Am I reading enough of the scientific literature? Should I read for breadth or depth?

Question

Ever since starting graduate school I've tried to make scientific reading a part of my daily ritual; I track pages read using Beeminder, and the graph doesn't lie. It keeps me honest.

I aim to closely read and summarize 5 pages per day and skim a few other abstracts besides that. I spend about half my time looking at data and figures which doesn't contribute to my daily "page count." When I'm reading about a new topic these five pages can take several hours, but on topics I have more background in five pages might only take an hour per day.

I guess since everybody defines "read" in a different way it's hard to get an objective answer about how much reading is enough. How much people read seems like a bit of a sensitive topic among real-life colleagues because everyone has a bit of anxiety that they aren't reading enough. But for those further along in their academic path, I'd like to hear how you approached the literature early in your graduate school career and what you think is a sufficient amount.

I guess this all distills down into two main topics:

When deciding what to read each day, should I focus on depth or breadth?

Is five pages of close reading per day enough? I know it doesn't sound like much, but it takes significant mental energy to meet that goal. And consistently reading 5 pages per day adds up to a lot over time.

Edited to add: I mostly read about petrology, volcanology, structural geology, and tectonics if that makes a difference. By "page" I mean "page of text" so if I'm reading a structural geology paper with lots of maps and figures I discount for those and a "ten page" paper becomes a 5 page paper for my purposes.

Answer 1

My experience is almost exclusively with mathematics papers, and applies little or not at all to other fields.

Much of eykanal's post applies to math as well, but one big difference is that math papers are much more varied in their structure, not having an actual experiment to tie them together. A good paper will generally explain its organization in the introduction, however.

One point worth emphasizing is that reading a paper from front to back, trying to understand everything at each step, is usually inefficient. The most common instance is that a paper often starts with definitions which may be hard to make sense of without understanding the theorems they're used in. It's generally more effective to skim the paper several times, trying to understand more and more with each pass.

Relatedly, you'll eventually pick up the skill of picking out the most interesting ideas from a paper without reading the whole thing. Early on, though, it's probably better to read things carefully; it's very easy to fool yourself into thinking you've understood something.

As to your main question, about breadth versus depth, your first priority has to be depth, because that's what you'll ultimately need to be able to do your own research and get your degree. But if you're learning enough to do that, you want as much breadth as possible. It actually gets harder and harder to learn completely new things as you get on in your career, even when there may be direct benefits to your research to do doing so. Laying the foundations of a broad understanding of your field while in graduate school will pay off later.

Answer 2

This post refers to research in the STEM fields, and may not be applicable to other research topics.

One of my biggest epiphanies in research came when I learned how to read a paper. Reading scientific publications is completely different from reading literature or news. At the beginning of your research career, you can expect to spend a full day (if not more) reading through a single 8-page paper. Some tips follow:

• Most papers are divided into "Intro", "Methods", "Results", "Discussion". These are roughly broken down as follows:

  • Intro - Read this for background. There will be nothing "new" in this section. You will find it very useful to read the intro section to as many papers as you can get your hands on. While you do this, you will become fairly depressed that so much research has already been done, and you will wonder what you can do to add to the field. Speak with your advisor, he has many good ideas.
  • Methods - This will take you a VERY long time to read initially, because they go into ridiculous detail. They do this so that you, the researcher reading and interested in replicating their results, can do so. If you don't understand everything here at first, don't worry. NOTE: If you finish reading the methods section and still want to know how something worked, email the author! This is research; the guy who wrote the paper is likely another grad student/postdoc like you. He'd love to hear from you.

  • Results - This is the meat of the paper. Read this very carefully to find out what they found. Between this section and the methods section you will determine what went right, what went wrong, what is new, and what they should have done that they didn't that you can now research and publish and become a superstar.

    When you cite a paper, you will be citing from this section. If you find yourself citing a paper based on something in the Intro, you're just citing another citation.

  • Discussion - This is the author's thoughts on what the results mean. Take note of this; the author is using his or her expertise to interpret the results. If you disagree with something he or she says here, and you can back up your findings, more power to you.

• Most accomplished researchers don't actually read papers; they just read figures. A good paper will be completely in the figures. (This is particularly true in some biological sciences fields, less so elsewhere.)

• Take notes on the papers you read. Keep those notes. My method was to keep my notes in a 3-ring binder, put a little post-it tab with the author's name, and then put the paper in there as well with the notes, so each "tab" is my notes and the paper. You will read hundreds of papers during your academic career. You will want to remember what you've read.
• This is a very arduous process, and the learning curve is steep. Don't be discouraged! Reading papers is a skill, and the more you read the more proficient at understanding them you'll become.

