Abstract. This is a brief (usually one paragraph) summary of the whole paper, including the problem, the method for solving it (when not obvious), the results, and the conclusions suggested or drawn. Do not write the abstract as a hasty afterthought. Look at it as a real exercise in cramming the most information in one paragraph. The reader should not have to read any of the rest of the paper in order to understand the abstract fully. Its purpose is to allow the reader to decide whether to read the paper or not. A reader who does not want to read the paper should be able to read the abstract instead. When you write an abstract, remember Strunk & White's admonition, ``Omit needless words.''
Introduction. Tell the reader what the problem is, what question you will try to answer, and why it is important. It might be important for practical reasons or for theoretical (or methodological) reasons having to do with the development of a scholarly discipline. Don't neglect either type of reason.
If the problem is a very basic one, you may state the problem first and then review what has already been found out about it. If the problem is one that grows out of past literature, review the history of how it arose. But do not forget to mention the basic issues behind the research tradition in question, the practical or theoretical concerns that inspired it. (Sometimes there don't seem to be any. In this case, you have probably chosen the wrong topic.)
Your literature review should be appropriate to the kind of paper you are writing. If it is a thesis, you should strive for completeness, both in reviewing all the relevant literature and in making the main arguments clear to a reader who is unfamiliar with that literature. For a course paper or journal article, it is sufficient to review the main papers that are directly relevant. Again, you should assume that your reader has not read them, but you need not go into detail. You should review only those points that are relevant to the arguments you will make. Do not say that ``X found Y'' or ``demonstrated'' if X's conclusions don't follow from X's results. You can use words like ``X claimed to show that Y'' or ``suggested that'' when you are not sure. If you see a flaw, you can add, ``However ...''. Try to avoid expressions like ``Unfortunately, Smith and Jones neglected to examine [precisely what you are examining].'' It might have been unfortunate for them or for the field, but it is fortunate for you, and everyone knows it.
The introduction should lead up to, and conclude with, a statement of how you intend to approach your question and why your approach is an improvement on past efforts (or why it is worth undertaking even if it isn't). This is essentially what is new about your approach, your particular contribution. It need not be anything great. Something like ``applying X's method to test Y's theory'' is good enough.
Method. This section gives the details of how you went about your project. It is usually divided into subsections such as subjects, materials, and procedure. These subheadings are standard ones, but they are not always appropriate, and other subheadings are acceptable. The point of subheadings is that the reader may want to skip this section entirely and return to it later in the paper. The subheadings should make it easy to find relevant details.
Results. This is a summary of what you actually found. It is not a dump of your unanalyzed data, nor merely a report of whether your statistical tests were significant, but somewhere in between. It should contain whatever summary statistics will help readers see for themselves what happened, such as means and standard deviations of various conditions, and raw correlations, when these are relevant. It should also contain the results of statistical tests. Make sure to do and report just those tests that are relevant to the question that inspired your project. If you must include your raw data (and sometimes there is good reason to do this), put them in an appendix. (Notice that the word ``data'' is a plural noun meaning, roughly, facts.)
Graphs, charts, and tables are often useful in this section (and elsewhere, but less often). They should be labeled consecutively either as Figures or Tables, depending on whether a typesetter could be expected to set them, (yes for tables, no for figures), e.g., Figure 1, Figure 2, Table 1, etc. Each one should have a caption explaining clearly what it is, if possible without relying on anything in the text. (Figure captions are on a separate sheet so that the typesetter can set them, but for course papers, this is not necessary.) The text should tell the reader when to look at the figures and tables (``As shown in Figure 1 ...''), and it should point out the important points, but it should not simply repeat in writing what they say.
Figures and tables are supposed to go at the end of the paper, but this is for the benefit of the typesetter. Most professors (except nitpickers) prefer the tables and figures close to where they are needed.
Discussion. It is a good idea to begin the discussion with a summary of the results, for the benefit of the reader who wants to skip the results section (and to remind the reader who didn't skip it but got interrupted by a phone call and forgot it).
In the rest of this section, you return to your original question and tell the reader what your results have to say about it (``The results indicate that ...'') and what they do not have to say (``However, the results are inconclusive concerning ...'' or ``do not speak to the question of''). In each case, tell why. Try to think of objections that someone might make to the conclusions that you draw (whether the objections are correct or not) and either answer them or qualify your conclusions to take them into account (``Of course, these conclusions assume that the subjects were telling the truth, which might not be the case''). You may also say why you think the objections are weak even if they are possible (``On the other hand, there was no reason for the subjects to lie''). Your task here is not to do a sales pitch for some idea but rather to help the reader understand exactly what can and cannot be concluded.
The discussion section may be combined with the results. The advantage of this is that it puts the results in the context of the issues that generate them. The disadvantage is that the flow of the discussion gets interrupted with a lot of statistics, etc.
The discussion section is also the place to say anything else you want to say that does not go anywhere else. You may reflect on the implications of your results, or your methods, or whatever, for other issues that were not the main point of the paper. You can talk about how your project should have been done, and why. Or you can make a more general argument, for which your results are only a part.
Note that some of these things may be quite creative, but none of them amounts to simply reporting ``your own ideas'' without support. You should report your own ideas -- when you can support them with arguments and reply to potential arguments against them. If you can't do this, maybe your ideas need to be changed. You can also make suggestions that might be true, labeled as such, but then try to state the alternative too.
It is often a good idea to end the paper with a general statement of main message. More generally, one type of well-constructed paper will reveal its main ideas to a reader who actually reads only the first and last paragraph and the first and last sentence of every intervening paragraph, and this principle applies especially to the discussion section by itself.
