How to write a reference letter?

• I have been asked to write recommendation letters for an undergrad who is applying to grad schools. Should I just automatically use a separate sheet, or should I use their form if they provide a lot of space? Any general tips for the structure of the letter? I'm only a post-doc, so this is my first time writing a letter of reference. One of the schools wanted 4 letters, so I agreed to write one for her (and once I've written one, we figured that I might as well send one to all the schools). She was under my direct supervision in the lab for about a year, so while I can't say much about her academic work, I do know about her research qualifications (she made 10-15 peptides by solid phase fmoc synthesis, then went on to use mol. bio. techniques to express 2 different full length proteins, in both cases doing all the steps including purification and characteriazation). Am I doing her a favor with the schools that only asked for 3 letters? Any guides or examples on the web?

• Answer:

  I've written a few references for undergrads for jobs, and I'm a grad student. As long as they're getting other recommendations too, there's no problem with a supervising post-doc writing a letter. Here's some things you can think about mentioning, from some letters I've written in the past: --the amount of control the student had over their project --if the student picks up new protocols quickly and has "good hands" --if the student understood the "big picture" of their research project and/or had good ideas about future directions for their work (there are always students who can do what you tell them, but you can tell they don't have any clue what is going on) --if the student was willing to work long and odd hours --if the student would ask questions and seemed genuinely interested in learning more and really understanding the work --how did they deal with problems in their research? Did they ask questions when they didn't understand things? --if he/she ever did any presentations of their work, wrote up part of a paper, or did anything publicly to demonstrate that they performed and understood the research --are they bright, personable, trustworthy, an asset to your lab? Will you miss them when they're gone? If you can honestly highly rank this student among other students you've worked with in one aspect or another, then that's worth a mention. Something like "so-and-so is the most dependable/promising/enthusiastic/etc student I've trained." But don't over do it--nothing is more annoying than a recommendation letter in which somebody is over-hyping a person. Not every student can be the best you've ever met. Be honest, but try to stay positive. If you can't say anything positive, it is best to say you can't write the letter--in my opinion.