How to read a scientific paper


If you’ve been wondering how the pros do it, here’s a guide to dissecting a science paper.

Don’t be intimidated: it’s a description for how to really take every detail of the paper apart, and it’s a rough outline of what I do before talking about a paper on the blog. But it’s also a little bit of overkill for most papers. I read a lot of papers, and I can’t possibly analyze them as thoroughly as that article prescribes, and I take shortcuts — often, the methods are the most boring part, and I’ll just skim over them rather than doing the thorough diagramming recommended. I’ll go back and cover them thoroughly if I find other parts of the paper provocative, though.

The other course I’m teaching this term is an independent writing course, though, in which the students have to produce a well-researched term paper. I’ll have to send them a note telling them to read this article now.

Comments

  1. natashatasha says

    I really wish we were taught how to read papers in undergrad. Thanks for the useful resource!

  2. Matt G says

    A handy tool, but I hate seeing the word “proof” in the context of science. This isn’t mathematics! Or theology!! The word “evidence” is more appropriate.

  3. Bicarbonate (formerly Elizabeth Hamilton) says

    I usually read title, then key words, look at number of authors and affiliations, first line or two of abstract, jump to conclusion, go back to abstract, then begin methods, then introduction, then methods thoroughly, perouse discussion. Think about it. come back later and read more or less in he order this guy suggests but the methods section is really where the meat and/or bullshit can be found.

  4. chigau (違う) says

    As an undergrad, I only ever read the abstract and conclusions.
    I was a straight C student.

  5. David Marjanović says

    All this extroverted writing and drawing… my head spins.

    I’ve been doing a lecture, in-class project, and takehome project on this for a few years now in my freshman colloquium: http://www.geol.umd.edu/sgc/lectures/paperanatomy.html

    I like that; it covers largely different points though.

    …And in the 2nd-to-last paragraph, you have “nearly all over”, where by “over” you actually mean “other”. Took me a while to find out. :-)

  6. george gonzalez says

    Well, that is a fine guide, but is only helpful to a small subset of the folks out there.

    You see the critical piece, that can’t be taught, is to have the smarts to be able to discern reasonable-sounding plattitudes and equivocation and fuzzy statistical data and techniques from real solid logical thinking and real solid data and analysis.

    And if you have that bit of skeptical moxie, you probably don’t need this guide.

    If you don’t have that, having the guide is only marginally helpful. You’re probably intellectually lazy, and will not follow up on other reviews of the scientific paper.

    It’s sort of a chicken and instructions-on how-to lay-an-egg kind of quandary.

  7. says

    In my experience, people have very different approaches to how they read papers, and no method is right or wrong just as long as you synthesize all the information. Brains work differently in terms of how they process it.

    To me it makes no sense to read the Abstract last. The first thing I want to know is, “is this paper worth reading?”, and I can’t know that if I don’t read the Abstract. The title alone is not sufficient. The Introduction can usually be skipped and returned to later if it’s a subject you’re highly familiar with, but it’s likely to contain useful references. I could go on but I’d just be describing my own approach, and my point is that it’s different for everyone.

    The way the author has described the process makes it sound more like doing homework — something you do purely as an exercise in skill development rather than as a means of obtaining useful information. I guess that’s okay the first few times you read a scientific paper, but when you have a stack of dozens of them on your desk, it’s a lot of effort spent for little gain.

