Skip to Main Content
It looks like you're using Internet Explorer 11 or older. This website works best with modern browsers such as the latest versions of Chrome, Firefox, Safari, and Edge. If you continue with this browser, you may see unexpected results.

Evaluating Scientific Information (if you're not an expert): How to Read & Analyze a Scientific Research Article

tips for critically evaluating scientific research articles and news reports about research

Understanding a Scientific Research Article

When you get lost in a sea of unfamiliar terms and abbreviations - Wikipedia to the rescue!  Use it to look up any terms (or concepts, or theories) you don't know.

Wikipedia entry for IMRAD

Anatomy of a Scientific Research Article

For more than 40 years, research papers in the sciences and social sciences have been organized by IMRAD format. (More accurately, it should be TAB/IMRaC/DR sections, but that's difficult to say.) This stands for:

  • Title
  • Abstract
  • Background Introduction (which includes the research's hypothesis or thesis statement)
  • Research Methods
  • Results [and]
  • Conclusion / Discussion (which sometimes includes recommendations for further research)
  • References cited in the paper

However:

Reading a research paper from start to finish can be more confusing than helpful! 

How to skim a scientific research article

Most scientific research articles are organized in the same way - and they make the most sense if you don't read them from start to finish, but read them out of order, section by section. Start with the abstract (the author's summary of their paper).

Then skip to the conclusions and discussion sections to learn what the research study found.

Next up as step 3 -  if the article still looks useful - read the Introduction. This should give you background context for this research study, why it matters and how it relates to other research studies.

And only then, when you think this article is still relevant to your question, go back to skim the description of the research project. and its methods. Be sure to note the size of the study (just a few? or thousands?) and if it was a controlled study (the "gold standard" for science is a double-blind study).

Image credits:  Kapi'olani Community College's Library & Learning Resources webpage on how to read a scholarly article.

Analyzing a Scientific Research Article (When You're not a Scientist)

Even if you are not an expert in that topic, there are still ways to judge the validity or accuracy of a scientific research paper. Start with the basic questions for evaluating a website (either the PROVEN or the CRAAP tests), then also ask:

- Was this particular item peer-reviewed? Not all articles in a peer-review journal go through that process (opinion pieces do not, for example). Check whether there are any details about the peer review for that specific article on the paper itself. If you don't see any details, look up the editorial process for that journal.

- Who paid for the research project? Was it funded by a government grant, or by a company or agency that could have a vested interest in the study outcomes? (Authors of scientific papers are also supposed to disclose any financial relationships with interested parties, tor any potential conflicts of interest.) While no agency is likely to find research studies out of simple curiosity, if the funder has an obviously vested interest in getting a particular result from the research study, examine that study very critically.

- Who did the research? Was it conducted in one lab by graduate students or  by researchers who all worked for the same project director? Or was the work carried out in more than one place, or by authors at different institutions?

- If the research study looked at multiple instances, how big was the sample size? Sometimes this is expressed as N=[a number]. Does that sample seem large enough to include all possible explanations for what's being studied? (For example, the now-discredited article that suggested a connection between the MMR vaccine and autism was based on only 12 children.)

- Does the research project actually study the population in question, or does it use a surrogate? Because it's not ethical to knowingly cause harm to people, nearly all research into human health uses some other organism instead (specially-bred lab mice, for example); or it looks at factors associated with the actual research problem, to infer that it might work the same way in people. But the more inferences made in the study, the greater chance that's not the case. Correlation is not causation!

- Are the statistics credible? Even if you're not a statistician, there are a couple of tell-tale things to look out for. If there are tables or graphs, can you easily figure out what information those represent? Are the data points in the graphs constant (not skipping numbers in the series)? If the Results section talks about percentages, does it also give confidence intervals and standard deviations  - and if it does, are those significant? (Tip: the confidence interval should be at least 95%, and/or have a low standard deviation.)

- Look over the citations in the Reference list. Are there many different authors cited, or just a small number (or worse, only the authors of that same paper)?