Say it isn't so

Blogging on Peer-Reviewed Research

The NYT reports on a this article by Tomas Grim of the Dept of Zoology at Palacky Univ purporting to show a negative effect on numbers of scientific publications for scientists correlated with increasing beer consumption.

According to the study, published in February in Oikos, a highly respected scientific journal, the more beer a scientist drinks, the less likely the scientist is to publish a paper or to have a paper cited by another researcher, a measure of a paper's quality and importance.

The results were not, however, a matter of a few scientists having had too many brews to be able to stumble back to the lab. Publication did not simply drop off among the heaviest drinkers. Instead, scientific performance steadily declined with increasing beer consumption across the board, from scientists who primly sip at two or three beers over a year to the sort who average knocking back more than two a day.

However, looking at the paper I'm somewhat confused, and not just from the willingness to generalize to all scientists from a single country's avian ecologists. For one, the scales have to be a goof. Check out the first figure.
i-4fd1c38060d2e0cc51fa3d5f3f89e27a-oik_16551_f1.gif

Who drinks 2 liters of beer a year? That's basically teetotalling. Even 6 liters a year (the high end of his effect) would be a very small amount. Is this just alcohol in the beer? At 18ml/12oz beer that would mean each liter corresponds to ~50 beers. At 6 liters that's still only 300 beers or less than one a day. If instead the author means 2-6 liters/day/person/year that may make more sense. But 6 liters of alcohol a day? Maybe the Czech's are worthy rivals for beer drinking but that's now an unbelievably high amount for a non-hobo. How about 100-600 liters of beer a year? One liter is roughly 3 x 12oz beers. That would be a minimum of about 1 beers a day for the left side of the scale (although that starts at 2 so really about 1-2 beers a day is the lowest group), versus people who have about 5-6 beers a day.

I'm having difficulties understanding the quantities of alcohol we're talking about here. Can anyone enlighten me? If, as the NYT article suggests, the mid range was with 2 beers a day (which would fit with my 100L scaling above), I have even more trouble believing this silly hypothesis that the depressive effects of moderate alcohol negatively impact scientific work. After all the data is pretty level with a +/- bounce of 0.5 from 2-4 liters, or approximately 1-4 beers a day. Then there is a group of 5-6 beer/day drinkers who yank the line down giving it a pretty poor r-square value. I think this is a confusing paper with inadequate data and an improper line fit. At 5 or more beers a day you're talking about pretty heavy use (not that I haven't thrown back more than 5 in a day but not every day). Isn't this really a study showing that alcoholic avian ecologists don't publish as much as non-alcoholic avian ecologists?

Tomáš Grim (2008)
A possible role of social activity to explain differences in publication output among ecologists
doi:10.1111/j.2008.0030-1299.16551.x

More like this

I refuse to accept the results of this study: After years of argument over the roles of factors like genius, sex and dumb luck, a new study shows that something entirely unexpected and considerably sudsier may be at play in determining the success or failure of scientists -- beer. According to the…
What is the difference between Free Access Beer and Open Access Beer? You go to a bar to get your Free Access Beer. You sit down. You show your ID. The barista gives you a bottle. You don't need to pay anything for it - it's free, after all. You take your own bottle-opener from your pocket and…
A New York Times article has appeared about a study on the effects of excessive beer drinking on scientific productivity. The study, (Tomas Grim, "A possible role of social activity to explain differences in publication output among ecologists." Oikos 117 (4), 484-487) done by the aptly named…
As you may have already heard, ScienceBlogs is expecting to reach the 1 million comments mark sometime this month. And our generous overlords want to throw a party...actually several parties. They've encouraged ScienceBloggers to take their readers out for drinks, and the parties are happening all…

Do people on average really drink one or more alcoholic drinks (never mind beer for a minute; let's generalise) per day?! I will admit I'm not much of a drinker -- I used to be able to count my alcohol consumption in drinks per year on at most two hands, but now I'd say I have drunk a lot if I have one drink a week. Still, I'm kind of baffled by the idea that there are people out there who aren't alcoholics who might drink more than one drink per day.

Maybe that's why most people seem so frickin' weird to me -- all yawl are half in the bag all the time by my standards...

By Interrobang (not verified) on 20 Mar 2008 #permalink

Interrobang are you drunk?! A drink a day is not a lot by anyone's standards unless you're 10. I mean, I don't believe in God except for the fact that he gave you a liver - live a little.

I remember when I worked at Rockefeller it was normal for scientists with "not-alcoholic" reputations (and me) to drink a couple of beers at lunch - how much the world has changed. There was an on-premise bar we would go to while waiting for very slow tests to complete and it was a place of interesting debate and discussion.

I admit I have changed with society in the intervening 20 years (the RU bar has since closed) and tend to not drink during daylight hours (unless it's saturday and I'm at the free Don Julio bar at SXSW).

And I definitely don't think there is a correlation between dry countries/counties and higher scientific accomplishment!

Maybe there's something about the scientific process I'm not getting here, but I'm more astonished at the way people can churn out negative numbers of publications. Or is anti-science counted as minus points?

The y-axis reflects some RR value for likelihood of publication he made up, not number of publications. I think each tick represents one less publication than expected per year.

Fundamentally I find this paper a bit annoying because I don't think the conclusions match the data, or an r-squared of 0.5 represents a good linear fit.

Wine, the way to go.

By Ian Findlay (not verified) on 20 Mar 2008 #permalink

Thanks, Mark. All in the labelling, eh?

I never have and never will produce a scientific paper, and for about a microsecond I thought about helping you out by going teetotal, but then I thought, what the hell, and had a beer.

Being a psychologist it's always irritating when researchers outside of the behavioral sciences delve into psychological research and think they will do just fine. I'm not advocating for any type of disciplinary ethnocentrism here, just that research like this isn't as easy as it seems, and there are many methodological and analytical concerns that are unique to the behavioral sciences.

Besides the obvious issues in quantifying beer consumption, it doesn't seem apparent that the author has controlled for other demographic variables, like age, gender, ethnicity, current occupation, institute that granted the participant's degree, or field of study. While I don't dispute the argument that beer consumption CAN influence academic productivity, any suggested conclusions being made here are based entirely on correlations, and we all know that correlation is not sufficient to imply causation.

Isn't it more likely that younger individuals in the field drink more beer, and being in the beginning of their career are also less productive? Or possibly those individuals that are less productive have more free time, and spend it drinking beer? Or maybe the types of jobs that the less productive individuals are in allow for greater beer consumption, but not the opportunity to publish as frequently?

Here publications are a proxy variable for what is actually being measured here, competence in one's field, which is ultimately determined by a number of factors. Beer consumption obviously would not affect all of those factors related to publication rates, so an effective study would look at the relationship between those behavioral factors and all possible influences on those factors, positive or negative.

The method that was actually used seems about as sound as asking the researchers what kind of car they drove and then concluding that those researchers with nicer cars were more productive because they had more reliable transportation.

I suspect the editor of Oikos let this through because he thought it was amusing (this same editor entertained us a few years ago by demonstrating what happens when you put chocolate into beer). I doubt he expected anyone to take it seriously.

Incidentally, for those who are confused, it's in the April issue of Oikos, not the February issue. They've got themselves a bit muddled over the dates. However, the February issue does have Daniel Rankin's paper "Can punishment maintain sex?".