Reading Science in College

Via Bookslut, there's an article in the Chronicle of Higher Education about whether reading is really important:

Is it always a good thing to read an entire book? When I was a graduate student, it dawned on me that I often had the most intelligent things to say about books I'd only half- or quarter-read. I was surprised by my observation -- it didn't seem to make sense. But it just seemed to work out that professors preferred my insightful and trenchant comments on, say, the first part of Tristram Shandy than on the whole wandering thing.

In that way, a little knowledge can be a practical thing.

(I can't swear that that link won't disappear behind a paywall in the near future, so look quickly...)

Though they're talking about very different things, this made me think of Timothy Burke's essay on How to Read in College:

The first thing you should know about reading in college is that it bears little or no resemblance to the sort of reading you do for pleasure, or for your own edification.

Professors assign more than you can possibly read in any normal fashion.

We know it, at least most of us do.You have to make strategic decisions about what to read and how to read it. You're reading for particular reasons: to get background on important issues, to illuminate some of the central issues in a single session of one course, to raise questions for discussion. That calls for a certain kind of smash-and-grab approach to reading.You can't afford to dilly-dally and stop to smell the lilies.

The rest of the article is a set of tips on how to effectively and efficiently extract the key points from a sample of academic writing, with an illustrative example from the field of history. The first time I read it, though, I was surprised at how applicable it was to the sciences.

Now, the nature of my classes is such that I don't tend to assign a great deal of reading in the usual sense. There's usually a textbook, and students can sort of follow along (only not really, because I tend to diverge from the text quite a bit), but that's not at all the same thing as reading the work product of professional academics.

In thinking about it, though, I realized that Burke's article is really an excellent description of the way I approach many scientific papers. It also nails the problem that students have when I give them journal articles to read.

It's extremely rare for me to sit down and read every word of a paper from a journal. I've done it a few times, when I needed to be really sure about something, but in general, unless I'm an author or a referee, I skim through the paper looking for key words to find whatever I'm looking for.

This process is simultaneously harder and easier than Burke's historical example. It's harder because the length constraints for the top physics journals require a density of prose that can make skimming difficult. Thorny technical problems can frequently be hidden in a single compact sentence at the end of an otherwise innocuous paragraph, and you can get into a lot of trouble by missing those details.

It's easier because most papers have figures or equations. When I'm looking over a paper, I'm usually after the key results, either because I need a number to feed into a calculation or a lecture, or because I want to be able to give a good description of it in a bog post. In most of physics, the key data are presented graphically, and you can usually find the critical results by leafing through and looking at the figures. There's usually a capsule description of the results in the caption, and if you need more information, you can scan through the text to find where the relevant figure is discussed.

Another factor making life easier is the presence of a useful abstract. Some articles in the humanities and social sciences have abstract-like summary paragraphs at the beginning, but I rarely find them all that useful. The important thing is the development of the argument, which usually can't be boiled down to a single up-front paragraph. In science, however, the important result is frequently a single number, which is often reported in the abstract. There have been a number of times when I got everything I really needed from a literature search out of the abstract on INSPEC.

(This includes at least two cases in which I subsequently got the articles from inter-library loan, only to discover that they were in another language, and the abstract had been translated into English before being posted. One of those was in French, which I could sort of puzzle out, but the other was in a Russian journal, and wasn't even in a ahcracter set I could recognize.)

Much of the rest of the process is the same, though. You can save yourself a lot of reading by knowing a few things about the basic structure of a scientific paper. The first few paragraphs are almost always a survey of previous work, to place the new results in context, and prove that the authors have done their background reading. If you already know the field, this can safely be skipped. If you don't already know the field, this may be the only part you need to read.

Shortly after that, there's usually a paragraph that starts with a phrase like "In this paper, we show..." which usually gives a compact summary of the important points. They're usually listed in the order in which they appear, so if you only care about the third, this can be a helpful hint about how far to skip ahead to start looking for the reelvant bits.

Other "signpost" phrases are phrases like "The core of the experiment is..." or "In a second experiment..." (these are taken from the microwave photon paper I wrote about last week). Those indicate the start of the description of a new measurement, and can be important reading.

The other crucial structural feature to know is that there's usually a paragraph shortly before the end that sums up the important findings, which is another good place to look for what you need to know. If there are important technical notes or caveats, they'll usually be hinted at here, which will tell you whether you need to dig through the technical details in the main body of the article.

The main trap that students fall into is usually the introductory historical material. This is often presented in an extremely schematic fashion, and is sometimes little more than a list of citations. I know that, and tend to just skip lightly over those sections (unless I'm really looking for historical background), but students will soemtimes get hung up in trying to puzzle out the meaning of every little reference, which is often pointless (he says, having done this a few times as a student...).

