Scientific Writting
Citing information from others' publications is common. However, except in rare instances, these should not be verbatim quotes (ie. copied out). In all cases, the concept or sentence you use must be referenced to the individual who wrote/thought of it. You do not wish to take the chance that you look like your are trying to take credit for someone else's original ideas as one of your own -- plagiarism.

Review articles are very useful but they should be read, digested and synthesized by you rather than directly copied in your writing. Aside from the question of who did the original writing (you or the review article author), you may not want to agree with the bias expressed. Review articles are always written with a bias that you need to identify. Your own publications (papers, essays, exams etc.,) should reflect your own interpretation, and not just reflect someone else's.

Writing the Results section of a poster, paper etc., is a learned skill that you will acquire in the lab, usually after quite a bit of practice. Basically, you want to be sure that the data you choose to show is representative of the whole picture. This is (i) to avoid misleading others and (ii) to protect yourself when others try to repeat your work. There is no problem in selecting a particular experiment (for instance, one out of 4 performed) which clearly demonstrates your findings. Not all experiments work out identically and you shouldn't expect them to. At the same time, you do not want to choose the one experiment which gives much bigger differences between groups than all the others, because it is "nicest", because this would be misleading. Basically, you need to ask yourself if the data you show fairly represents your observations. It's not technical, just honest.

It can be permissible to exclude some experiments from consideration if you can justify the reason for doing so (ie. enough mice died by accident or from bleeding over the course of the experiment that you are concerned about the validity of the conclusion; the positive or negative control failed to work; the viability of the cells was low; the radioisotope was old; the culture was contaminated, etc). These experiments would clearly be misleading, so excluding them is appropriate. Ask yourself if excluding data is going to falsely enhance (or detract from) the "impact" of the data that you do show. This is worked out in discussion with your advisor at the time by looking at all the relevant data.

Often, the best resolution in cases of uncertainty is to add an additional independent experiment to increase your confidence in the conclusions you are drawing.

Do you know why you are using the statistical approach that you are? You needn't be a statistician, but you need to know why you chose the method you did, and the assumptions it is based on (ie. normal distribution for parametric tests such as Student's T tests, usually non-parametric such as Mann-Whitney for highly variable data such as that obtained with human populations). If in any doubt, do not make assumptions. It is best to consult Mary Cheang (789 3730), Senior Systems Analyst and biostatistician in Community Health Sciences.

Other common points regarding writing:

Misspelled reports/theses or overheads are terribly unprofessional. Use a spell checker before showing your document to anyone. Make sure that your figures and tables are clear and easily interpretable.

Discussions should not be a chronological recitation of your data – they are intended to discuss it. This is your opportunity to demonstrate your critical ability to synthesize and interpret scientific work. Check an issue of the Journal of Immunology. In general you will find that there is a brief recitation of the main findings (ie., not a list of observations made but a synthesis of the data) which are intended to (i) highlight the conclusions that have been drawn and (ii) focus the reader's attention on the novelty of the study. This is followed by an attempt to integrate your observations into the literature. This means a demonstration of your critical analysis - extracting the key points of difference / agreement/ weakness, etc., from the literature that you need to buttress your argument. It is not a recitation of "George found it went up 87%, Sally found it went down 19% and Sue didn't even look." That is just a compilation that can be done by a student who is asked to make a list of observations without commenting on them. Particularly since we know that one can find evidence for and against absolutely everything in the immunological literature, you need to make your conclusions – your judgement – clear.

A common problem as one appreciates the complexity of the literature on the topic they study is that some people attempt to cite every possible related study (both pro and con), something that can give the impression that you are arguing with yourself or haven't made up your mind and want to list all possible outcomes to be safe. It is much better to decide on the point you want to make for each paragraph, then state your hypothesis, assemble your evidence (your conclusions/ others data which you may synthesize and refer to or refute with a valid reason) and, if necessary, restate your conclusion at the end. Everyone will not agree, but if your argument is clearly presented, the reader will enjoy discussing the scientific merits of your case rather than spend most of their time trying to figure out what you believe yourself. Brief, simple writing is always better, and more difficult, than covering every single eventuality.

Novelty: This is difficult. You obviously don't want to overstate your case, claiming novelty for something already well known. At the same time, the summary paragraphs in initial drafts of theses are sometimes written to make it appear that you have successfully repeated what others have done and not much new. Is this the impression you wish to convey? It is natural to take comfort in the fact that you find what others do (reducing the probability that a prof will ask an awkward question), but you need to clearly identify what you have done that is novel. Be explicit about what you have discovered/contributed or they'll safely assume that the answer is nothing. Particularly for theses, you know much more about your data and the literature in your area than do the readers. Be explicit about your contributions and don't assume that the reader will intuitively grasp the significance of your work.

Once you have completed a draft that you feel you can be proud of, ask a colleague (post-doc or senior student in the lab) to read it and critique it for you. This will benefit both of you.


GENERAL POINTS FOR THE PREPARTION OF A SCIENTIFIC PAPER OR REPORT

Developed from Instructions to Authors published in Infection and Immunity.

Organization and Format

One of the best ways to improve your writing is to examine carefully good papers that you have read, paying particular attention to their style rather than just reading them for content, as you usually do.


REGULAR PAPER

Regular papers/theses/research reports should include the elements described in this section.

A. Abstract

Usually limit the abstract to one double-spaced page, two pages for theses.

