Friday, April 14, 2006

The 'Et.al.' Effect

In academic economics, the convention in papers with many authors is to order them alphabetically. This is in contrast to other academic disciplines that have their own norms including ordering based on relative contribution.

It was always somewhat surprising that economists might choose to 'hide the individual contribution' by adopting a convention that meant that name-order should bear no signal. In the 'cuteness over substance' phase of my academic career, I wrote a paper on this topic. The paper (with Maxim Engers, Simon Grant and Stephen King) was published in the Journal of Political Economy. We demonstrated how this might be the natural result of bargaining amongst co-authors. Put simply, to deviate from an alphabetical ordering norm would cost the author with a surname earlier in the alphabet alot more than would be the gain to the other author. Hence, this would only occur rarely; reinforcing the norm. [Oh, and yes our 4 authored paper was ordered alphabetically but only after vigorous bargaining].

The only effect we thought this would have would be to reduce paper quality -- after all, without a signal of individual contribution, free riding would ensue. Now it turns out that there is something more sinister at work: your surname matters for real academic success in economics.

In a paper that will appear in the Journal of Economic Perspectives soon, Liran Einav and Leeat Yariv look at the actual effect of a surname's first letter on academic performance in economics. They find that those with surnames earlier in the alphabet are more likely to have tenure at a Top 10 economics department and win awards such as the Nobel prize. There is real alphabetic discrimination in economics and it is not accounted for differences in nationalities, age and other trends in academic co-authorship. If you look at psychology, for instance, none of this happens and surnames have no effect on success.

How can this be? After all, an alphabetical norm was supposed to send no signal and not be meaningful as it appears to be. Einav et.al. suggest several possible channels for this effect. The first is that in citing multi-authored work, others use the first named author plus 'et.al' to save on space. Maybe the 'et.al.' obscures all. Similarly, citation measures such as the Social Sciences Citation Index used to similarly only report first authors. This may distort the number of attributed citations for others and harm their chances of promotion or awards. Of course, what is strange is that the surname effect only appears for Top 10 departments. Go to others and surnames do not matter.

Another possibility is that this effect influences participation decisions for multi-authored projects. It is suggested that a disadvantaged surname holder will opt out of those projects but this, in turn, would disadvantage them academically as those projects may be more successful. This, however, isn't borne out at surname is not a predictor of such participation.

So if surname matters and it matters more 'higher up' in the profession, what is causing it? In the tradition of Larry Summers, I suppose that we can't rule out that the causation runs the other way. Maybe those with surnames earlier alphabetically are better economists. As with all discrimination, it is hard to imagine how that could be.

Of course, it could be a selection effect. Good students with names later in the alphabet here that they will be ordered last. They make choices between economics and psychology and go for the latter. Of course, this would suggest that a reverse discrimination effect should occur elsewhere but it may not show up because the other options are diffuse.

This could explain why the surname effect is not showing up as we move down the academic rankings. It could be there but it could be offset by the reverse discrimination as those denied tenure at top departments move their way down.

Something more is at work here. This seems to me to be a real puzzle that will do a nice job in diverting academic economic research towards the cute.

7 comments:

hc said...

Joshua you say it yourself. Isn't this just 'cute'? Does anyone care? Maybe you can learn something about bargaining with this work - or something else - but why do it? Is it just to display technical virtuousity?

Economics does have aesthetic dimensions so this type of work might pass it on a curiousity-aesthetic heading. But I do wonder.

Anonymous said...

Harry,

I don't get what you are saying. Care aboit what?

Joshua Gans said...

Harry is obviously part of the favoured 'C' class in academic life. Rabee you are in the unfavoured 'T' class. The privileged always protect the system!

hc said...

Rabee, aboit -> about (you always correct my spelling!). Care? Does it really matter about authorship orderings? Is answering this type of question an aesthetic concern or is it developed to display technical vituosity? Or indeed is there intrinsic interest in the issue either because you can learn from it or because you lie awake at night agonising over it? My guess is not either of the latter answers as even Joshua describes this type of inquiry as 'cute' which sounds a bit pejorative or suggestive of impracticality. But that still leaves me wondering.

I am not making strong claims. I don't know.

Joshua, 'C' class doesn't sound much (even 'favoured' C) but neither does 'T' class especially as it is 'unfavoured'. I don't know what you are suggesting.

Kim said...

But is it just cute, do you think? As a law academic, I think there's an issue of interest here. In law, co-authored pieces were rare until relatively recently; they are gradually becoming more common. As such, there is no settled convention, yet, I would say on how to order author names. Where I am, we've settled on an approach that says if there is real inequality in contribution, names are ordered so that the greater the contribution, the earlier in the list of names you will be ranked. Where contributions are roughly equal, the names are put alphabetically.

Now, as a W, I felt hard done by with this approach: I've ended up on the end of a few lists as a result. The reasoning is alphabetical, but in the absence of a very clear convention, some might read that as indicating a lesser contribution.

If there was a real effect from alphabetic placement, this would strengthen my feeling that alphabetical ordering should be abandoned.

Anonymous said...

Joshua, I think this is an effect that has been observed in other fields. For example, "A" surnames figure disproportionately in the astronaut corps.

One explanation is that, as children at school, those with early surnames become accustomed to being called early for various class tasks, encouraging them to concentrate more and also to fight more fiercely for top rankings.

Anonymous said...

I just found 25 missing citations to coauthored unpublished papers of mine. Now I understand why a famous individual in the Z-class puts his name first on coauthored working papers and alphabetically when the paper is published.