Reading (and understanding) Scientific Articles

So you’ve started using PubTrawlr and you’re ready to dive deeper. You follow some links to the primary source, open then up….and it’s gobbledygook.

Take a deep breath. This a very common experience. Articles are not easy to read. This isn’t the latest from Stephen King. There is likely to be a ton of jargon. The writing can be unclear, probably not that much better than this post.

With that in mind, this post takes you through some of the basic steps and questions to keep in mind while you work through the article. We update it frequently in order to have it reflect the most up-to-date experience and reflections of the PubTrawlr community.

The Big Picture

Before you do anything else, page through the whole article. These aren’t mystery novels; you don’t have to worry about spoilers. You want to get a good sense of the flow of the article from the opening premises, to the complexity of the methods, to the visualizations.

Really, we can boil any (scientific) article down to four central questions:

  1. What’s the central question?
  2. How do they try to answer that question?
  3. What did they find?
  4. What does it mean?

Let’s take each of these in turn.

What’s the central question? Any article should be trying to move science forward. An article should be trying to say something that has not been said before. Generally, this comes about by posing a novel question. Sometimes, though, the questions look backwards to find out, “what have we found so far?” This is most common in research synthesis or metanalysis that try aggregate findings over time.

The article should be boiled down to one central concept. A good place to start that is by looking at the abstract. Abstracts lay out the story of the article, and are the BEST place to begin. Take this article in the image below.

Right off the bat, we can see the what and and the why of this article. This helps to justify the implied question, “[For women of Haitian descent], What are the perceived benefits of participation in evangelical communities?”

How do they try to answer that question? Actually, let’s back up first. Think about the central question, how would you try to answer it? What type of ways would be helpful in getting closer to this answer? There’s always lots of different ways to approach a scientific problem. Authors have the responsibility of saying why they picked a certain strategy and why they believed it would get them closer to the answer.

You’re usually find this information in the methods. Here, the author will talk about how they measures what they’re interesting, how they collected their data (and from whom), and what types of statistical methods they used to make sense of the data.

What did they find? It’s helpful to skim the article first and spending a minute or two on the figures and tables. There are a few types of figures. There are the theoretical figures that show relationships between underlying constructs. Take the below figure, which is something that the Dawn Chorus Group uses to think through how they develop implementation supports.

In this figure, we see a few box showing “things” that we are concerned about. We also see arrows moving between those, which show the relationships (some going only one way, some going two ways.) We also have a circle that surrounds all the factors, with one of the arrows going to the outside of the circle.

Then, there are the figure that demonstrate the relationships between data. Take a look at the axes. These are the lines running at the bottom and side of the graph. They show values and the relationships between them.

This figure so the “net sentiment” over the middle half of 2020. We can see wild swings, with positive and negative values jumping up and down before settling into a general, flat trend. Looking at this, what can you say about the mood of COVID-19 stories over time?

What does it mean? Before you even get to this point, you should spend some time yourself trying to determine what the big take home messages are. I had a professor who once claimed that he never read conclusion sections. That’s a bit extreme.

The conclusion is where the authors spend some time discussion about what the implications of their findings. If the study was particular well done, then you should have a pretty good idea about what should happen next.

A word of caution. Publication Bias

It’s hard to get something published (usually, there are some pretty terrible journals our there). One problem is that it is hard to publish something that doesn’t work, or a study that didn’t find anything. The articles that you encounter are more likely to be positive, meaning that they have found *Something*. Keep this in mind when you are reading. In almost all cases, it is likely that the results are skewed to the positive. Don’t forget to read with a critical eye!

Summing it all up

Don’t worry if you don’t understanding everything. Unless you’re of the experts in the field, you will find it hard to dig through articles. However, by trying to keep those four questions in your head, you can get past the 60% threshold of understanding and use that to further you exploration. Happy reading!

I am a gigantic boat!

Last updated December 8, 2021