How to effectively read scientific papers?

[May90]

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?

[Kujako]

With a snifter of scotch, in a wood paneled room lit only by a fireplace, with jazz on the record player. The corduroy jacket (with elbow patches) is optional of course. The beard and glasses are not.

[May90]

With a snifter of scotch, in a wood paneled room lit only by a fireplace, with jazz on the record player. The corduroy jacket (with elbow patches) is optional of course. The beard and glasses are not.

I'm afraid I've failed then... I don't wear glasses.

[A dot Ham]

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?

You are asking the least scholarly group of people ever. The majority of people here are what is called academically dishonest. They purposely spread misinformation and will always cherry pick from published articles to suit whatever side of the argument they choose to be on that day.

[PC2]

What are you reading that is not interesting?

I think you should be reading neuroscience, computer science, or physics.

[unbound]

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 notes as you read the articles. I find it is easier if I take a break every 15 to 20 minutes as well if it is particularly complex or a subject I'm not very familiar with.

As far as memory goes, I can remember the gist of them quite well years later, but I'll have to return to my notes or the actual article for details which, from my perspective, is perfectly fine.

[May90]

You are asking the least scholarly group of people ever. The majority of people here are what is called academically dishonest. They purposely spread misinformation and will always cherry pick from published articles to suit whatever side of the argument they choose to be on that day.

There are quite a few graduate students and postdocs here, as I've noticed. Sure, it probably would make sense to ask this question on PhysicsForums, for example, but I'm curious what people here have to say on this as well!

What are you reading that is not interesting?

I think you should be reading neuroscience, computer science, or physics.

What kind of science papers you reading?

I mostly read articles related to star and galaxy evolution. Don't get me wrong, I love the field, and the topics articles are about are very interesting. Sadly, often an article is so densely packed with technical information that it gets very tiring long before I've gotten to the part describing actual results. And if I hop to the results section right away, skipping everything else, then I won't understand well where the results came from and what assumptions they emerged from.

EDIT: what kind of program you in? genuine curiosity

U Notre Dame, Department of Physics. We have a very strong faculty on astrophysics, and pretty good funding!

[Endus]

A useful skill to pick up, if you want to make the effort, is speed-reading. I can go through a 30 page paper in less than 5 minutes. The key to it is that you're not trying to read every word, you're trying to skim and skip. It's basically the next step up in reading; you start out by reading each letter, and sounding them out, but eventually you read entire words at once. With speed-reading, you're reading entire sentence clauses in the same kind of speed. Your comprehension goes down, though. I can retain about 60% after a run-through.

That's not a bad thing, though. When you've got 50+ sources to go through, you want to get a quick handle on what they contain, as a first-run, and then you can settle down and dig into the few you picked out as most relevant in greater depth. Also, like Darkdeiii mentioned, taking notes in that second reading is helpful, either with hardcopy and jotting stuff in the margins and wielding a highlighter like a sword, or with a word document alongside what you're reading online.

Also, especially with science papers, be sure you know what you're reading, and why; if you're interested in their conclusions primarily, skip the methodological defense section. You can go back to that when you decide you need to fact-check their methods. The reverse, obviously, is true if you're more interested in comparing methodologies between studies.

[NED funded]

It depends on the topic I'm reading. For things that aren't very technical i just use a pen to mark important stuff or parts that I feel are wrong so I can check later. For very technical stuff I read the whole thing once, then check my textbooks, later I read the whole thing again. Think for around 10 minutes of the implications and then read it one final time to get a better understanding of the methodology ( which imo is the most difficult part)

As for length it varies depending on the article. However it takes me around 3 hours for basic stuff and an entire day for the technical stuff. On that note fuck sticky wages and everything directly related to it.

[Celista]

I usually take notes, but I only bother if I think there's anything significant I can take away from the articles. Really it's good to keep in mind your own research as you do this, it gives you a focus point that helps determine whether something is worth note-taking/brain storage for.

Skimming the article first, reading the headings, subheadings, proposed hypothesis (usually made prominent in the article in some fashion) + results section of paper is a good first read. Then going back again, taking notes, for a second read.

For articles where 1-2 sentences are only relevant to whatever it is I'm reading scientific papers for, I keep a file where I can copy+ paste with citation.

For articles that are more important/relevant, hypothesis, results, whether the hypothesis was supported, flaws in methodology (sometimes there are a lot of them). Theoretical articles, I try to find out whether the theory was empirically tested in a later study.

As to how long it takes to read...it varies quite a bit. Length of article, clarity in writing, presence of unnecessary jargon, how familiar I am with a particular area of research etc all are relevant.

[Annoying]

I had to take notes to do effective research for my thesis. Grabbed a stack of computation books (the ones with grids, not lines -- I prefer them for drawing examples and such), title of the paper you're reading at the top, note down anything interesting, write underlined notes about things you don't understand. Try to link things together that are similar but different areas. Really depends on the type of research you're doing and what you're doing it for.

