Advice to read papers well

[Logic]

Anyone have any advice to read papers quickly while still retaining the details? I'm mostly talking about scientific papers but this kind of advice can probably be extended to any field. 

[AP]

I recommend Raul Pacheco's blog on this: http://www.raulpacheco.org/resources/reading-strategies/ He provides a myriad of reading strategies. 

[TakeruK]

Wow, that blog link is really interesting and I think I can learn a lot from that! I can share how I read scientific papers but I don't think I have such a detailed system.

It does depend a lot on what I want to get out of the paper because that dictates how much time/energy I spend on the paper. Since you asked about reading papers "quickly", I'll talk about the three methods I use for the lowest level of reading.

Level 0, for "keeping up" reading: The goal of this reading is to ensure I know about the latest papers. Every day, the pre-print server I subscribe to sends me a list of paper titles and abstracts for my subfield. This is how I mostly "keep up". The other way I "keep up" is to follow social media for my subfield---there are several facebook groups for various areas of interest to me and papers are discussed there. I also stay connected with my colleagues and see what papers they find interesting.

When I do this type of reading, I will read the paper title and the abstract. If I remain interested, I might also skim the introduction since that's how people put their work in context with the field. Once you are familiar with the subject area, skimming for key words and citations will tell you what area of the field they are building on. The main goal of the Introduction reading is to figure out what important question they are addressing and why I should care. Then, I skip right to the conclusion and see what their results are. I do all of this reading/skimming without taking notes. Usually over my morning coffee. I try to spend about 5 minutes per paper.

At this point, I decide whether or not to spend more time on the article. If I find the motivation compelling and the results significant, I will put the paper in my Mendeley library for further reading (see below) at a later time (sometimes right away if I'm really excited about it). Otherwise, I move onto the next one.

Level 1, for "cataloguing" papers. The goal is now to take good summary notes on these papers so that I am able to find more details when I need them in the future. I may never need them in the future. I download the PDF and import it to my Mendeley library. I assign it a unique ID (usually AuthorLastNameYear plus additional letters for extra papers in the same year). I have some Mendeley tags that I use to organize papers by topics (like Gmail labels) that I also assign. I also have some special tags like "citeThisinPaperX" so that I don't forget to cite the paper when I get around to writing it. After adding this meta-data, I read the abstract, intro and conclusions again.

This time, I will highlight key words. The goal is that if I am looking for details about X in the future, I want to be able to quickly flip through this PDF on Mendeley and have these important words pop out to me. I will also read over the methods section to ensure I know what they are doing. Often, this allows me to classify the paper further (using tags if appropriate). I'll also skim the rest of the paper, particularly the discussion to see their interpretation and any caveats. I use the "Notes" field of Mendeley to write a very short summary of the paper. Here, I also mention anything interesting that comes up or if I have any concerns, e.g. a new method, an unorthodox method, interesting interpretation, problematic assumptions, etc. I think this type of reading takes me about 30 minutes to do per paper.

Level 2, for "understanding" a paper. This is a more careful reading. I actually read every word this time instead of just skimming. I use more highlighting. I also now use the sticky notes annotation tool in Mendeley. One big use of the sticky notes is to make sure I can understand the authors' flow of logic in each section. After each section, if I found the logic hard to follow, I try to figure it out and then write a sticky note that summarizes what I thought they were trying to say in that area. I also add sticky notes for Methods. Since these notes are searchable, I try to use standardized terms so that I can find related papers easily. For the discussion and results sections, I use sticky notes to summarize each and every one of their individual findings (whereas in the above sections, I focus more on the big picture results). I add my own commentary to their interpretations where appropriate. The goal is to annotate the paper enough that when I view the paper in Mendeley, I can click the notes tab and see just the list of my sticky notes. I should be able to reconstruct the main argument from the sticky notes without having to read the paper again.  If I have done the "Level 0" and "Level 1" readings before this, then this more in-depth reading usually takes 1-2 hours, depending on the paper length and whether I already know a lot about the field. Sometimes it can take up to a half-day if I'm trying to read way outside of my area. I sometimes also print out a paper to read at this depth because it's a little easier for me to read things in print and when I want to be able to read it while away from my desk. I try not to print too much because it wastes paper and I will have to spend time importing my annotations into Mendeley.

I usually do "Level 0" and "Level 1" every day (or at least try to). The "Level 2" reading happens maybe once per week (a little bit more recently because there's lots of cool stuff). This is the depth I read to when I want to be able to speak about a paper at a Journal Club or something informal. Deeper reading levels, for me, include reading related papers , comparing results across papers, and making even more notes. I don't usually seek out papers to read to this depth level. Instead, I only do it when I need to, for example, when starting a new area of research and needing to know the foundational papers, or when I really need a question answered.

[Hope.for.the.best]

19 hours ago, Logic said:

Anyone have any advice to read papers quickly while still retaining the details? I'm mostly talking about scientific papers but this kind of advice can probably be extended to any field. 

I start with the abstract. If I am interested in the paper, then I will read the last paragraph of the introduction (containing major findings) and the conclusion. If I need more information, I will read the figure legends and relevant parts of the methods. Only if I am very interested in the study do I read the whole article. I usually start writing about the papers as soon as I find them useful. 

The strategy that works for everyone is different. Try any strategies that may help you read faster, and jot down notes. Then read the whole paper and see if your notes cover the details. This will help you to find out what works the best for you. 

[TakeruK]

3 hours ago, Hope.for.the.best said:

I start with the abstract. If I am interested in the paper, then I will read the last paragraph of the introduction (containing major findings) and the conclusion. If I need more information, I will read the figure legends and relevant parts of the methods. Only if I am very interested in the study do I read the whole article. I usually start writing about the papers as soon as I find them useful.

Oops I forgot to mention figures! Good advice here. For papers in my field, a figure really is worth at least 1000 words and I try to study the figures and reconstruct the story from the series of figures and section titles. This is helpful in the stages where I don't read all of the text.

In doing this, you also quickly figure out what types of figures are effective at telling a story and what kind of figures add more confusion than information! Good to digest this and incorporate into your own figures and papers.

 

[Hope.for.the.best]

7 hours ago, TakeruK said:

Oops I forgot to mention figures! Good advice here. For papers in my field, a figure really is worth at least 1000 words and I try to study the figures and reconstruct the story from the series of figures and section titles. This is helpful in the stages where I don't read all of the text.

In doing this, you also quickly figure out what types of figures are effective at telling a story and what kind of figures add more confusion than information! Good to digest this and incorporate into your own figures and papers.

 

That's the advice from my tutor. She actually got students to just read the figure legends and understand the papers. It was hard for everyone at first, but this strategy is very handy to understand a large number of papers within a short time. 

Another strategy is to shortlist your reading. Find some recent reviews on your topic. This way, you don't need to go through a lot of papers to understand the literature. Should you need to do any reading, it will be those cited in the reviews. 

[AP]

On 12/12/2017 at 2:50 PM, TakeruK said:

Wow, that blog link is really interesting and I think I can learn a lot from that! I can share how I read scientific papers but I don't think I have such a detailed system.

I know! I don't implement his suggestions verbatim, but I came up with a system that works for me. He has pretty good tips!  

[ClassicsCandidate]

On 12/18/2017 at 12:29 PM, AP said:

I know! I don't implement his suggestions verbatim, but I came up with a system that works for me. He has pretty good tips!  

I agree that that's a great resource, thanks for sharing it!