How to effectively read scientific papers?

[Fasc]

As a graduate student/researcher, I have to read a lot of scientific articles. I've always been quite terrible at it: reading a 30 pages long article I'm not familiar with can take many hours sometimes (including the time to google unknown terms and results, look through some of the referenced papers, etc.), and in the end I often remember few points from it, as the amount of information is just too big and can be presented rather chaotically. Plus, the language is usually very dry, making it hard to keep the interest and focus - more than once did I catch myself nodding off while reading the same sentence again and again, trying to understand what it means.

I'm curious what your approach to reading scientific papers is, if you do that. Are there any tricks/methods helping make this process easier? So far I've only noticed that taking notes helps a lot, as you force yourself to concentrate, when a pen is in your hand and you are writing - however, the notes get lost, and the amount of effort it takes in case of large/complicated articles is ludicrous.

So, how do you usually read articles? How effective is your approach? How much do you remember from the articles you've read a while ago? How long does it take for you to read one average-sized article?

Take the legal approach and brief them. It is like taking notes but more formalized than that and will seem very very clunky and difficult at first, but trust me you'll improve over time and it'll make what you're reading go faster and smoother. You'll be forced to distill sentences and concepts into their core arguments and will realize that some of what you're reading is actually grade A BS depending upon your own level of knowledge and expertise.

If you're REALLY still struggling, bite the bullet and take some technical writing courses/tutorials or any course that teaches you fundamental mechanics to language. Language is assembled in specific ways and despite all the flowery language and terribly misplaced clauses in a lot of papers, the better you know how it all fits the easier it is to dissect and digest it.

Sadly there is no quick fix, just practice practice practice.

[anyaka21]

It might take a while, but as your reading and taking notes, and if you understand it, how would you rewrite it? Then rewrite it.

[Kraenen]

Go with abstract as a screening process. From there usually skim the intro (last paragraph usually has some hypotheses), scan the methods and results for major keywords and pick up on any fudging of data. Discussion is mostly the same. All around you're trying to quickly pick up on the main idea of each paragraph. Train your eyes to see key words.

Also take brief notes. Just read your pdf and have word open, jot down a very small point and make sure you don't miss critical details. Eventually you'll have a couple paragraphs of summary and you'll know what the jist of the article is.

If it's extremely dense, you'll need to read it a few times, making sure you get a break from it in between.

- - - Updated - - -

Do that, but if you're trying to prove something, make sure you go back and read the methodological defense parts if you intend on including them in your own paper/thesis/whatever. Don't want a huge hole to be opened by completely missing something that defeats the part of the paper you're citing.

True. Method is very important in the later years (year 4+grad+phd) and you might sometimes pick up on some falsified/exaggerated data, even if not you can pick up on some good criticisms of their measurements or operational definitions, etc...

[Underverse]

It'll usually be hard. Even on topics I know a lot about, I end up having to reread things several times to really understand it. Usually I'll just look at figures and then skim the text for details on things I don't understand - unless I'm presenting the paper for some reason.

[Illuminance]

Read the abstract completely. If you don't understand it fully, stop and look up things until you do. The abstract is the cliff notes for the paper.

Then, I usually skip straight to the figures. If you read and understand each figure/figure title/figure legend, you can recreate the "story" of the paper without having to read any of the actual writing. The intro and discussion are simply there for context and perspective. The results just state what's in the figures. This is fairly easy to do once you have some expertise in your field.

If you are just starting out, it's going to be a slog.

[Dsonsion]

I usually read the introductory parts first, skip the method, read the discussion and conclusion and come back to the method having in mind what will be important there, taking the conclusions into account.

[Acidz]

Isn't this one of the first things they teach you at university?

[twiddler]

SQ3R and some method of taking notes. Takes about a night to read shorter papers in which my background is fuzzy and the material is impossible to efficiently abstract, but papers on other topics (e.g. social graphs) have been both great and efficient reads as there is that layer of analogous connection. I don't think it is too different from reading graduate level text in mathematics and CS.

[Fasc]

Isn't this one of the first things they teach you at university?

Not necessarily. Technical writing (and consequently reading) are not requisites of a LOT of majors, to include positions that lead to research. Engineering for instance mandates technical writing as often our presentations and speeches are observed by non-engineers so we have to learn how to speak/write simply, define all our terms, be concise, etc etc. Law school mandates similar reading/writing skills be taught to 1L's so they can slog through 50+ pages of opinions to get to the core arguments, evidence, legal theory, actions, holdings, etc.

