Tag Archives: letters of recommendation

How to write (and read) references

I believe he has a perfectly charming wife

DarthI have a very simple approach to writing references, inspired by the story behind the quotation  “I believe he has a perfectly charming wife”. According to a story circulating in Oxford and Cambridge in the 1970s the quotation is a complete reference, written according to the principle that a reference should say all the good things that you can honestly say about its subject.

I used to believe that the principle was sufficient. A reference like the quotation, that says nothing about the candidate, is a dire warning. However, in the real world, where references are commonly leaked to their subjects, it is not enough, so I have added another principle. I include important and relevant facts but I take care to express them a really positive way. I’d say that I use two principles to guide my reference writing.

  • Use facts to tell the story.
  • Express the facts in a really positive and upbeat way.


In addition to these principles I find it helps to use a standard structure. I write a reference in  three parts. The first describes how and how well I know the subject; the second covers things I know directly because of their relationship with or work for me and the third covers what I know indirectly because of their work with or for other people.

Starting the reference with a description of how and how well I know the subject helps the reader to decide how much weight to put on my assessment. Writing brilliant undergraduate essays for me in fortnightly tutorials 10 years ago is less relevant than working under my direct management line last year.  This is also the first part that I write because it helps me to compose the more difficult parts of reference both by allowing me to warm up by writing something easy to write  and by jogging my memory for the really important facts that will follow.

Positive references

Facts improve a positive reference because they allow the reader to make their own judgements. It’s frustrating to read a reference that really only tells you that the writer likes the subject of the reference and thinks they are wonderful. Saying what the subject has done and what were the consequences is much more useful.  Bear in mind that things that are important to you may not matter to the reader and vice versa, so it is much more useful to them to get information about how subject has done their job than to know you think they are wonderful.

References for people you hardly know

Facts are very useful when you have to write a reference for someone you hardly know. Most former undergraduates come into this category. When I was an academic I had 30 or 40 tutees whom I would meet seven or eight times a year, in groups of ten. When they left after three years I could just about remember their names. Fortunately a detailed marks transcript provides an excellent set of facts around which to frame a reference.

Negative References

Facts are pure gold when you have to say something negative in a reference, particularly now that even a confidential reference can be obtained through a Freedom of Information request. My approach is to state facts that give a very clear message but to put a positive spin on them.

I once had to write a reference for someone, let’s call them Dr D*. Dr  D was a delightful person but  never quite learned how to run an experiment in my lab. They had applied for a job in Professor Q’s lab, which was much more complex than mine. I thought they would be a disaster and I thought it was important to say so clearly in my reference. A bad post-doc can cripple a PIs prospects. In my reference I wrote about how difficult Dr D found the complexities of the lab and how hard they worked to learn what they needed. I said that after more than a year in the lab Dr D knew about 90% of what was needed to carry out an experiment.

This statement was defensible if Dr D should ever get to see it. It was arguably a bit generous. There was a catalogue of minor disasters to justify it, each of them excused by the statements like “It’s very complicated to operate that piece of equipment”, or “I keep forgetting that you always have to do Y before Z“, and the like. Although I had stated an uncomfortable truth, it was a truth supported by evidence and I had stated it in a very positive way.

The statement was a clear warning to Professor Q*.  The glass that was 90% full should have been overflowing.  Imagine my surprise when Professor Q gave Dr D a job. I was less surprised a few years later when Professor Q took me to task for writing such a glowing reference for Dr D. She clearly hadn’t read it carefully.

Reading a reference

It is essential that you read a reference carefully and look for statements of fact. Sift the facts and decide which of them allow you to draw important conclusions. Usually the conclusions will be run of the mill. They will confirm important but dull facts like employment history, reporting lines, and achievements. For this reason most employers require a reference from the line manager. I always insist on one. I don’t expect to learn much from what it says but I do think it’s important that someone has a good enough relationship with their boss that they can ask them for a reference.

References almost always contain a lot of unimportant stuff about who likes (or is prepared to say that they like) whom.

Just occasionally a reference will contain something really important. It could cripple your lab to employ a post-doc who can’t quite do an experiment after a more than a year of struggling to learn how. Professor Q discovered this the hard way. She could just have read the reference.

*Neither Professor Q nor Dr D exists. I have invented them both to make a point.