Reading a Scientific Paper
GUIDELINES FOR READING A RESEARCH MANUSCRIPT|
For critical analysis of this particular paper that you are reading
I generally approach critiquing a manuscript in three stages.
- My key guideline for reading a paper critically for results (as when I review a paper for a journal) is to take notes as I read. First, what do they claim in the abstract. Write it down in point form (there shouldn't be more than 4 to 6 key points).
- Then skim the introduction. Is there anything that is likely to be pertinent? You don't have to know all the history but they might mention something here that you will need to refer to later.
- Then skim the materials and methods checking mainly the titles of the sections and subsections. What techniques are they describing? Link what you see to what you yourself have done or read about before. You don't need to read all the nitty gritty but know what they've written so you can refer back easily.
- Then read the results and flag the figures. Can you see their results in their figures? Does what they say make sense to you already (without going into the details of their procedures)? Do the results you can see tie with what they claim in the abstract?
If you can see that their results are true and back up what they claim in the abstract, then, if you have reason, go more into detail in their procedures, introduction, and conclusion.
Write down what they claim based on the abstract and leave it to stew for a bit.
Then see what they did (results and figures and maybe techniques if I needed that to help me understand what they did)
Then go through the whole thing critically commenting on writing and fine points.
For a procedure that might apply to your work
My guideline for reading a paper when you want to know if what the procedure was and what it gave them would be to skim the abstract for what they claim, but only make a note of the pertinent part that used that particular technique. Then check the results and especially the figures and see if you can see that what they claim is true. Then go into the text of the results and the discussion (especially if this is a techniques-type paper) and then read the relevant parts of the procedure carefully. I suggest this route because there is no point going through the details of a procedure if it didn't do what they say it did. (The proper procedure might do it, but they aren't likely to have written the procedure properly if they can't get the results and claims to agree ....)
For background to an area of research we are doing
- If it's mainly the technique you are interested in, then see above for procedure.
If it's history you are interested in and want an overview of the field, then only read the introduction and the abstract. The introduction often gives you enough history of an area, the abstract should tell you what they did/found that moves the area forward.
If it's history you are interested in and want to know what these particular people did in this paper, then don't sweat the procedures at all and look at abstract, intro, results, discussion. If the paper is significant, then we will have to look more closely to see if their results are feasible given what they say their techniques are, etc., but that comes later and only if the paper is significant.
If it's to see if their results match their abstract, then proceed as Procedure above. We shouldn't cite a paper based solely on the results or discussion given in an abstract.
If it's to see if we should cite a paper, start with the abstract. Is this paper likely to be relevant to the work we want to publish. Was their work important for our particular work? Does it enhance the value of our work? Does our work have more meaning in the context of their published work? Generally, journals do NOT want a comprehensive bibliography. Aiming for 20 to 40 citations seems reasonable to me for most journals. Probably about 10 to 15 citations will arise from the techniques. So, that's not many other articles that we ought to cite. That doesn't mean that we only need to be aware of about 20 to 30 in our particular little area of research though. That's one reason why reading papers for background sometimes is important.
copyright 1999 by Elizabeth Willott
Last update July 25, 2000