Reference Letter Writers
- Be as specific as possible.
Provide anecdotal evidence to make your case.
- Give percentages and specify comparison group
Instead of "One of my best students" specify the top %, in which categories? Out of how many? Such information also helps comparisons that take place against nominees from other disciplines. For top-tier awards, like Vanier, consider that being in the top 15%-20% (or worse) is not considered particularly impressive.
- Avoid words like "excellent" to characterize the candidate and restrict value judgments to a minimum.
Let the value judgment be of the reviewer, and provide relevant facts to substantiate your case (as in (1)).
- Structure your letter and make sure it flows logically from one paragraph to the next.
A letter lacking this flow might be seen to be a cut-and-paste job -- this does not impress anyone.
- Avoid possibly ambiguous statements.
For example, a sentence like "... this award will help his/her productivity ..." or "... this award will improve quality ..." can be taken to obliquely admit a deficiency of the candidate (did the letter writer intend to convey this message?). Also avoid producing a sense of entitlement for the award which usually irritates the reader. Consider having a colleague (ideally from another area) "debug" the letter for you.
- A short letter, unless it makes a very bold statement, is probably a weak letter.
Not being able to fill a single page for a candidate you consider to be a top candidate calls into question whether the reference really knows anything about the candidate. Any reasonable letter, which backs with facts its recommendation is likely to require 1.5-2 pages to get into sufficient details. For top-awards, if the reference believes that the candidate is indeed exceptional, then this is not the time to be lazy.
- Don't talk about yourself.
For instance, don't include why you chose a particular research topic.
- Structure your proposal and use headings.
For example, NSERC tells you what topics to include -- structure your proposal and use headings based on those topics. Do not use too much space on generic background. Clearly identify WHAT you propose to do, HOW you plan to do it, and WHY it is important.
- Write for scientists or engineers who are not familiar with your topic.
Minimize the domain-specific acronyms you use and define them (on first encounter) before you use them.
- DO NOT list NSERC renewal as a separate scholarship, DO NOT list scholarships that are not based on academic merit, DO NOT list course projects as publications. BUT ... DO list NSERC USRA's.
- Mention special circumstances only when they are applicable to your case.
Do not include common circumstances like: "I didn't realize how important it was to do my homework," or "I had a part-time job."
- If you've had RAs, describe them in your statement of research experience.
- Read and proofread your entire application, many times.
Get others to read it. Use a spell checker. If you are not a native English speaker, seek the help of someone who is.
- For research-related awards, it is important to get a letter from your supervisor(s) / proposed supervisor(s).
DO NOT get letters from a teaching assistant, student, or postdoc.
- When required, for the particular award, provide evidence of community involvement, leadership, teamwork, communication skills etc.
This is usually done as part of the CV devoted to extra-curricular activities (or quasi-curricular ones, like study groups, tutoring, math Olympiads, programming contests, etc.). There is substance in such activities that could be very well appreciated by the reviewers.
- The highest ranked applications should be good in all categories.
Providing an excuse for low grades, or whose proposal is poorly-written can be perceived as a mistake.
- If there is something that sets the candidate outside the norm and is imposed by the particular study plan for the student (e.g., increased make up course load) the departmental cover letter (if any) should explain it.
Questions could be raised why two files for similar candidates include differences, e.g., in the number and nature of courses taken. Care should be taken to check whether a candidate has withdrawn from a course (usually a red flag for any committee) and provide an explanation, if there are extraordinary circumstances.
This version by yannis@cs with input provided from stewart@cs and macg@cs