  8. bortedwards says

    Plotting out diagrams and mind-maps is all very well, and something I find useful when really tearing a paper to pieces, but I have yet to find an adequate solution to doing this electronically. There is no physical way I can print out and annotate even a subset of the papers I (should) read, let alone then index and search them in the future, so all my papers are electronic (‘Papers’ is great software for filing). The compromise being that scribbling in the margins, on the back, drawing diagrams, doodling, underlining etc is very agricultural in electronic form. iAnnotate is not half bad, but still freehand scribble is seldom legible, slower than thought-speed, and forget nuanced/shaded diagrams or visual cues in pencil pressure etc.
    Maybe I’m asking too much, maybe I need to suck it up and set up a more prescribed but robust note taking system, or maybe I just wasn’t born with a silver stylus in my mouth and prefer outmoded methods.
    Or maybe every time I annotate a hard copy paper I should scan it to PDF… ;)

  9. ChasCPeterson says

    scribbling in the margins, on the back, drawing diagrams, doodling, underlining etc is very agricultural in electronic form.

    hm. Have you tried a stylus instead of a parsnip?

  10. ChasCPeterson says

    In my experience, people have very different approaches to how they read papers, and no method is right or wrong just as long as you synthesize all the information. Brains work differently in terms of how they process it.

    yep. This right here^.
    To me, this is one of the real strengths of the (superficially constraining) standard paper format–because I already know which kinds of information belong in which sections, I can read papers in different ways depending on what I’m looking for from them.

    But for those who are trying to teach people to read scientific literature, the author’s process of “writing to read” (i.e. writing down a paraphrase of the title, summarizing the hypothesis in one’s own words, etc.) is very effective in my experience.

  11. ChasCPeterson says

    Tom Holtz: if you’re accepting typo reports, I have another:

    An abstract: a short (typically 1-2 paragraph) long that states…

  12. babanani says

    PZ,

    If you are able to share notes or links related to your independent writing course, I can promise you that there is at least a small audience that would be very interested in seeing them. Yes, I am asking for you to work for free, but my Mom told me “Don’t be backwards about being forwards.”

    Thanks!

  13. says

    We (American Society of Plant Biologists) have a program that provides high school teachers with access to our journals. As part of this program, we put together two resources. One is a general guide about how to read a scientific paper, which includes sections on peer review and ethics, and the other is a guided walk through a paper published in Plant Physiology.
    The links are here:
    http://journalaccess.aspb.org/
    http://journalaccess.aspb.org/ReadaSciPaper
    http://journalaccess.aspb.org/CaseStudy

  14. carlie says

    Bard College did a workshop several years ago (oh lord, it’s been almost a decade) on “writing to read scientific texts”, and I use a modified version of that format in my nonmajors classes. It’s one of those “read each section and try to summarize it in a single sentence” kinds of things, but the most important part is understanding that you can skip most of it and still get the main concept of the paper. The second most important part is understanding that skipping hard words is ok too. What I see most in students, especially poorer readers in general, is a fixation on reading every.single.word.in.order., and it’s like trying to swim through molasses. Tedious, and as soon as they find a word they don’t know they get stuck. Schools really do all readers a disservice by not teaching how to skim.

  15. blgmnts says


    The abstract is that dense first paragraph at the very beginning of a paper. In fact, that’s often the only part of a paper that many non-scientists read when they’re trying to build a scientific argument. (This is a terrible practice—don’t do it.).

    Unfortunately, the abstract often is the only part that people outside of academia can access due to pay-walling.

  16. carlie says

    I usually tell my students (nonmajors) not to read the abstract, because that’s the part under a word count limit, so that’s where all the shortcut jargon goes and is therefore the hardest part to understand (the sticks nix hick pix effect). So yes, that being the only access point is a real problem.

  17. David Marjanović says

    To me it makes no sense to read the Abstract last. The first thing I want to know is, “is this paper worth reading?”, and I can’t know that if I don’t read the Abstract.

    The context was that you already know the paper is important to read because it’s in your narrow subfield. In that case, not wanting to be influenced by the flashiest, showiest conclusions makes sense: you only want to know what exactly the paper says and how well supported the conclusions are.

    that’s the part under a word count limit, so that’s where all the shortcut jargon goes

    It’s also the part potential reviewers usually read before deciding whether to accept or decline reviewing, and no doubt the part editors read first; it’s therefore a bit prone to “look, this paper is awesome enough for your picky journal”.