Its the question everyone has asked at one time or another. Its also the question every supervisor is asked by almost all of her or his research students. ‘How many references should I have in my thesis/chapter/conference paper/article?’ Put simply – How many citations are enough?
When asked this question my approach has always been to say, ‘As many as the piece needs’ and then to go on and talk about what references are for, why we include them, and how to decide which earlier writers to cite. You’ll know the sort of thing. You’ll generally need a reference if:
- you make an assertion of fact that is not part of the common knowledge of the society. For example, if you say ‘the sun rises in the East’, that doesn’t require a reference. If, on the other had, you say ‘As we all know, the greater bearded senior professor is most commonly found in the Senior Common Room during daylight hours’, a reference to the source of this tidbit of information would be most useful;
- you make a statement of the ‘Researchers (or studies) have found…’ type;
- you quote someone or some document (or even yourself!);
- you use someone’s general argument or structure;
- you want to close off a line of argument. For example, you write something like, ‘This last point raises a number of interesting questions about the nature of the day-to-day activities of the greater bearded senior professor, but these have been addressed elsewhere.’ Tell the reader where that ‘elsewhere’ is and be thankful you don’t have to do the research!
- you want to guide the reader to some interesting areas or readings that you don’t have space to pursue in your current piece of writing or which don’t quite fit your argument.
But did I think science would come to my aid in answering the ‘How many references are enough?’ question? I did not. But it has.
I found the following frame in a very recent post in PhDComics. Its a Figure from a 2002 article by Abt and Garfield and it shows clearly that there is a relationship between the average length of a paper and the number of references both of which vary across disciplinary areas but the relationship between the two is sufficiently strong for a formula to be derived which will guide you in the number of references to include. Here’s the Figure and the formula.
Of course, its not quite as simple as that and different papers will require different numbers of references even if they’re within the same discipline. But this will give you a guide and a sense of whether you’re in the right ballpark area. If you find you’re not, there may be very good reasons why not, but at least this will help to make you think about those reasons and be prepared to answer any questions about your citation practice.
In the meantime, isn’t science great!
A short Post Script (8 Dec 2015). I have just returned from a writing retreat where I presented on a couple of issues and introduced the formula for calculating how many references a paper should include. One of the participants caught me after the presentation and said that he’d run the formula on his draft and it said that for its length the paper should have 72 references. He actually had 68. Another satisfied customer!!