Concisely summarize:

  1. the background - Why did you bother to do the study?
  2. the intent - What did you set out to do?
  3. the results obtained, without presenting extensive experimental detail
  4. heir significance.

B. Introduction

The introduction should supply sufficient background information to allow the reader to understand and evaluate the results of the present study without referring to previous publications on the topic. The introduction should clearly provide the rationale for the study. Choose references carefully to provide the most salient background rather than an exhaustive review of the topic. Often the last paragraph is a brief summary of the major findings of the work. An explicit statement of the hypothesis you will test is an excellent idea. You don't want to give the mistaken impression that you were just working along, hoping something would jump out at you.

An important point is to ensure that your introduction is focused and does not ramble. In theses where this can be a common problem, a good trick is to use subheadings for each subsection so that the reader can see the focus of each section in advance.

C. Materials and Methods

The Materials and Methods section should include sufficient technical information to allow the experiments to be repeated. When centrifugation conditions are critical, give enough information to enable another investigator to repeat the procedure: make of centrifuge, model of rotor, temperature, time at maximum speed, and centrifugal force (X g rather than revolutions per minute). For commonly used materials and methods (eg., media and protein determinations), a simple reference is sufficient. If several alternative methods are commonly used, it is helpful to identify the method briefly as well as to cite the references. For example, it is preferable to state "cells were broken by ultrasonic treatment as previously described (9)", rather than " cells were broken as previously described (9)". The reader should be allowed to assess the method without constant reference to previous publications.

Assays for cytokines, Ab etc., should provide explicit detection limits and state clearly the specific activity of the standard used so that comparisons can be made with other literature.

Your subjects, animal or human, should be clearly identified in terms of their characteristics, number of experiments, number per group, etc.

D. Results

In the Results section, begin most paragraphs with the rationale or objective of that section. Follow with the specific experimental approach taken. It is an often good idea to highlight the reasons for your choice of this method in preference to others. Do not assume that the reader will know why you made the choice of strategy that you did.

What are the main findings that you want to focus their attention on? Identify the one or two major points that are important in your figure/table. Do not bury your audience with extensive presentation of everything that can be seen in your experiment.

Reserve extensive interpretation of the results for the Discussion section. However, you should include a single sentence identifying what you believe to be the relevance of this experiment and how it connects with the plan of your research. Do no assume others will necessarily derive your implied conclusion. Be explicit.

Present the results as concisely as possible in one of the following: text, table(s) or figure(s). Don't make filler figures to beef up a paper. It is better to have a short report with one very important figure than nine figures of fluff that obscure your one important finding, and make the reader skim your paper so quickly they miss it. Avoid extensive use of graphs to present data that might be more concisely presented in the text or tables. Think carefully about which means of presentation will most effectively convey your point. All tabular data should usually be accompanied by standard errors of the means (SEM). The number or replicate determinations (or animals/humans) used for making such calculations must also be included. All statements concerning the significance of the differences observed should be accompanied by probability values given in parentheses. The statistical procedure used should be stated in Materials and Methods.

E. Discussion

The Discussion should begin with a brief one paragraph summary of the novelty/importance of your findings. It should provide an interpretation of the results, not merely a restatement of your observations, in relation to previously published work and to the experimental system at hand. It should not be a reiteration of the introduction. It is in the discussion in particular that you have the opportunity and obligation to critically analyse your own and others' findings, and demonstrate your ability to do this well. What are the important differences between your own and previously published work? How does it affect our view of the field? You must make clear your contribution. If you just exactly repeated someone else's work, no one is interested. Therefore, the discussion should explicitly state what is different or novel and important about your work. It is ok, sometimes important, to criticize specific aspects of others' work, but his needs to be done carefully and with grace. You should also be sure of your facts. Re-read the particular papers very carefully to be sure you have not misinterpreted.

There will usually be many small discrepancies between your work and that of others. It is usually best to ignore these, unless they are critical or your discussion becomes bogged down in trivial details. If in doubt, discuss it with your supervisor.

Your paper or thesis should tell a coherent story that is connected from beginning to end. Some observations will not fit and will "require further research" but after reading your work, your reader should feel able to give a brief statement of what you wanted to do, what you found and why it matters.


Common problems that drive your readers crazy... like your supervisor!

  1. Make sure your subject and verb agree: Conditions are tested, not is tested. As your sentences get longer (and they should not), it becomes easy to forget the subject of the sentence. Identify it and make sure that it agrees. It is very boring for your supervisor to do this for you.
  2. Use a spell checker before giving your report to anyone, including your supervisor!
  3. Keep your sentences short or your point becomes obscure. Resist the temptation to add too many modifying clauses. If you find a sentence becomes too long, break it into two or three. It is better to have two short, clear sentences than one that is long and confusing.
  4. After you have proof-read your document carefully, read it aloud, to an empty room if necessary. This will make you break up the run-on sentences into more manageable ones. Note that where you pause in your reading, a comma should be inserted. One sentence equals one thought. Remember not to add so many modifying clauses that it becomes impossible for the reader to get the point.

Specific Faults

  1. We effect changes and their consequences affect us. We measure effects.
  2. Cells or animals are not primed. They are immunized.
  3. Mice are not cultured (Unless, of course, they have been to the opera lately!) Spleen cells from mice immunized with xyz are cultured.
  4. We don't measure IL-4. We measure IL-4 production, levels or synthesis.

Written by: Dr. Kent T. HayGlass
Last revision: April 30, 2004