References. This is a list of the articles cited. Usually, articles are mentioned in the text by author and date, e.g., Baron (1988), and the references at the end are listed alphabetically by author. Each discipline and each journal has its own conventions about references. These usually insure uniformity, but they don't even help the typesetter. The important thing is that you give the reader what she needs to find the articles you have cited. For journals, both the volume and the year are usually needed as well as the page numbers, because mistakes are common. If you really want to do it ``right'' pick a journal and imitate the style.
Footnotes. Sometimes you want to say something that isn't quite necessary. This is the time to use a footnote. If you can get away without using them, it saves the reader's eyes. But sometimes it's hard to resist making rather extensive, but rather tangential remarks. These go in footnotes, not the text. The really eager reader will read them. Others will not.
You may assume your reader is intelligent, but he reads only your paper, not your mind. Therefore, when you use any terms that are not obvious, you must make sure to define them so as to remove any relevant ambiguity. A good way to do this is with both an abstract definition and an example: ``I use the word 'dyslexia' to refer to seriously impaired reading despite normal instruction, vision, hearing, and language ability. By this definition, a retardate could be called dyslexic if his reading is far behind his speech.'' There are practically no ``standard definitions'' in fields like psychology, so you must choose your terms and your definitions of them so as to capture what you want to say without flying the face of other people's definitions of the same terms. Because terms are so important in academic discourse, do not use more than one term for the same idea (no matter what you learned in 9th grade English).
Style. Academic writing may seem pompous and convoluted to you. A lot of it is, but the best is not. Do not use words just because they sound academic (especially when you aren't sure what they really mean).
The major rule of syntax is this: write so that a reader could parse your sentences -- that is, figure out what modifies what, what is the object of what, and so on -- without understanding what they mean. The syntax should help the reader figure out the meaning; the reader should not need the meaning to decipher the syntax. For example, put ``only'' just before what it modifies (``Smith suggested that only men are susceptible to this effect,'' not ``Smith only suggested that men ...'') to avoid ambiguity of syntax, even if you think the meaning is clear from context. Of course, pay attention to correct usage as well. Make sure you know the rules for using commas; many people do not. (Strunk and White, "The elements of style," provide an excellent review of the roles, as well as many fine suggestions for elegance as well as clarity.)
When you read, pay attention to the different ways that people indicate the relationship of their work to the truth. Words such as indicate, demonstrate, prove (not used outside of mathematics), test (a hypothesis), hypothesize, suggest, assert, question, claim, conclude, argue, discover, define, and assume do have very specific meanings in academic discourse.
I've been correcting graduate student papers and editing journal articles for more than twenty-five years. I see the same errors of writing over and over. Here are some to avoid:
Vacant Lead Sentences. The first sentences of each section, and the first sentences of each paragraph as well, are the most important sentences. They should state, in plain English, your main points. Then the details can follow.
Right:Qualifiers and Caveats. Don't squander the opportunity to write forcefully by beginning with secondary points and caveats. They belong in the body of the paragraph or section, but not as openers.
Results. Cognitive therapy prevented relapse better than drug therapy. Drug therapy did better than no therapy at all. Analysis of covariance...
Results. We performed four analyses of covariance on our data, first transforming them to z scores. We then did paired comparisons using a Bonferroni correction...
Distinguish between strong and weak statements. Good scientific writing uses qualifiers and caveats sparingly. Qualifiers apply to marginal results, arguable statements, speculations, and potential artifacts. They do not apply to strong findings, well-confirmed statements, or bedrock theory. "Seem", "appear", "indicate", "may", "suggest" and the like are meaningful verbs. They are not to be used reflexively.
Right:Big words and long sentences. Most readers are busy. Many readers are lazy. Many readers just scan. Help these readers by using short sentences and plain words. Whenever a big word tempts you, look hard for a plain word. Whenever a long sentence tempts you, find a way to break it up. The big word and the long sentence must increase accuracy a lot to make up for impeding reading.
Because volume was barely significant, water-deprivation may lower hunger. Electric shock, however, increased hunger two-fold.
Our findings suggest that electric shock may increase hunger. It also appears that water-deprivation seems to lower hunger.
Wrong:Overwriting. Omit words and ideas that the reader already knows. Overwriting slows the reader down and does not increase accuracy at all.
Thus, by assigning this group to the wait-list condition, treatment effects would not be artificially inflated by including the higher income group with a better prognosis in the initial treatment phase.
Richer people have less depression. So we biassed against our hypothesis by putting more of them in the wait-list control.
Wrong:The Royal "We" and the Passive Voice. Poor writers turn to the awkward passive voice to avoid saying "I did such and such". The first person, used sparingly, is fine. Write forcefully and use the active voice whenever you can.
The wait list control group, when compared to the attention control group, the drug treatment group and the psychotherapy treament group did worse than the attention control group, and much worse than the experimental drug treatment group and the psychotherapy treatment group.
Psychotherapy and drugs did better than attention alone and much better than no treatment.
Right:Citations in the middle. Don't break up sentences with citations. This small increase in accuracy slows the reader to a crawl. If you can manage it, group all your citations at the end of the paragraph.
I propose that animals can learn about noncontingency and, when they do, they become helpless.
It is suggested that animals can learn about noncontingency. When noncontingency is learned by an animal, helplessness results.
Direction of statistical effects. Always state the direction along with its significance.
The interaction between drug and weight was highly significant (F (2,31)=14.56, p<.001).
Small doses of the drug put small rats to sleep right away, while big rats stayed awake even with very large doses (F weightXdose (2,31)=14.56, p<.001).
Comments to firstname.lastname@example.org