How to Read a Paper - S. Keshav

Not summarized yet.

https://web.stanford.edu/class/ee384m/Handouts/HowtoReadPaper.pdf

How to Read a Scientific Paper Efficiently and Critically

These are just my notes (almost verbatim) from the video “How to Read a Scientific Paper Efficiently and Critically” of the channel “Scientific Writing with Karen L. McKee”.

https://youtu.be/lXJeU2dzzWo

Scenario 1: Reading a single paper

The general scientific paper structure

  • Abstract
  • Introduction
  • Methods
  • Results
  • Discussion
  • Conclusions

First Goal: Get an overview of the paper

Check

  • Abstract
  • Highlights
  • Conclusions (skip to them)

Why?

Get overall idea of what was found and why the author thinks it is important.

Scanning figures and tables

  • Basically check if the figures and tables agree with the conclusions or outcome said by the author.

  • Strive to get your own impression of the data/evidence before reading the author’s interpreation.

Reading the introduction

Read the instruction trying to find answers to the following questions:

  • What’s the stated topic of the paper and why is it important?
  • What research has been done so far and what information is needed?
  • Is the literature cited appropiate, relevant, and up-to-date?
  • What problems, questions or hypotheses are addressed in the study?

Write the answers to these questions in your own words after reading the introduction.

How they went about answering their questions?

Basically go to the materials and methods part (the methodology)

If you are familiar with the phenomena under study you may skim through the part that describes it and only come back if you has questions. Otherwise, read it more purposefully.

Pay close attention to the experimental design and analyses.

Questions to ask:

  • Is the experiment well-designed?

Basically does it have appropiate controls, randomization, and replication?

  • Are the analytical methods appropiate and up-to-date?

  • What precautions were taken to minimize bias or operator error?

  • Were the statistical tests appropiate?

After reading the methodology ponder about the following questions for some time:

  • What would I have done differently?
  • Was this an in-depth and rigurous study or not?, based on what you have read up to this point?

Make note of your answers.

Skim through the results

Having already scrutinized the figures and tables you already have an idea of the results. The results section thus, adds details or data summaries not readily apparent on the figures and tables.

Answer these questions:

  • Did the study provide clear answers to the questions or hypotheses addressed?
  • Were the findings not clear-cut?

There is nothing wrong with having results not providing complete answers. Nevertheless, if that happens then in the discussion section there should be details about why was this the case.

Reading the discussions

Read and try to find answers to the following questions.

  • How does the author interpret the main findings?
  • Does the author describe alternative explanations for the outcome?
  • How does the main findings compare to previous research in general terms? Do they agree or disagree?
  • What are the main strenghts and limitations of the study? How do they affect the interpretation?

These questions should have been answered by the author in the discussion.

If they dont, the quality of the work or its writing could be improved.

Rereading the conclusions

Reread and answer:

  • Does the paper describe something new or is it a confirmation of prior work?
  • What are the broader implications of the findings?
  • Do the results have relevance to other types of systems/phenomena,etc?
  • Do the findings have any practical applications?
  • What additional needs to be done in the future?

Rereading the whole paper

Useful to corfirm points based on some sections of subsections of the paper.

This is just one of the many approaches for paper reading

So, you know.. it works for her.

Other strategies

  • Reading the paper closer from start to finish, and then go back and ask the questions listed previously.

Main values of reading papers

Be efficient and discerning

Scenario 2: Reading multiple papers

Reading a set of papers in preparation for writing a paper.

Goals:

  • Identify potentially relevant papers by doing literature research using keywords.

Some papers may be more important than others, some may not be relevant at all.

Main strategy:

  • Use the title and abstract to quickly discern those suitable for further reading.

  • Find papers in databases, ie “Web of Science”.

  • Strategically use logical connectors when looking for keywords.

  • You may require the search terms to appear in the title.

  • Using sort on the results wisely:

Recently vs more citations tradeoff.

Some allow to sort by “relevance”.

  • Refine/specialize/narrow the search if too many papers come up. Reduce 70 to 20 papers ie.

  • Scan abstract of each of the results to discern which to read in more detail.

  • Papers directly relevant should have mentioned the search terms in the abstract. That may constitute a good way to discern.

  • Additional papers can be found by looking for the cited literature on the papers you already have found.

  • You may export the list of citations and papers for easier manipulation inside research software and citations manager like mendeley.