I like researching things. Not in a write-a-research-paper-for-school-with-a-long-works-cited-of-things-I-didn’t-actually-read kind of way. But more in a procrastinating, sidetracked, or found-something-super-duper-interesting-and/or-personally-relevant kind of way.
My “researching” used to take me down a lot of Wikipedia rabbit holes. And while I’m not discounting Wikipedia as a useful tool (sorry high school teachers), I have upped my research game a tad. Since taking a crap-ton of science and psychology classes in college, I’ve turned to more academic sources. (Even as I typed that, I imagined “academic” being said in a really post British accent.)
Reading a Scientific Paper
You may have noticed that in my last post I linked to several scholarly articles from journals focusing on biology, physiology, and psychology. I also mentioned the National Institutes of Health (NIH) and the database PubMed. If you want to be a discerning consumer of science and knowledge, sometimes you have to go straight to the source. Because while news outlets and science blogs can be great ways to broadcast research results to the general public, the writers at these sources have their own filter through which they see the world. (Yours truly included!)
On the flipside, there is still a chance of you reading an original piece of research and having your own interpretations tainted by your own biases, generalizations, and misunderstandings. It’s hard to be completely objective, but understanding how to read stuff written by actual scientists doing actual experiments is a good way to start.
First off, it is important to understand the difference between a primary scientific paper, in which the author(s)/researcher(s) carried out the study(s) themselves, and a literature review or meta-study, in which the author(s) read and analyzed many studies that they did not perform themselves.
I am going to focus on primary papers here, as I feel that those are a.) the most essential and b.) the most challenging to read!
Understand the Pieces
Normally any primary research paper in the sciences (biology, chemistry, genetics, psychology, etc.) will be made up of the same basic pieces. The first step to reading a scientific paper is to understand what those pieces are there for.
- The abstract is your map to the rest of the paper. In college courses where we were taught to write in the style of scientific journals, the professors had us include 1-2 sentence summaries of each of the subsequent sections in the abstract. That means that an abstract gives you a brief overview of the context of the study (the introduction), the methodology of the study (the methods), the major findings (the results), and the conclusions that the authors/researchers drew from this study (the discussion).
- Reading the abstract is a great way to tell if you are interested in reading the rest of the paper, but it is not a good idea to read the abstract, assume that you are an expert on the topic, and then spend Thanksgiving lecturing everyone on what the mashed potatoes are going to do to their blood sugar levels.
- Here you’ll find the background information that puts the particular study in the context of other research. This will tell you why this study matters.
- The researcher’s hypothesis (or predictions) will likely be at the end of this section. This will tell you what they were hoping to learn from their experiments.
- This is exactly what it sounds like. The methods section tells you, pretty objectively, exactly what procedures were carried out; if they used human participants, animal subjects, or tissue cultures; and what materials they used.
- The results section will include both raw data–the numbers generated by the experiment without any statistics magic thrown in–and the statistics magic.
- Here there be lots of tables, graphs, and figures. These should report the main data from the experiment and include captions to tell you what the heck is going on in each.
- Using various statistical tests, the researchers determined whether or not their results are statistically significant–that is, that it is much more likely that the results were achieved because of a variable that the researchers intentionally manipulated rather than by chance alone.
- Here is where the authors merge the cold hard facts given in the results section and start interpreting their findings in the broader scheme of things. This is also where they will choose to accept or reject their original hypotheses.
- While this is often the most interesting part of a scientific paper and the part that convinces you of its importance, it is also the part most likely to be tainted by researcher bias. Look out for statements that seem like a bit of a stretch based on the actual data given in the results section.
- Acknowledgements and References
- The acknowledgements are where the researchers thank anyone or any group that contributed to their work, financially or otherwise. Read this section to look for any conflict of interest.
- The references are where the researchers list all the sources that they referenced throughout the paper. If you are interested in digging deep into the topic or if you want to double-check the paper’s authors’ interpretations of any earlier studies, check out some of the works they have cited.
Know the Terms
Scientists have a really strong jargon game. I suggest that, when you’re reading a scientific journal article, you keep a separate tab open on your computer to look up the basic definitions of terms with which you are not familiar.
Correlation ≠ Causation
The battle-cry of statisticians everywhere: Correlation does not imply causation!
I want that line engraved on my tombstone, tattooed across my forehead, and printed on business cards for me to hand out to random passerby. If you take nothing else away from this post, please remember that correlation does not imply causation.
To understand why this is true, let’s look at an example that Bloomberg Magazine provided in an article on the subject. They plotted the number of active Facebook users on the same graph as the yield on Greek government bonds (in order to track Greek’s growing financial crisis).
Based on this graph, it almost looks like the rise of Facebook caused Greece’s financial collapse. But wait…what if Greek’s financial collapse led to a growth in the popularity of Facebook? Or what if a third factor drove both? Or what if a third factor drove the growth of Facebook and a fourth, different factor drove the financial crisis? (Ding ding–I think we have a winner with that last one!)
And therein lies the problem with equating correlation to causation: it is very hard to isolate a definitive cause for an event. Observational data–such as the data on Facebook and Greece shown in that graph–can only determine correlation. Sometimes correlation does imply a relationship, but sometimes it implies nothing other than coincidence.
Controlled trials are the best way to determine causation, but even they are not perfect. In a controlled experiment, researchers try to keep every variable consistent except for the one that they are testing. For example, if someone wanted to test whether or not plants respond better to classical music than rock music, they would want to ensure that everything other than the type of music was being kept consistent between the two sets of plants.
- The plants would need to get the same amount of water, at the same times.
- They would need to receive sunlight in equal amount, intensity, and frequency.
- They would need to be kept isolated from all non-musical sounds, and the rock plants should not be within earshot (plant-shot??) of the classical music, and vice versa.
Even once all of these factors have been controlled for, however, there are things the researchers may not have thought of in advance. Did they check the acidity of the soil before beginning the experiment? What if one plant was already diseased? Many things can confound (a.k.a. mess up) an experiment.
Read Other Sources
Just like you (probably) wouldn’t buy the first house you saw with a “For Sale” sign in the yard, you shouldn’t trust only one source to tell you everything. Look for other sources with similar (or conflicting!) results. Read review papers and meta-analyses to see what other scientists thought of the experiments you read about. Find blog posts and popular science articles to see how other people are reading it. Ultimately, it’s your job to remain critical and discerning, but it always helps to find out about what other smart, discerning people are saying.
Which transitions me nicely to my next point. While I’ve done my best to give you a starting guide to reading scientific papers, there are lots of other resources out there to help you further your journey. After all, if you are a critical, discerning reader, you should be thinking, Why should I trust this girl with a blog? Sure she says she’s getting her degree in neuroscience, but what does that really count for?
Valid questions, my friends! I hope you trust me, but, without further ado, here are some additional resources:
A library guide from Duke defines different types of studies.
Thomson Reuters provides a glossary of scientific terminology.
The Minitab Blog explains how statistical significance works.
Rice University has their own guide on reading scientific articles, in case you trust the experts more than me.
If you’re looking for great places to find articles to start reading, allow me to recommend PubMed and Google Scholar.
Note that many scientific articles are not free to read online from their original sources, but if you do some hunting, you can often find them on university webpages. If you attend a university or college yourself, your school might pay for access to these types of journals, so ask!