Reader query: how many recommendation letters?

A reader writes,

I was wondering about your earlier post on the Philosopher's Cocoon where you said that the more outside letters you had the more interviews you got. How many letters did you end up sending? I could probably get six strong letters. Three of them would be from outside both my current department and the department where I earned my PhD. But I wondered if it would be helpful to get that many? Also, is it mark against one if one sends more letters than the ad asks for?

To answer the reader's first question, I did best on the market with six or seven letters. Alas, as I will now suggest, I'm not sure there is a "sweet spot" for how many letters to send. I suspect it is largely a matter of judgment, as well as trial-and-error. Fortunately, as I will also suggest, the evidence I've come across indicates that one should not worry too much (for the most part) about sending too many letters. Generally speaking, and certain cases aside (see below), candidates appear to do fine on the market utilizing anywhere between four to eight letters.

Interestingly, I came across a facebook thread on this very topic earlier this week, and the answers people gave were pretty telling: they were pretty much all over the map. A lot of people said they thought five letters are optimal--yet others reported doing splendidly on the market with seven or even eight letters. Two answers appeared to be the most common, however:

1. Send no more letters than the job ad asks for (if it asks for N letters, send no more N),
2. Otherwise, five is probably optimal (three from faculty in your grad department, and two from outside people).

Are these good rules to follow? I'm not entirely sure.

First, there is plausibly a difference between a job ad that requests "N letters of recommendation" and one that explicitly states "no more than N letters." In my experience, both types of ads exist--yet the first type of ad is simply ambiguous: it could mean "no more than N", or it could mean, "at least N." Fortunately, oftentimes the online application software the department school uses will clear things up, only allowing N letter uploads (in which case, just go with N!). Other times, however, the application software has no upper limit. In these cases, I would suggest that the safe thing to do is to simply shoot a quick email requesting clarification from the contact person in the job ad. I suspect no harm can come of this. Indeed, if anything, it might make you look conscientious, and at the very least it will enable you to get a clear answer, sending the right amount of letters.

Second, while five letters sounds pretty good as a rule of thumb (and one definitely needs one or two "outside letters" after one has been out of grad school for a year or more), my own experience--and the reported experiences of many people on the aforementioned facebook thread--go against the "five letter rule." I did far better on the market with six and seven letters than I did with five, for instance, and other people on the facebook thread reported doing just fine with seven or even eight(!) letters. Accordingly, I don't think there are clear grounds to trust the five-letter rule of thumb as anything more than just that: a rule of thumb. Setting aside cases where a department explicitly says "no more than N", all of the evidence I've come across is simply unclear: some people think more than five is too many, yet others do really well on the market with more than five.

What to do, then? I guess the upside of the evidence--at least as I see it--is this: aside from sending more than N letters when a job ad says, "no more than N" (which seems like an obviously bad idea), anything between four to eight letters seems okay. One should use one's best judgment, and try to learn by trial and error. This is what I did, and it seemed to work. Three years ago I sent six letters and did okay on the market. The next year I send five and did a little worse. Then, last year, I send seven and did a whole lot better. Hard to pick out a clear pattern! :/

Anyway, what do you all think? Better yet, what has your experience been? Job candidates: how many letters correlated with your best job-market performance? And search committee members: do you have any clear preferences when it comes to letters? Thanks, in advance, to everyone who chimes in!

Comments

[Lost]

I have somewhat related question: are there people from which it would look bad if you *didn't* have a letter?

I ask because I had two advisers in graduate school and two advisers as a postdoc. If it looks bad to not include the letter of an adviser, then I'm rapidly approaching a large number of letters.

[Marcus Arvan]

Lost: I can only answer on the basis of personal experience, but I think it does look bad if you don't have letters from certain people--specifically your advisors.

A few years ago, after getting two new outside letters, I dropped one of my grad adviser's letters from my portfolio. I did so because I worried both that (A) I had too many letters (seven), and (B) his letter might not be the strongest. Alas, that year, my # of interviews precipitously dropped. The next two years, when I included his letter again (using up to seven letters), my # of interviews soared. So, although this is admittedly an anecdotal experience, it was a striking one. The number of interviews I had correlated strongly with whether his letter was in my file--and I certainly didn't appear to be hurt by sending up to seven letters.