Q: What is a 'reference'?
A: A reference is a pointer to a published source which confirms some claim that you are making. Note that posting to usenet is not normally considered to be a form of publication; references to prior posts do not count.
In the better circles the reference should exist, it should be reputable, and it should actually support the claim you are making. The following advice is for the ethically challenged.
Q: How do I find references that support my position.
A: Some would say that you should be familiar with the topic under discussion and the work that has been done in that area. For those fortunate to be so blessed it is usually easy to locate the requisite references.
For most of us this is an excessively onerous requirement and we needs must have recourse to more devious methods. The usual difficulties are either (a) we know very little about the topic under discussion or (b) we are advancing opinions widely at variance with standard opinion in the area in question. This requires some ingenuity in the matter of supplying references. Here are some tips for the novice.
As a general rule people will not check your references, particularly if they are difficult to find. This is good. Quite often references really do not support your claims in a satisfactory manner. There is the occasional pedant, however, who will attempt to check them. Defeat these gentry by adapting a reference style that makes it very difficult to actually find the references. Here are some guidelines:
If you combine these techniques, you might get something like this:
MacRae et al., CJES, v.33, p.1475.
This could be anything from a near-useless abstract to a full paper. Readers will have to guess what journal it is and then hope that the library has volume 33, and that there are not a great number of MacRaes if they have to go looking for the full citation using the various abstracting indexes.
In general it is fairly safe to fabricate references since almost nobody actually looks up a reference. There is the rare pedant who will try to look them up but they can be defeated by the tactics outlined above. Here are some guidelines for fabricating references.
As a general rule the fabricated reference should sound as though it were a real reference. The title should have the proper tone of academic turgidity; likewise the author's names should be plausible. Suppress the urge to satirize titles and names.
The location of the supposed reference should be hard to check. At the same time it should sound as though it had scientific weight. Some good choices here are Russian and Japanese journals. The journals need not exist; it is merely necessary that the titles sound as though they existed. As a matter of course avoid the better known international journals, e.g. Nature and Science. Do not use German, French or English sources - there is too much danger that some wise acre will recognize them as a fabrication. Likewise avoid Indian sources - there are far too many Indian nationals posting to the net (many of them students in the US or Europe) for this to be safe. Italian and Indonesian are fairly safe but do not carry much weight. Eastern European sources are good as long as they aren't Balkan.
As a general rule references with dates should be recent. As a matter of safety your maxim should be that recent references are from obscure sources whereas the less obscure sources should be old. Most libraries do not carry back journals for an extended period of time.
Also obscure universities in the US carry very little weight whereas obscure universities in obscure places carry much weight.
Do not restrict yourself to journal articles. Symposia are an excellent source for fabricated references. In this regard it is useful to remember that much material is presented at symposia that does not appear in the proceedings. You can use real symposia. You can, if you choose, use well known people provided that (a) they are now dead and (b) they attended the symposium in question.
You can also invent conferences. These can be easily made up (pick a topic and a location), easily sound legitimate, and publications from real conferences often do have limited distribution, so they are often not widely available. It is often difficult to check whether the conference even took place at the supposed venue (particularly if a few decades ago), and it will be difficult or impossible (depending upon how it is cited) for people to determine whether or not it is a full paper or only an unsatisfying abstract that isn't worth the trouble (perhaps making it even less likely a reader will check the reference to find out).
It is not advisable to fabricate titles of books; they are well-catalogued these days.
A general guide is that fabricated references should not be prominent in your argument. You can use them to establish essential points but you should not call attention to them.
This is a powerful technique and a major art form. The amateur uses the "quotation out of context" technique. Avoid this. People are likely to catch you at it and you will lose credibility. You can, however, use somebody else's use of out of context quotation - the blame attaches to the person you are quoting.
Large collections of contextless quotes for a wide variety of topics are available. Many are on the WWW which facilitates a quick cut-and-paste. You need not fear using these sources (at worst you will be accused of sloppy scholarship) but, like fabricated references, they should not be prominent.
The best technique is to use a mixture of specific quotation on a single point and general observations which are a plausible misreading of the cited author. In this regard you should be careful when using Eldredge and Gould because the misrepresentations of them are so well known and so frequently used.
The merit of Eldredge and Gould's writings is that they offer so much incredible opportunity for misrepresentation. Even conventional scientists have been occasionally confused about what they have said. The best strategy is to exploit the other author's confusion about Eldredge and Gould, and use their material, rather than Eldredge and Gould's original scientific material.
A useful technique is to base most of your use of an author on their entire body of work. Thus, although you give at one point a specific reference (actual and detailed) to a statement by Mayr, the bulk of your argument uses statements such as "Mayr says" without bothering to note where he says it.
Some authors can be used without references except for a pro forma reference to their most widely known work. Darwin is a very convenient author; almost anything can be attributed as "Darwin says". Likewise Kuhn and Popper, the noted philosophers of science, can be used to support almost anything.
Citing Personal Communications with noted Authorities is a valuable technique but it is almost invariably misused. The usual usage is to claim that the Poobah has confirmed some outrageous claim; this is quite unconvincing. The real function of the Personal Communication is to establish that you and the Authority are peers and are terms of easy communication, that you are a co-eval authority.
To this end you want the Personal Communication to seem to be something that the Great Person might have said. The simplest and most obvious way to do this is to take something that they actually did say, massage it a bit so that it is not an obvious plagarism, and represent it as a Personal Communication.
A particularly useful variant of this is the Personal Communication that disagrees with you on a minor point which you defend. Thus:
Mayr  objects that geographic isolation is a de facto requirement for speciation; however recent work by Slewspinski  shows that this often is not the case. Ernst Mayr, personal communication
 Slewspinski, Variation in Certain Cold Water Molluscs in the Kuriles, Bio. Bulletin, v17, p33.
As a general rule it is not advisable to actually fabricate numbers. What one does instead is to pull them out of context. With only a slight bit of ingenuity you can find numbers to support the point you want to make simply by ignoring the qualifying context. Do not, however, use the fancier techniques from How to lie with statistics. Too many people have read the book and get a perverse pleasure out of exposing the gamier tricks.
For example, you could cite data consisting of 11 samples, and discuss the spread in the data and how discordant that range is, without mentioning that 10 out of 11 data points are completely concordant, putting the explanation for the discordant 11th data point in a footnote. This works well enough that it has been used in publications.
I wish to thank Andrew MacRae who has provided many useful and valuable suggestions.
This page was last updated July 1, 1998.