Heterogeneity in quality of recommendation letters

[good_tea]

Hello,

I just finished registering my letter writers and I was thinking about how universities would interpret differences in evaluation. In most recommendation letter forms, the recommender has to specify how good the student is (top 1%, top 2%, top 10%, bottom 50%, best ever, etc... - possibilities vary from place to place). For many (most?) of us, different recommenders are likely to mark us differently. How will this be accessed? Do you think more weight will be given to the strongest or to the weakest evaluation? I think (other things more or less equal) special weight might go to the strongest one b/c it might reflect more knowledge. For example, a professor that one does research with might have a stronge basis to confidently state that a student is top 1% compared to someone with whom the student took, say, took regular classes.

Another related question concerns how highly differently ranked universities regard themselves. Are top 20 happy with top 10%? top 5%? top 2%? More importantly, what do you think would be positively harmful? top 20%?

I realise these are more or less impossible questions to answer, but I thought it is an interesting topic for speculation.

- Good Tea (I am ironically drinking coffee right now)

[Gecko]

Haha, I would think most writers would be selling their candidates as the best thing since sliced cheese - if not, you should target some other writers..

[AREStudentHopeful]

In reply to Gecko's post: they'll only sell you if they think you're worth the sale. By writing a letter of reference that states that you're the best student they have ever had, they're putting their neck out there for you. If they don't think you could handle the load at the school you're applying to, it's in their best interest to say so to maintain their reputation.

[walt526]

As far as the threshold ranking for a "good letter," it depends a lot on the perceived quality of the undergrad institution. For example, "Top 20%" from Stanford means something quite different than "Top 20%" from a CSU.

It's also the case that if the letter of reference author is unknown (or simply doesn't write a lot of letters to top programs) than his/her claim of "best ever" or whatever won't carry as much weight as an established professor saying "Top 5%."

So to your original question regarding heterogeneity in the letter of reference ratings... I'm 99% certain that no systematic analysis is done comparing average ratings of two candidates. It's more of a causal way to quickly survey whether or not the letter of reference is worth reading insofar as promoting an applicant (if the rating indicates a weak or negative opinion, then it's most likely not even read).

As a general rule of thumb, any rating below the "median" selection (inclusive) is probably considered as mediocre.

[Golden Rule]

The text matters a lot more than the numbers I'm sure. An informative letter of reference that says you're top 20%, but still can convince an ad com you're well-suited for grad school, is worth more than a vacuous one that says you're top 1%.

[good_tea]

Thanks for the answers. I know that they don't actually aggregate the numbers statistically. I was thinking more along the lines of one recommendation being *extremely* strong and one being strong but not that strong. I guess the text does matter a lot and they would look at what the basis for each recommender's assessement is.

Does somebody know how many people are in the admissions committe typically? Is there anyone that will always be in the commitee (e.g. head of department for instance)?

- Good Tea (I am drinking tea this time).

[Gecko]

The initial perusal will be by the less senior profs I believe, since they have better things to do with their time