1. Cultivate good references.
Get to know a few professors well enough that they can write you a strong reference letter. A letter stating, "She was in my class. She sat in the front row. She got 85%." usually isn't worth much. There are many ways for students to get to know professors. Take small upper-level courses in which the professors can get to know you. Ask questions in class or after class. Meet with your professor if you have questions about the course or your career plans or opportunities in your field of study. Check out the Western Work Study Program and Western's Volunteer Program. Consider doing an Honours thesis or independent study course. Consider doing off-campus volunteer work in London. This page contains other good ideas about getting refs.
Grad Students and Post-docs
Get to know professors other than your advisor. For example, have discussions with members of your advisory committee and colleagues that your advisor works closely with. If possible, take the opportunity to work on projects with other faculty members (assuming it won't hamper your degree progress and your advisor doesn't object).
2. Determine who your strongest references will be for the target audience.
Who knows you the best?
Who thinks most highly of you?
Who will carry the most weight with the target audience?
If you are unsure whether your referee thinks highly of you, you may wish to ask, "Can you write a strong letter of recommendation for me?" Most reference letters are positive and differences between candidates fall between the good and the great ones. As a speaker in a grant-writing seminar once said about academic competitions, "everybody is a large egg so you have to be an extra-large egg." Except under unusual circumstances, you should ask for references from your advisor and others who know your research well rather than others who don't. If you do not ask someone who you've recently worked closely with, reviewers may assume it's because that reference would not have been positive.
If you're applying for grad school, a reference from a professor with whom you have done research will count for a lot more than one from your boss at Walmart.
References from friends and family are useless -- don't bother.
Grad students and Post-docs
For scholarships, particularly at the Ph.D. level, you should ask the people who are most familiar with your research, particularly your recent research. Master's students may not know many professors and may have to ask course instructors and professors who don't know them that well. However, for Ph.D. awards, research counts much more than coursework (whereas, Masters awards tend to weight grades heavily). It is best to ask your advisor plus another professor (e.g., an advisory or thesis committee member) who knows your research work.
3. Ask your referees to write letters, ideally NOT at the last minute. You should give at least two weeks notice from the date the referee needs to send the letter.
Yes, your referee is most likely going to leave the task till the last possible nanosecond, but if you're late asking them, it's even more of a gamble. You might discover that they're away at a conference or they may be swamped with dozens of requests in peak reference seasons (early fall scholarship time and end of calendar year grad school application time). A last minute request is likely to make them grumpier when they fill out the ranking forms. Bad planning on your part does not constitute an emergency on your referee's part.
4. Minimize the amount of effort your referee must expend. Don't make them work unnecessarily to do you a favour.
If you have multiple letters, compile a list containing one section for each letter that includes:
- the agency, program or potential employer;
- If you are applying to an agency that offers multiple awards or a location with multiple job postings, make it clear which one you are applying to (and make it clear if you are applying to more than one).
- the address if it is to be mailed directly;
- Format the address so that it can just be cut and pasted into letters and envelopes. It is even better to give your referee address labels to put on the envelopes.
- Most faculty have funding for postage, though if your referee is poorly funded (e.g., a post-doc), it's a courteous gesture to put on the postage.
- If you've really left it till the very last minute, offer to pay for a FedEx
- the date the reference is due and when it must be sent or when you would like to pick it up
- any additional forms, ideally filled out with all your information possible including:
- your address
- your PIN
- the award you're applying for (e.g., for NSERC, Master's vs. Doctoral award)
- which form your referee should fill out (e.g., for OGS, there are two letters, one from the "person most familiar with your research" and one for "a person familiar with your research)
- If possible, figure out what reference letters you need and ask for them all at once rather than one or a few at a time.
- Reference letters usually ask how long the referee has known you. If it's been awhile, it helps to remind your referee of the dates when met them or worked with them.
- If there is anything in particular you think your referee should mention in the letter, discuss it with them. Consider giving your referee a copy of your grades (unofficial okay), CV/resume, writing samples, reference letters from prior employers if it would help them highlight your skills and accomplishments.
5. Make sure the reference letter gets sent in time.
Don't just assume your professor is organized and efficient enough to remember to write and send the letter. A surprising number of reference letters never get sent. Usually, it's a case of absent-mindedness, but it can be taken as a negative evaluation. Remind your referee as the deadline approaches and be sure they follow through.
For students asking me for letters: To be totally sure that I don't forget, calculate the date by which the letter must be mailed or picked up. One week before this date, send me a gentle reminder by e-mail. Then 2-3 days before, start nagging me with increasing frequency until I tell you it's done.