How to write a Paper
David GoldbergInstitute of Psychiatry
Course for Young Psychiatrists
Nairobi: 23th March 2007, 10.45
Before anything else….
Consult a statistician – how many patients will I need? [to show a difference if one exists (POWER); and to avoid making false claim (SIGNIFICANCE)]
Are the intended patients representative?
Get yourself a supervisor!
Discuss the work at every stage with him or her.
Ask the statistician
How will the results be processed – what statistical tests are most suitable?
Will you, or will the statistician, carry out the calculations?
(If the latter, the statistician is on the “by line”)
Who WILL the authors be?
- they are all meant to contribute to the paper
- what order will the authors appear?
- who will be first author?
- who will WRITE the first draft? (you!)These things are best decided well beforehand
The Pilot studyMost research needs one! (It may not be possible to decide on the numbers of patients required until one has been done)
Do the procedures work? Are the measures understood by the patients?
See your supervisor – can the fieldwork begin now?
When should you start writing the actual paper?
The first steps should be taken while you are doing the fieldwork – make a careful note of your method, as you will need this when you write up
At this stage, decide on who the authors are going to be; who is going to write the paper; and who will be first author
Processing the results
If you carried out the calculations yourself, take the results to the same statistician, and discuss them.
Then discuss them with your supervisor – is there enough for a paper, or should some additional work be done?
Are the findings important enough to merit publication?
Was your original hypothesis correct?
If YES, excellent – go ahead
If NO, is a negative finding important enough to publish? If it is, go ahead
If not, has your study shown anything else that might be interesting?
…assuming that you’re going ahead:
Which journal is likely to take it?
general or specialist?
local to your country, or international?
what are the publication delays?
how likely is your chosen journal to take it?
The Impact Factor 2004
No. of citations / No. of articles • Nature 32.2• Lancet 21.7• Arch Gen Psychiat 11.2• Am. J Psychiat 7.6• BMJ 7.0• Schiz Bull 6.6• Biol. Psychiat 6.2• J Clin Psych 4.8• Brit J Psych 4.2• Cognit Psychol 4.0• PsRxPsSom 4.0• Sz Research 3.9
• Am J Geriat Psy 3.5• JAmAcadChPsy 3.5• Psychosomat Med 3.4• J Psychiat Res 3.1• Addiction 3.1• Psychol Med 2.9• J Psychosom Res 2.8• Sz Bulletin 2.6• Psych.Serv. 2.3• Acta Psych Scand 2.3• Can. J Psych 2.2• Psychiatry 2.2
Having chosen a journal with a reasonable chance of taking your paper (and whose readers you wish to communicate with!)
Go to the library, leave through a recent issue:
how long are the papers?
What are the instructions to authors?
What has to be submitted?
What are the HEADINGS that they use?
Do the writing in this order:
• Title• Method• Results• Introduction• Discussion• Summary, conclusions, implications• Abstract
TITLE: Importance of “now read on….”
Assertive community treatment versus traditional community care: a randomised controlled trial with comprehensive outcome measures
Assertive community treatment is superior to standard community care
- which is better?
The paper must tell a story
The message must be clear – especially in the Abstract!
The Italians call it the “linea rosa”
Eliminate all that is irrelevant to your message
INTRODUCTION: What the paper is about; end with aim of this study
Don’t start like this:“Many studies have addressed the problem of assertive community treatment (1 – 17) ..…”
This would be better:“Mental health staff working in the community cannot be sure of the advantages to be gained by assertive community treatment. Several uncontrolled studies have reported great advantages (1-3), but better designed studies reveal conflicting results (4-7). Recently, the treatment received by the control group has been the focus of interest…”
What does the reader need to know?
Subjects: who, where and when. Don’t forget sample size – POWER of the study!