May is a physicist. Technical writing may not even be a thing until the graduate program itself and even then, given how the GRE works, it may very well be assumed you have good reading/writing skills upon entering. I actually don't know of many fields that explicitly teach technical writing for research purposes.

[Tennis]

As a graduate student/researcher, I have to read a lot of scientific articles. I've always been quite terrible at it: reading a 30 pages long article I'm not familiar with can take many hours sometimes (including the time to google unknown terms and results, look through some of the referenced papers, etc.), and in the end I often remember few points from it, as the amount of information is just too big and can be presented rather chaotically. Plus, the language is usually very dry, making it hard to keep the interest and focus - more than once did I catch myself nodding off while reading the same sentence again and again, trying to understand what it means.

I'm curious what your approach to reading scientific papers is, if you do that. Are there any tricks/methods helping make this process easier? So far I've only noticed that taking notes helps a lot, as you force yourself to concentrate, when a pen is in your hand and you are writing - however, the notes get lost, and the amount of effort it takes in case of large/complicated articles is ludicrous.

So, how do you usually read articles? How effective is your approach? How much do you remember from the articles you've read a while ago? How long does it take for you to read one average-sized article?

It's normal for it to take hours to read complex material. You just need to be patient.

If you're really in a hurry then skimming and making notes may help. I usually just try to set aside a lot of time and just focus intensely on reading the articles.

[Deleted]

If I need to read say 30 pages of technical stuff I generally skim through it, let it sit in my mind for a few hours while I think about it / do other stuff. Then reread later in a normal speed. I feel the insight i gained through skimming and a bit of thought makes second reading a lot easier compared to just reading it.

[Endus]

Isn't this one of the first things they teach you at university?

In general, it's "taught" through making them do it, and then criticising their grasp of the paper and such. And if they don't glean enough, they get poor feedback, and teach themselves to read more thoroughly.

Professors generally have more important things to do than teach people to read well.

[Deleted]

Sounds like a flaw in the education system.

Isn't there some kind of software to help people screen and select the points of interest and connections then help manipulate and construct whatever they're trying to achieve.

[satimy]

I always look for the data and how it was collected. Bad data collection ruins science

[kail]

For me, I tend to read the abstract/summary first to see if an article, report or paper is relevant for my own assignment. After that, I look at the headers that tend to organize and break up the material. Rarely do I ever read the whole thing. If it's a matter of you not understanding any of the material whatsoever, it may not be relevant to your field or you are simply not at the required level.

[Spectral]

So, not an improvement in speed of reading papers, but for things I actually care about (or more importantly, am reviewing for publication), I outright skip the text of the results and dive into the figures first. If I need the introduction to understand the context, I read that, but above all, I want to dig into the data and see what it looks like - I need just enough context to interpret it coherently and then I'll start forming my thoughts on what it says. This helps me avoid bias when reading the author's narrative of what their data says. I know both from reading and writing that story-telling has a big impact on how someone sees data.

When I need to get through content more quickly, I succumb to the same abstract skimming as everyone else. Then I'll look for the punch-line figure - most papers have one or two figures that really tell the most important piece of the story and everything else is just supportive or exploratory, at least in the fields that I'm familiar with and qualified to read meaningfully.

Discussions I only read if I'm interested on much elaboration. There's usually not much there there.

Methods I read in a relatively cursory fashion unless I'm reviewing the paper or need to use something from them. The majority of methodology in the work I'm familiar with is pretty routine and the specifics of sampling and such are usually spelled out well enough in the results section. I don't usually need to know exactly how many microliters of antibody you used for your ELISA.

As for retention, I was never very good at remembering individual papers. I'm good at remembering the total body of literature, but the specific author and year never really stuck. I'm really impressed by PIs that have a databank of that stuff in their head.

[Garnier Fructis]

Don't get stuck on particular items that you don't understand, sometimes it becomes clear what they mean later on.

[Spectral]

Professors generally have more important things to do than teach people to read well.

I cannot imagine a more important skill for anyone in any intellectual pursuit to have than reading well.

[Celista]

I cannot imagine a more important skill for anyone in any intellectual pursuit to have than reading well.

I am amazed when I meet highly educated individuals who flatly state that they hate reading. I've met sciences PhDs, physicians, attorneys who all claimed to hate reading (whether academic journals or otherwise).

[ro9ue]

Cliffnotes.