Read on Twitter
Today I did a presentation for @hgse master& #39;s students on how to read a research paper in 15 minutes. Thought I& #39;d share my suggestions. Thread follows.
(Research nerds: This is how to read research for PRACTITIONERS, not researchers - I promise I love your methods section.)
(Research nerds: This is how to read research for PRACTITIONERS, not researchers - I promise I love your methods section.)
The audience for research papers is other researchers, not practitioners--so practitioners have to dig to find what they want to know.
Also, a potentially controversial opinion: I think practitioners should generally look for meta-analyses, synthetic literature reviews, and other summaries. That one paper you came across randomly might not be representative of the literature overall.
But, if you DO want to read a research paper*, where should you focus? My recommendation is R3I: relevance, impact, importance, inference.
*I focused mainly on quantitative studies attempting to establish a causal claim, though much of this advice applies more generally.
*I focused mainly on quantitative studies attempting to establish a causal claim, though much of this advice applies more generally.
RELEVANCE comes first. Are the intervention & outcomes ones you care about? Is the context similar - not just student demog but also governance, historical and cultural context, etc? Do we know why & how this intervention worked & whether you can create the same conditions?
If the study is relevant, then move on to the three Is.
IMPACT: What is the direction & magnitude of the impact?
IMPORTANCE: Are the findings statistically and practically significant?
INFERENCE: Can we infer that the treatment *caused* the outcome?
IMPACT: What is the direction & magnitude of the impact?
IMPORTANCE: Are the findings statistically and practically significant?
INFERENCE: Can we infer that the treatment *caused* the outcome?
Where to find R3I in an academic paper: the abstract, intervention & context, and discussion sections
Practitioners, I hereby grant you permission to skip all the Greek, guilt-free. If something about the methods is super-important for interpretation or limits how we can generalize from the findings, it& #39;s the researchers& #39; obligation to mention it in the abstract or discussion.
Finally, no presentation would be complete without a random meme. (Hat tip to my fabulous teaching fellow, Parastoo Massoumi, for the design concept.)
Hope this thread is helpful!
Hope this thread is helpful!