Measures: don’t describe well known instruments (reference only), fully describe new ones
Eliminate ALL unnecessary tables
Consider when histograms, graphs or tables should be used
Always set your own tables, including only essential information the reader needs
Do not say the same thing in text and table: text tells story of how you arrived at your conclusions, tables give the data
Do not give absurd numbers of decimal places
First, discuss your new results in the light of the studies quoted in the introduction
Then, admit the limitation of your study
Finally, consider the implications of your results for the mental illness services
Try to end on a “high”
Should it be structured? (usually, “yes”)
The only thing most people will read!
It must tell exactly the same story as the paper
Declaration of Interest
People who helped, but did not contribute to writing the paper.
The funding organisation.
Any commitments you or your co-authors or your employing organisation to sponsors or those contributing financially to your organisation.
KeywordsImportant – may determine who reads paper!
Time to send drafts by e-mail to your supervisor and your co-authors. Everyone must see it!
All your writing can be done on your computer, using mode in Word (Tools; track changes) that allows you to follow your co-authors suggestions
Allow your results to be on computer paper until you have finally decided upon what will be used in the final paper
Expect to do many drafts
Decide what can be left out
Do NOT include irrelevant data, just because you have it!
Always use a spell check; but remember this only helps with certain mis-typed words
If writing in a foreign language, ask someone who is fluent in it to read your final draft
Always aim to sleep on it: you see quite different problems when you re-read in the morning!Finally, print a hard copy and proof-read that as well: your eyes see things on the printed page that they miss on the screen of your computer.
Almost ready, but not quite!
Is your Abstract fully structured?
Are all the references included, and in the proper format?
Have you cited any references which you are no longer quoting (from earlier drafts, usually)
Have you thought of “key words”
Have all the authors signed “copyright release” forms
Send it off, and hope for the best!
Reviewing a Paper
Which Journal has sent it?
If you aren’t familiar with the Editor’s requirements, read the accompanying guidance notes carefully.
Sometimes, authors’ anonymity is respected
Usually, your anonymity is respected
Only do what the Editor asks you to
• Read the summary or abstract• Skim read the whole thing• Separate the tables from the text,
lay them side by side on your desk• Turn on your computer: two
separate fields – minor comments, and major
• Off you go! Go through line by line
Must be in the journal’s format
Can you understand it?
- it should accurately describe the paper that follows
When you have finished reading the whole paper, return to the Abstract
- did it leave important points out?
In many journals, there is a maximum permitted length. Even if there isn’t one, it must not be too long.
Is it all necessary?
Has anything important been left out?
Does it end with a clearly stated “Aim of the present paper”?
MethodIf this is not clearly described, the paper is of little value. Are you quite clear what they did?
Do you know who the subjects were?
Are you clear about what the measures were?
Did the study have adequate power?
Have appropriate statistical tests been carried out?
This is where the tables are needed, as you read.
Are all the tables necessary? (could some of the findings be expressed more easily in the text?)
Could any of the tables be simplified?
DiscussionThis should discuss the new findings in the light of existing knowledge (which should have been in the introduction)
It should mention the limitations of the study, and not gloss over them
It should also cover the implications of the findings
The referencesHave important previous work been cited?
Have unnecessary papers been included?
Are they in the correct format?
Have all references mentioned in the text been included?
- to check this, make a small pencil tick in the reference list against each paper cited in the text
- are any missing?
- are there some that were not cited?
In your “main text”…
Make suggestions that will be sent to the author[s]
Don’t be rude!
In your “subsidiary text”…
If there are grammatical mistakes, typos or stylistic problems, put these in your subsidiary text. They are very useful for the author[s] in revising their paper.
Standing back …..
Thinking about the whole paper, how important are the findings?
Are they new, or just confirming what is already known?
Should the Editor be advised to publish quickly?
Don’t be hyper-critical. Do as you would be done by!
Confidential comments to the author can be made in two ways:
- there is sometimes a special form, usually including an overall grade
- if there isn’t, and you need to make a comment in confidence, write a separate letter