On Letters of Recommendation

A Conceptual View of Letters of Recommendation

It is tempting to believe that LORs are extremely useful in determining the most successful students. They are not. Letters of Recommendation provide little predictive power of academic success relative to other predictors such as grades and GMAT scores. In concert with other predictors, LORs are remarkably weak marginal contributors to the prediction of academic success. To some extent one could justify LORs based on the need to distinguish students who are already at the right tail of the grade/GPA distribution but it is unclear if they add much.

But, of course, LORs persist, presumably for a reason beyond path dependence. One very useful aspect of LORs is that they provide a narrative evaluation of the student. This sort of "high context" information is difficult to obtain from other sources of information such as grades, tests scores and other standardized or quasi-standardized measures. In addition, unlike other forms of narrative input (e.g. student essays and interviews) LORs come from an external source are thus viewed to be less likely to be corrupted by student influence.

Schools are generally guarded about revealing how they evaluate LORs and might not genuinely know the impact of the LORs on the probability of admission.

Regardless, LORs are important because they are a true "wild card" in the admissions process.

LORs are unusual in the sense that they are blind score, blind content, single-shot, externally provided instruments.

Blind score: Students are not made aware of their performance on this instrument. Contrast this with the GMAT where you are given a precise indication of your standing relative to other candidates.

Blind content: Depending on the school, you are generally restricted from seeing or even verifying the content of a Letter of Recommendation. In contrast, you are able to look at a copy of your college transcript to verify that it is correct. In the case of the GMAT, you are given a pretty good idea of the content of the instrument and how you will be evaluated.

Single-shot: In any given year, you are given only one chance to submit a set of LORs to a specific school. In contrast, candidates can take the GMAT many times and take the best score.

Externally provided- GMAT and LORs are both externally provided. In contrast, essays are provided by the student.

How is all of this important?

It shows that you must be very careful regarding how you select your letter writers:

They might write weak letters
They might be racist, sexist, bigoted, jingoistic etc. and adjust the strength of their recommendations in accordance with their prejudice (remember- if you cannot read the letter how could you ever challenge the truth or origin of the author’s assertions?)
They might be confused or just plain spiteful and include misleading or factually incorrect negative information about you.

To top it off, you have little ability to ever learn what was written about you in a confidential LOR (regardless of its veracity) and whether it was used against you. Of course, readers of LORs might disregard some weak recommendations as the product of angry or confused writers. One advantage of having several LORs per student is that you can see if a weak opinion is shared among the letter writers. Nonetheless, this is not something that one should rely upon (plus the reader might wonder if the author of the harsh review was simply more honest or observant than the other writers were). I have read many letters of recommendation over the years. Most of the LORs I have read were of limited relevance since the overwhelming majority were so positive (it seems that every candidates was somehow “extraordinaryâ€