A safe and supportive forum for philosophers.
Owned & Moderated by Marcus Arvan (University of Tampa)
Contact: marvan@ut.edu
Blog mission & moderation policy
One issue that philosophers on the job-market face is how long to stay on the market in pursuit of a tenure-track job. Although each case is different (it took me seven years, and I've known a few people it took over 10 years), Lee Elkin posted this interesting analysis of how long it takes on average for those who do get TT jobs. Evidently the mean time is 1.86 years post-PhD. Here's a graph of the distribution:
There’s going to be bias here toward lower numbers, though, because of the structure of grad programs? I stayed at my PhD institution teaching but not defending for 1.5 years on the market (the first year very selective, the second year full-on), so in the data I’d look like a zero, but really I’m more like a 1.5.
I assume this doesn’t count people who never get a TT job, so it’s safe to interpret this as your odds of getting a TT job in yr n for any n. It supports something I’ve generally suspected: if you don’t have a TT job in the first two years on the market, your odds of ever getting one go way down. While there clearly are success stories of people getting a TT job after 5+ years, it looks to me like anyone still on the market after year two should be seriously planning an exit strategy.
a philosopher: that might be true – but don’t we need to know the number of people still trying to get a TT job for each yr n? I don’t think, without a lot of additional assumptions, that this graph supports the claim that this is your odds of getting a TT job for any n.
For example (and I don’t know if this is true), IF it were true that people 4, 5, 6 etc. years out were increasingly likely to stop searching for a job, this could affect the odds of those remaining at that stage.
So for most people would this mean their 3rd year on the market? As the first year is as ABD? I have to say I find this a big surprising, as with the top programs who have lots of students, it is getting more and more common to take post doc position for several years before getting a TT job.
Chris, perhaps, but I don’t really see what special assumptions I need. Also, if I’m 5 yrs out (or whatever) and so are you, and you drop out of the market, that shouldn’t affect my odds of getting a job in any special way beyond the fact that now there are fewer people competing for a job.
The calculation I’m doing is pretty simple. I believe the following equation is right:
P(Get TT job during time covered by chart) = # of jobs handed out / # of people on the market during that time.
The chart tells me, regarding the people awarded the jobs being counted by that top number, how long they were on the market. So I don’t see why the following equation for the relevant conditional probability doesn’t also hold:
P(Get TT job during time covered by chart | Have been on market for >=n yrs) = # of jobs handed out to people on the market for >=n yrs / # of people on the market during that time.
According to the chart, this new top number approaches zero pretty quick after 2 yrs, so that means your odds of getting a job go way down after those first two years.
Also, if you want to use Bayes, the relevant likelihood — P(Have been on the market for >=n yrs | Get TT job during time covered by chart) — clearly goes to zero quickly after n=2, so Bayes formula suggests the same conclusion.
A philosopher,
P (Get TT job during time covered by chart | Have been on the market for >= n years) = # of jobs handed out to people on the market for >= n years / # of people on the market during that time
Is that the right formula? I would have thought it would look more like:
P (get a TT job | N+ years) = P (TT jobs to N+ years people) / P (N+ years person)
As a thought experiment, imagine that there are 10 people on the market, and that every year there are 5 hires. Also imagine that every year there is only 1 N+ years person on the market. (Perhaps everyone else drops out.) Imagine that it just so happens that every year, that person gets a TT job.
In that case, calculating using your formula gets us:
P(Get TT job during time covered by chart | Have been on market for >=n yrs) = # of jobs handed out to people on the market for >=n yrs / # of people on the market during that time = 1 / 10 = .1
But it sure seems like it should be:
P (TT jobs to N+ years people) / P (N+ years person) = (1/10) / (1/10) = 1.
(Like I said, it’s late, and my math is iffy.)
Philjobs is probably the best place to get data on this topic, but it’s a sure bet that there are going to be limitations about what you can infer from the information that’s on there. A lot of schools — especially community colleges — do not advertise on Philjobs, and those colleges do have tenure-track positions (or the equivalent). The trends of who they hire could be different than what’s listed above.
Also, I think this graphical representation is not as useful as strict percentages would be. It looks like about 1/3 of TT jobs in the data set are taken by people fresh out of graduate school, 1/6 by people who have been out of grad school for 1 year, and the rest by people who have been out of graduate school by two years or more.
This graph also creates the impression that your circumstances become more hopeless the longer you are on the market, but I think that inference should be resisted. People often go off the market after a few years because they get tired of moving around the country, they want to start a family and need greater financial stability, they grow weary of academia, or some other idiosyncratic personal reason. In all such cases, they might eventually have success if they keep at it, but circumstances make it impractical or unwise for them to do so. My suspicion is that the distribution is caused in large part because there are so many fewer people on the market 3+ years post-PhD than those who are fresh out of graduate school.
As a follow up, I think your formulae assumes no correlation time on the market and the likelihood of getting a job (that’s the only way to get the non-relative formula you offered). But I took it that the independence of the variables was what was at stake—hence we need more info!
Craig and Chris are correct, I am wrong. Sorry for sowing confusion. I should have thought a bit more and broke out pen and paper earlier. No one look at my bad math too closely…
Leave a Reply