Of course, given that it's a rare physics paper that runs ten pages, skimming isn't nearly as important for the sanity of a physics student as a history major. I'm not assigning hundreds of pages at a time, and while it might take as much time to completely work through a four-page physics paper as it does to skim a hundred-page history monograph, that's not necessarily a Bad Thing.

It was striking, though, to see things that I do in my professional life as a scientist being so accurately captured by advice aimed at undergraduate humanities majors.

Returning to the original Chronicle article... Actually, I don't really have much to say about that, other than that I love the idea of a 100-level English class called "The First Half of Long Novels." Really, all it did was remind me of the Burke piece (which includes a disclaimer noting that you shouldn't actually skim works of literature), which I read ages ago but hadn't commented on... That's blogdom for you.

More like this

The Chronicle of Higher Education has an article about online literacy this week (time-limited link, look quickly), and I'm sure you'll be shocked to learn that the author is pessimistic. The article cites distressing findings from new research: In the eye-tracking test, only one in six subjects…
My course this term is on time and timekeeping, but is also intended as a general "research methods" class. This was conceived by people in the humanities, where the idea of generic research methods makes a lot more sense than in the sciences (where there's a lot more specialization by subfield),…
Over at, Jo Walton is surprised that people skim over boring bits of novels. While she explicitly excludes non-fiction from her discussion, this immediately made me think of Timothy Burke's How to Read in College, which offers tips to prospective humanities and social science majors on how…
The Mad Biologist points to and agrees with a post by Jonathan Eisen with the dramatic title "Why I Am Ashamed to Have a Paper in Science. Eisen's gripe is mostly about Science not being Open Access, but he throws in a complaint about length restrictions, which is what the Mad Biologist latches on…

A math professor of mine once told our class that "you haven't read a [math] book until you've worked all the problems."

Of course I have this physics book buying habit, which is better now that I don't have access to an academic bookstore. But most March Meetings I come back with a handful of books that, sadly, I still haven't "had time" to look through come the next March Meeting.

Reading the introductions to papers is, for me, often the most enlightening part. I think this is one area where we could relax standards of plagarism. I know that you're supposed to pretend that each publication represents new, interesting, and original work, but in reality much science that's both worth doing and worth publishing is incremental. Couldn't we just agree on one best opening paragraph about the context and importance of the general line of inquiry and share it, instead of finding new ways to list the same potential applications and same prior work?

You'll be happy to know that I didn't read this whole entry. I just skimmed to get to the interesting parts.

With papers it also seems to depend how far out of your field it is. I had to read a bunch of geology papers this quarter, and some of them were impossible to understand simply because I didn't know the jargon.

Also, I've found that reading the textbooks for science classes is immensely useful. I never used to, but found that it supplemented the lectures and helped me remember the material better.


Also, it's not a rare physics paper that reaches 10 pages. It's a rare physics paper in many of the subfields. JGRs, echhhhh. It's clear that back in the 30s Ferrarro had published his thesis or something close to when I cracked open the third paper that was about 50 pages long. My god, and all that before digital computers to run the numbers with...

For most of last year just about every paper I read was abstract -> methods. Occasionally I'd actually glance at the data table, but not often.

Yes, I noticed I was doing this recently when refereeing for a conference. I would skim over the paper, trying to get all the important points, then write a one-paragraph summary of it. If I couldn't summarize it, I knew I hadn't read it well enough. If, during the summarization, questions arose, I would go back to certain sections for more details.

Also, after skimming, I tried to outline how I would go about duplicating the experiment. If I couldn't do that, I knew I hadn't read some parts carefully enough (or there just wasn't enough detail).

If you've read enough papers, you know the usual structure. It's not really hard to figure out which points are necessary to understand the paper and which are tangential details.

I should also add that I was kind of depressed when I realized I was skimming like that. If I'm not reading every word of a paper, then probably no one is reading every word of my papers!

The sociologist Harry Collins wrote a paper on how carefully people actually read physics papers. It turns out that physicists have little time so they only read a paper truly carefully if they have social reasons for believing it to be important. That would be information like the names or affiliations of the authors and that sort of thing, or being told by their advisor. Collins' book on gravitation waves was fascinating and also covers this subject.

Because of this, the papers of amateur physicists are routinely ignored. I've run into the same problem. My solution was to remove all the high brow Clifford / operator algebra mathematics and reduce the results to a simple freshman level trig that anyone can understand at a glance. You then mail the calculation in to someone who is already pushing towards that sort of thing. See hep-ph/0605074 for an example.