Lucky starts in academia

The academic job market in philosophy is very competitive. But competition does not imply that success is determined by ability or by hard work. David Chalmers and Robert Frank are both very successful academics, one in philosophy, the other in economics. Both have commented recently on the significant role of luck in their lives and careers. They both are to be commended for these statements, which express sentiments that are too frequently ignored or downplayed. It is not very comfortable to admit that we are recipients of privileges that are not deserved.

But however valuable comments like these are, they do not help prospective students quantify how much luck is involved in achieving a more modest degree of success. Put aside the fame and attention that Chalmers and Frank have achieved: what does it take to simply get a tenured job with a reasonable teaching load? For the vast majority of academics like me, I suspect, we are usually incredibly grateful for the job we have. And if any new jobs are advertised in our departments, we marvel – nervously – at how strong the field is. But I’m not sure our students fully appreciate this.

That said, I don’t have any data to offer that directly illustrate the chances of getting a job. I thought it might be useful, however, for students thinking about pursuing an academic career to see some actual data on how hard it is to get publications. In Australia at least, for early career academics, sheer quantity of refereed publications is an important factor in your likelihood of getting a job. I suspect it is at least somewhat important everywhere. (I believe it is less important in the US than in Australia at present, but that’s largely based on hearsay.)

Michael Huemer has written about the difficulty of getting published (and read), and you can learn a lot from his refreshingly frank comments. But he does not give precise figures on rejection rates. Inspired by Frank’s and Chalmers’ comments, below I estimate the degree of luck involved in my own initial forays into academic publishing.

Having completed my doctorate in September 2003, I applied for a permanent position at Monash in January 2005. I was ranked second by the committee, but fortunately the person who got the offer turned it down, so I got the job. (This is a sort of luck that is hard to quantify!) That is where I have been ever since.

By the time I applied, I had made (at least) 17 submissions to journals, not including resubmissions after a round of refereeing. (It is possible that I have overlooked a submission – my records are pretty good, but not perfect.) Two of those submissions were made very shortly before I applied, so I’ll set those aside, as there was almost no chance of a verdict on them prior to the interview. So that leaves 15 submissions.

I had 5 papers and 10 rejected from those 15 submissions. How lucky was that?

At time of writing, looking at my career as a whole, I have had 20 refereed papers accepted (almost always after one or two rounds of revision and resubmission), and 58 rejected, for a roughly 25% success rate. That rate has been lower in recent years, which I attribute to my aiming at more prestigious journals than at earlier career stages. Before the recent decline, my average success rate was around one in three. So I’m going to use 0.3 as an estimate of what my probability of success was, per submission, at that stage.

(Of course, it is not true that each submission is an independent event. One would hope that later submissions of a paper have greater chances of success, reflecting useful feedback from referees that has been incorporated. But despite that hope, the assumption that each submission is an independent event fits the data reasonably well – which is unsurprising given how much noise there is in the refereeing process.)

 

A cumulative distribution function, showing on the vertical axis, the probability of at least 3 (blue), 4 (orange), or 5 (green) acceptances, having made n submissions, on the horizontal axis. Based on the assumption that each submission is accepted with probability 0.3.

A cumulative distribution function, showing on the vertical axis, the probability of at least 3 (blue), 4 (orange), or 5 (green) acceptances, having made n submissions, on the horizontal axis. Based on the assumption that each submission is accepted with probability 0.3.

On these assumptions, I had a 0.48 chance of getting 5 or more papers accepted from 15 submissions. The chance of of getting at least 3 accepted was much stronger – nearly 0.9 – but I doubt that 3 papers would have been enough to get me an interview.

If we do the same calculation, but assume a base probability of success per submission of 0.25, the chance of getting five acceptances or more drops to just 0.31.

So take ten candidates, all with similar levels of philosophical ability as I had, all working similarly hard to write papers and get them submitted. Of those ten, we can expect roughly five of them to have CVs that looked significantly worse than mine over this period of time. Those are five people who may not have made a shortlist to be interviewed. Five people who might have been forced out of the academic game at an early stage. And that’s just taking into account a fairly readily analysed dimension of luck – many other factors, less fairly distributed, less easily controlled for, are likely to be in play, as Chalmers mentions.

I don’t draw any particularly new morals from this. It affirms the gratitude I already felt for my lucky situation! It perhaps reinforces that as well as having ability, sheer persistence is a very desirable trait in this environment. I'm also a bit shocked at how many unsuccessful submissions I had made – I suspect I would have guessed a much more rosy story, prior to going back and checking the records. That’s an unfortunate bias in memory; one that could give a misleading impression to prospective students.

Of course, this is a very limited sample, and it reflects my own idiosyncratic abilities, my choices about when to submit papers, and choices about where to submit them. I’d be interested if any colleagues with similar records of their early publication attempts could provide similar data for comparison, to develop a more representative picture.

And for the record, here are the particular journals I submitted to over this period:

AJP: 1 acceptance from 2 submissions
Analysis: 0/1
BJPS: 0/2
Journal of Applied Phil: 1/1
Nous: 0/2
PQ: 3/4
Phil Stud: 0/1
PPR: 0/2