PDA

View Full Version : Letters: Disutility of Choice



ColonelForbin
11-05-2014, 05:44 PM
This is my situation.. we can think of it as hypothetical, becuase I think it'll make a bit more sense that way:

I have four potential letter writers.. all four are going to be great to excellent. Now, the typical argument is that you should only send 3 letters, because on the margin, one of those letters is going to decrease the mean quality of your letters. However, in this case I have these four (potential) letters:

1) 2 yr full time RA
2) 3 yr part-time RA + thesis advisor + 2 courses
3) 1 yr part-time RA + field work overseas
4) 2 courses professor + strong personal relationship

In this case, I definitely choose professor 1 & 2. I'm certain that the letters from (3) and (4) will be glowing, but they will highlight different features: (3) likely project development and implementation and (4) my ability as one of his top students in the past X years. Think of both of these letters as being of the same overall quality but with focuses on very different and rather different skillsets that hold some clout in the application process.

I want to send all four letters to graduate departments that will let me -- I do not see any of these letters as substitutes for any of the other letters but rather as complements, and I think the overall quality of my letters will fall only if I omit one of the four.

I think this sits outside of the realm of "choose the best 3 letters", what are your thoughts?

chateauheart
11-05-2014, 06:02 PM
You can ask all four to write letters but randomize in selecting three of them for applications. This helps reduce the risk of the unlikely but potentially damaging possibility that one of the four is lukewarm or unflattering, or just poorly written. It also reduces the professors' average burden somewhat if you have a lot of applications.

HowDoIUserName
11-10-2014, 03:50 AM
You can ask all four to write letters but randomize in selecting three of them for applications. This helps reduce the risk of the unlikely but potentially damaging possibility that one of the four is lukewarm or unflattering, or just poorly written. It also reduces the professors' average burden somewhat if you have a lot of applications.

I'm curious, what is the consensus on what the OP suggested (i.e. sending more than 3 letters when the application in question allows for it and you are not convinced that there are quality differences between your top 3 and next best options)? Where the fourth letter is weaker than the third, obviously this lowers mean quality, but is it universally inadvisable?

NBZ
11-10-2014, 07:07 AM
I'm curious, what is the consensus on what the OP suggested (i.e. sending more than 3 letters when the application in question allows for it and you are not convinced that there are quality differences between your top 3 and next best options)? Where the fourth letter is weaker than the third, obviously this lowers mean quality, but is it universally inadvisable?

The consensus is that it is very often inadvisable, and I think it's not just because it lowers mean quality (because if mean quality were the only criterion everyone would just send their best letter and nothing else). It's more that most schools explicitly discourage applicants from doing so, which to me seems like a good enough reason to send just the three.

There is also no guarantee that the admissions committee will read all four letters. Suppose they randomly select three letters or select the three weakest letters (disclaimer: I have no idea whether they actually do this, but it seems like a reasonable enough procedure when you are comparing applicants who sent 3 letters with applicants who sent 4). In that case sending all 4 letters could backfire.

Of course if all 4 letters are read, then that allows your letters to bring out your various strengths and skill-sets. But I think the marginal benefit from that is small compared to all the things that might go wrong.

ColonelForbin
11-10-2014, 02:34 PM
Ok -- then in this case, I will use letters from #1 and #2... but now the conundrum: Letter (3) will focus on my ability to implement field experiments and my research and letter (4) will say I am a great student but be less research oriented (he knows of my research, but didn't advise me directly). The source of letter (4) is more senior than letter (3). Think emeritus versus assistant prof.

Perhaps I'll randomly select one of 3 & 4, but I'm leaning toward (3) since he can speak to my research potential a bit better. Ayiy, the marginal decisions that make this process such a PITA.

ColonelForbin
11-10-2014, 09:23 PM
There is also no guarantee that the admissions committee will read all four letters. Suppose they randomly select three letters or select the three weakest letters (disclaimer: I have no idea whether they actually do this, but it seems like a reasonable enough procedure when you are comparing applicants who sent 3 letters with applicants who sent 4). In that case sending all 4 letters could backfire.


I doubt this is what they would do -- they would have to read all four to determine which three are the weakest. I'm sure all four are included in your packet and the choice of each professor to thumb through all four recommendations is idiosyncratic at best. I agree though, I imagine some would be bothered by the fourth recommendation, but if I was on an Adcom, I'd read it, because if someone had the gumption to send four, then it must be saying something that the other three are not. If it didn't say something orthogonal to the other three letters -- I'd be disappointed in the judgment of the applicant to include the additional letter.

This being said -- I'm still on the fence about sending four letters to programs that allow it.

rekovu
11-10-2014, 10:37 PM
You should send three, but don't randomize. Send the letters to the schools at which they'll have most effect. E.g. if one of the letter-writers got a PhD at a particular school, or does work in a field that people at the school appreciate, etc., then make sure to send those letters to those schools. Don't have the Minnesota macro guy write a letter to Caltech.

ColonelForbin
11-13-2014, 01:38 AM
I've decided to send all four to programs that allow it, for those who absolutely do not allow it, I chose wisely.

My expected MB is greater than the MC of choosing at this point. If they read only three, so be it. If they read four, I think they'll get a better idea of my aptitude as a student and researcher.

I did dwell on this for quite some time... the four letters all add significant value and they all highlight different attributes. They aren't redundant, and in fact, I think they complement each other nicely.

I appreciate the feedback, but in this case I really do think I am an exception to the rule of only send three letters.​

Econhead
11-13-2014, 01:53 AM
I've decided to send all four to programs that allow it, for those who absolutely do not allow it, I chose wisely.

My expected MB is greater than the MC of choosing at this point. If they read only three, so be it. If they read four, I think they'll get a better idea of my aptitude as a student and researcher.

I did dwell on this for quite some time... the four letters all add significant value and they all highlight different attributes. They aren't redundant, and in fact, I think they complement each other nicely.

I appreciate the feedback, but in this case I really do think I am an exception to the rule of only send three letters.​

There have been a lot of comments on this forum that have indicated that, perhaps, only the strongest letter is read. (Perhaps perceived to be strongest based on some heuristic, like whether they recognize any of the letter writers-and if so, who they consider the most prestigious). If this is even close to the case, then I can only imagine all letters (3 or 4) would be read when they are having a difficult time making a decision (i.e. at the margin). If you were to try to quantify this in any way, consider that the strongest letter should give the adcoms a pretty good idea of what they want to know; from there each additional letter, even with new information, results in diminishing returns. It helps, but not as much as that first letter did (in terms of impression).

A few students that were admitted to ~T10 last cycle indicated that they thought their 4th letter made some type of difference. Perhaps you will have a similar experience.

The best of luck this admission cycle, Forbin.

ColonelForbin
11-13-2014, 01:59 AM
Thanks all. I'm relatively confident my highest ranked letter will open some door. But I think I'll be a marginal T10 applicant, so maybe those other three will matter.

Gracias.