Special Topics: Reading Scientific Manuscripts

Photo by  Roman Mager on  Unsplash

_“It doesn’t matter how beautiful your theory is, it doesn’t matter how smart you are. If it doesn’t agree with experiment, it’s wrong.” _

- Richard Fenyman

Like it or not, publishing a scientific article remains the primary productivity metric for academic researchers. “ Publish or perish” is no joke (and the name of a great citation tracker). Even outside of academia, the ability to quickly synthesize and extract information from these papers will provide necessary context for your work in the field. Numerous scholars have opined on the best way to learn this process . But they are likely coming from a related technical field. We are humanities folks. We can ponder the racial theory of Vonnegut one day, and the feminist critiques of Dickinson another. I would argue, however, less domain specific knowledge is needed to “get” a scientific article than you think. Regardless of field, I propose the following 5 steps to drastically improve your reading time and knowledge obtained from each manuscript:

1) Read backwards.

Similar to a journalist in a major periodical, scientists have, generally, one “big idea” they wish to convey to their audience. All that changes are the methods. Instead of extended metaphors and wordplay, we deal in variables, study designs, and causal pathways. Because the arc of the story remains consistent, I recommend that beginners start with the conclusion of a scientific article. This will ensure two things: (1) you have a “big picture” of the story the authors are telling, and (2) you can more easily predict how they conducted their experiment(s). This is like skipping ahead to the punchline of a joke—sure, it ruins the suspense, but it gives you the main ideas all at once [re-work this metaphor]. 

Before moving to our next step, quickly glance at any sources of funding mentioned after the conclusion. There are heated ongoing debates whether this belongs in science (I remain neutral as we do not have enough data yet to determine its influence and each source is different), but it can help you understand why the work is being done. 

2) Stop. Take two minutes to guess what comes next.

I know, this is weird. You wouldn’t pick up The Sun Also Rises, read “Isn’t it pretty to think so,” and then stop to think about how Hemingway set the stage. But this isn’t Hemingway. We can expect certain things to be said from science papers. If this is literally your first scientific article, take a stab! You will be surprised how well your guess matches the methods presented. Write this guess down in the margin and move to step three. Making it this far means you have committed to the broader scientific process of hypothesize, test, feedback and repeat - a process growing in popularity among study hacking circles.

3) Read the introduction deeply.

This advice also runs counter to conventional wisdom which tells you to more or less skim the introduction . This is wise for experienced scientists who only need to extract some context before plunging into the methods and results. You need it all. Be looking for how they discuss certain phenomenon—how do they phrase what they know? Is a relationship already established (“we know X influences Y”)? Is it unclear (“it is likely that X” or “emerging evidence suggests that Y”)? Context for the narrative will come. You need to meticulously pick out their linguistic tools. As you finish this close reading, circle concepts or phenomena of interest or confusion (for later articles!).

4) Skim the methods and results.

The methods and results are the cash cow of most scientific papers[cite]. Here I propose another tactic that departs from conventional wisdom. However, for our purposes, beginners need context and process more than getting bogged down in the details. Trying to read this section for comprehension early on is like reading senior theses in your freshman seminar. Instead,  circle or draw arrows to concepts you would like to eventually learn. This accomplishes two things: it primes you for actually learning them more deeply later, and it gives you a starting place to read more papers in the field. If you feel like you have to cover some of the results, focus on whatever they present in the first paragraph of the results section. This will typically offer a statistical summary of their findings supporting the “big idea."

5) Read the discussion like a play.

If you’ve taken an acting class or seen a live play (given you are trained in the humanities, this is a fair assumption) you know that almost all plays follow a formula. They contain characters, setting, dialogue and a resolution divided into acts. Science trades in variables, samples, comparison and discussion to deliver its biggest ideas. Framing a scientific manuscript as a play provides a solid framework to maintain comprehension in the face of difficult language or math.

For example, when I was first reading articles on survival analysis, I imagined the hazard ratio (change in survival over change in time) as an old-timey, sepia-toned scene where someone was stitching together tons of snapshot pictures for an overall movie of a dance. If you took one picture it would give you information on what the dancers were doing at the precise instant they were photographed, conditioned that they were still dancing. Swap out dancers for patients in a cardiothoracic surgery clinic, and the concept holds true.

Being able to extract useful information from scientific manuscripts touches many careers in and outside of science. Most importantly, if done in what I am calling the storytelling framework, it will connect you to our most ancient trades and practices as a species. Try it out and let me know how it goes in the comments below! What worked and what didn’t?

COVID-19 Relief Worker

Epidemiologist, Pre-Medical Student