Also:

Also, especially with science papers, be sure you know what you're reading, and why; if you're interested in their conclusions primarily, skip the methodological defense section. You can go back to that when you decide you need to fact-check their methods. The reverse, obviously, is true if you're more interested in comparing methodologies between studies.

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.

[Celista]

With a snifter of scotch, in a wood paneled room lit only by a fireplace, with jazz on the record player. The corduroy jacket (with elbow patches) is optional of course. The beard and glasses are not.

Beard optional for women. Pipe is not!

[Tonyman64]

I start with the abstract and then go through the figures and conclusion to get major points and see if it's worth reading more thoroughly.
After that it's what other posters are saying; read carefully and take notes. I've also found that things go faster and I comprehend better as I get more familiar with the area of research, but that's obviously not going to be helpful all the time. In terms of saving time the only thing I've found helpful is more quickly identifying when a paper isn't useful to me.

As for how long it takes me, a well written article in a familiar field can take ~30 minutes to an hour. A poorly written article in a less familiar field might take all day if I need to thoroughly understand it. I usually remember papers well enough to go and look at them if I think they might be relevant to something else. I generally don't remember all the details, but I don't think it's really reasonable to expect to.

[Endus]

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.

Yeah, I'm not really saying "don't bother with some sections", I'm saying "prioritize what you read, and be looking for specific information when you do". Papers aren't stories; they're infodumps. You don't have to (and often shouldn't) read them from start to finish, necessarily. That doesn't mean you shouldn't come BACK to sections you've skipped, but you should go at it like someone mining for information, not someone reading a story.

[Keeponrage]

The short answer would be - iterative classification.

Don't try to dig into details of every article you come across, especially if you are investigating an area you are not quite familiar with yet. Firstly, try to find a decent survey - in many cases this classification is already done there. Collect articles and read up their abstracts and introductions and do a first-pass classification based on them. Classification rules are yours and depend on your current understanding of the problem, i.e. they will evolve together with the level of your comprehension. By the end of the first pass you will be familiar with basic terms, sub-problems definitions and general approaches to their solutions.

Using this knowledge do second pass on articles u've collected already + additional articles you'll find useful, when you are armed with the knowledge from the first pass. But now you go a bit deeper in details. By the end of this pass you should extract extra classification major features and re-classify all articles u've collected once again.

Repeat this procedure by going as deep into each article as it allows you to extract new and even unique features. Stop when the amount of classes equals the amount of articles. At this moment you can clearly state what is unique in each article, since it is you who is defining these classes.

Write a survey by yourself! When you put your classification features on paper and explain their usefulness it will help a lot to systematize your understanding of this problem better and even find gaps you potentially can fill - this will be your scientific input!

Pick articles that match you research problem the most - and study them in detail and maybe model some stuff on PC (probably using some free stuff from web) - when you "touch" a problem in practice you'll level up you understanding of the problem by a huge amount.

[dextersmith]

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.

Sounds similar to software manuals and such. I like to reduce drowsiness with a nap before you start reading if you can. Food and taking notes relieves boredom, while music and tv distracts. Try not to take breaks cause it's hard to regain focus.

When taking notes from your articles, also write your thoughts, interpretations, and how you would employ the findings. Your investment will foster interest and focus, while letting you get more out of the experience. About notes getting lost, just don't lose them, you're not a child.

[Skulltaker]

There is no foolproof system applicable for everyone. My roommate in college read the stuff once and knew it by heart, but still never made it past the 2nd semester. Doesn't matter if you can memorize all muscles and nerves in the human body in half an hour if you vomit when you open one up. But I digress.

What helped me were notes. I usually had two notepads - the red and the blue. The blue one worked as an index - everything that had sunk in went in there, with a remark as to where to find it. The red one got everything I had trouble with, to reread.

[Grimbold21]

This is interesting.

I've just started my masters and we have to deal with 15-40 pages articles so far and there's one thing i've noticed with most of them. I understand that with a scientific paper you need to substantiate your case/claim, but goddamn i can get the main points of an article for under 5 pages. Im always like "yea, i get it, can we move on?"

As for method of reading, im a slow reader and not a fan of speed reading claims. I might take a whole afternoon or more to read 4 or 5 articles, mainly cause its good to take breaks and i take notes aswell.

[PC2]

U Notre Dame, Department of Physics. We have a very strong faculty on astrophysics, and pretty good funding!

Astrophysics is too boring for you. You need to go micro, to the planck scale.

[dextersmith]

There are quite a few graduate students and postdocs here, as I've noticed. Sure, it probably would make sense to ask this question on PhysicsForums, for example, but I'm curious what people here have to say on this as well!

Don't take my advice then, I dropped out of school in grade 6 and never looked back!