Answer 3

This is about my experience in computer engineering

I found that reading for breadth was the more important approach. The area of research I was interested in was pretty fluffly and ill-defined (I thought I could make a difference by organizing it better), so that many relevant articles were categorized in totally different areas. This meant I had to have a hummingbird approach: flittering around, but drilling down when I found an important vein of data. I also kept a journal where I'd put a citation and a very brief summary of the article, so that I could come back and say "I think I read something about this last September" and then go look in an older journal. Today, I'd have my own wiki at home to keep track of this. I used to have my own "library" of PDFs that I got through university access, but that removable hard drive was stolen.

Is five pages of close reading per day enough?

If you can stay consistently at 5 pages (or 1 article) every day, you will end up far ahead of other people who study only in spurts.

Answer 4

At the beginning read anything and everything (and take notes). You should always be reading something and writing something. The hardest part for most students in the sciences to get past is embracing the unknown.

You will probably feel the need to understand everything, right from the start. Unless you are exceptional, you probably will have to read the important papers several times. You are looking to develop a broad-scale understanding of your field. To know where your research fits in, you have to develop an understanding of where your field in. This takes time.

As time goes on, you'll pick and choose more carefully the papers you read closely. Often you can get the idea from just the abstract. If it sounds promising, then read on.

Answer 5

I'll chime in with a quite different opinion (or maybe a related opinion phrased in a different way), which grew out of advising/supervising PhD students and post-docs: reading for depth is a job requirement, but reading for breadth is what will make you stand out.

As a PhD student, you are required to read in depth the papers that directly pertain to your particular subfield. A PhD is the process of becoming an expert in your discipline, and you cannot do that without mastering the minute details of it, which you will only learn by reading in depth the papers published (and attending conferences, asking questions, etc.).

However, though becoming an expert is what gets you your PhD, if you want to continue further in research (whether academic or R&D), you will need to be able to show a quick understanding of new problems, to make connections between concepts in various areas of research, and propose creative solutions to the problems you have identified. This requires a casual knowledge of a large variety of fields, which will be only acquired by reading a large breadth of topics.

Answer 6

I agree with Henry about breadth vs. depth. You'll ultimately be judged on depth, so that has to be your first priority. However, breadth is quite valuable too. Many breakthroughs have come by applying standard techniques from one area to a new area.

The $.02 I want to add is that not all reading is created equal. Particularly when you're learning a new topic, well-written exposition is invaluable (in large part because it's so rare). As you progress, you'll develop a better intuition for what's worth reading. But when you're early in your career, I strongly encourage you to ask your adviser (or more senior students) which papers and books you should be reading. Personally, I've slogged through many manuscripts mired in myopia before encountering enlightened, engaging exposition. ...and that has made all the difference.

Answer 7

I personally think one should read both for breadth and depth. Read all sort of literature around your field, and a few outside your field. One can only come up with good ideas having a good general knowledge of science. When it comes to your own research area papers should be read carefully and critically to understand what is being done, how it is being done and if the interpretations and methods made and used really show that. As my PhD supervisor used to say, read atleast two papers a day even if you are busy with experiments. This gets easier as you go. For me in the first year it used to take a lot to time to read a paper. Towards the end, I was looking at the abstract, results/figures...if needed methods, and where confused check the discussion quickly to see how the authors explained their results. After a while, you rarely need to read the intro in your own field unless you want a refresher.

Answer 8

I have some points to mention on this:

1. Read papers relevant to your research: I mean, you will know slowly which conferences publish results in your interest area and which do you find relevant, so choose papers from top conferences or journals because those set the benchmarks

2. Datasets in paper convey a lot Browse in the result section of the papers and you will notice that the datasets on which the algorithm are tested should suit your requirement. For example: I am working on outdoor dataset and I see a paper showing results from indoor dataset, there are 99% chances of me dropping that paper.

3. Use abstract as filter I think this point is self explanatory

4. Scheduling your reading - difficult There may be a week where you will end up reading a lot of papers and there will be times where reading even one paper will not be possible.

Answer 9

This is about my experience in applied atmospheric physics

As others have said, it is often reading for depth that is more effective. Making sure that you understand the key concepts and connections made between the main concept and the subsidiary topics within a paper. Usually, I read the Abstract and Conclusion to garner the main points of what the paper sought to discover, what method they used (established, modified or new) and what were the overall results.

In many of the papers that I have read (and written), the method is often in stages corresponding to specific results in the results and discussion section (they are often merged in my field) - I tend to read the method, derivations therein and the results of each stage - which means flipping between each section. All the time taking notes.

At times (not very often), the method stage makes reference to another paper, so that paper is retrieved and worked through in the same way.

I have found this process to be quite quick, as each stage comprises a short paragraph and equations in the method; a corresponding sentence in the results, often with a table or graph and a sentence or two of explanation.