurements
The genetic evidence pulls these apart, and this is where the picture gets more useful. A two-sample Mendelian randomisation study drawing on genome-wide data from roughly 780,000 people reported that genetic instruments for more education were associated with a substantially reduced risk of alcohol dependence, on the order of half, per 3.61 additional years of schooling. A separate analysis of 334,507 UK Biobank participants, published in Molecular Psychiatry, found that genetically instrumented education predicted higher total intake, with the strongest signal for red wine.
More often, in smaller amounts, with food, spread across the week. Less dependence. That is a different behaviour from the one most readers picture when they hear that clever people drink more. These are also models built on genetic proxies for education, which is not the same variable as a childhood test score.
The plainer explanation
Alcohol in Britain and much of northern Europe has been bound up with professional social life for a long time. The conference dinner, the client lunch, the departmental drinks, the bottle brought to a friend’s house. People who do well on tests at ten tend to end up in occupations and social settings where drinking is scheduled into the calendar, with the disposable income to keep a wine rack stocked.
None of that requires a theory about the mind. It requires a theory about who gets invited to what.
The limits are worth stating. These cohorts are overwhelmingly white, Western, and born in the middle of the last century, and drinking cultures have shifted considerably since the 1970 cohort turned thirty. Adjusting for income and class reduces confounding without eliminating it.
If any of this reading prompts a question about your own drinking, a GP is a better place to take it than a set of cohort studies, which were never designed to answer anything about a single person.
What the literature supports is narrower than the headline version: in several British and American cohorts, higher childhood test scores predicted more frequent adult drinking, while the Swedish data points the other way on heavy use and harm. Which drinking culture a person ends up inside appears to matter more than how they scored at ten.
