On the alleged correlation between beer consumption and academic productivity.

Here’s the link to the paper in Oikos.

Before we get into philosophical discussions, however, let’s look at what they actually showed.

Alleged paper/beer consumption correlation

The first thing to notice is that this analysis is over the range of 1-6 liters/capita/year. That’s 1 pint every 6 months up to once a month. Now, I don’t know any Czech ornithologists personally, but I do know several Germans, some Polish, and a couple Hungarians. Their spread of beer consumption rates among them is more like 1-6 liters/capita/week. Therefore, unless Czech ornithologists have a significantly different consumption from the regional average, one must assume sampling error is present among such a rarefied population. When you look at it, and I know this is what passes for great results among ecologists, but the correlation really ain’t all that great, is it?

Now, as we all know but often forget, mere correlation doesn’t imply causation, so it could be just as likely that low productivity causes beer drinking or that some third factor causes both low productivity and beer drinking. What could that putative third factor be? Could it be that people who tend to…ahhh…misrepresent themselves tend to have higher publication rates (until peer-review catches up with them, of course), and would also, on this near-teetotaler end of the drinking scale, tend to under-report their consumption? So all they’ve really done here is show that people who lie on surveys get more publications!

To actually make one serious comment, let me say that it does make sense that someone who has no life at all will spend more time in the lab, but since 99.9% of all researchers worldwide already fall off the right side of the chart, how useful is this information?

3 thoughts on “On the alleged correlation between beer consumption and academic productivity.

  1. I agree with a lot of your post but I think that there is something about the axes on this figure that is unclear (at least to me). In the text Grim says that the consumption of the Bohemians had a median value of 200 litres/capita/year and the Moravians 37.5 litres/capita/year. This doesn’t seem to match with the scale on the x-axis. As you say the data from the graph indicate a much lower consumption. Many of the points on the y-axis (publications/capita/year) have negative values which seems to imply that these are not actual values- unless they are also counting retractions! The figure legend says the data has been Box-Cox transformed. I don’t really know this transformation but I guess the answer lies here somewhere.

  2. Yeah, they did something to the numbers. Maybe there are few enough Czech ornithologists that people could figure out who’s who from that sparse data, but honestly I just got a laugh out of it.

    The correlation is far too weak to be meaningful, anyways.

  3. Here’s my hypothesis: grad students and post-docs are bigger beer drinkers than senior researchers (more Friday-evening bashes, less family responsibilities), grad students and post-docs have fewer papers than senior researchers (just because of the amount of time they’ve been working), therefore people who drink a lot of beer publish less than people who drink a moderate amount of beer. Case solved 🙂

Leave a Reply

Your email address will not be published. Required fields are marked *