How to Read an Academic Research Paper

A Practical, Multi-Pass Strategy

Bogdan G. Popescu
bgpopescu@tec.mx

Tecnológico de Monterrey

Today’s Goals

  • Learn a multi-pass strategy for reading papers
  • Feel less intimidated by academic writing
  • Extract key ideas efficiently

By the end: You’ll have a repeatable reading workflow.

It’s Not Just You

Designed for experts:

  • Dense, specialized jargon
  • Complex statistical tables
  • Authors write for peer reviewers, not students

Good news:

  • Confusion is normal
  • Even professors re-read papers
  • Reading skill is rarely taught — but learnable
  • Your job: extract core ideas

Our Running Example

“Waking Up the Golden Dawn”

Full title: “Waking Up the Golden Dawn: Does Exposure to the Refugee Crisis Increase Support for Extreme-Right Parties?”

Published: Political Analysis (2019)

Authors: Dinas, Matakos, Xefteris & Hangartner

Source: Dinas et al. (2019), front page.
https://tinyurl.com/y8bddrky

What’s This Paper About?

  • Question: Did refugee arrivals boost far-right voting?
  • Setting: Greek islands, 2015 refugee crisis
  • Design: Islands received different refugee flows
  • Finding: ~2 percentage-point rise in Golden Dawn vote

The geography and timing of the refugee crisis in the Aegean Sea. Source: Dinas et al. (2019), Figure 1.

The Three-Pass Strategy

Don’t Read Start to Finish!

Bad strategy: Open page 1, read every word to the end.

Good strategy: Make multiple passes with different goals.

Based on: Keshav’s (2007) “three-pass approach.”

The Three Passes at a Glance

%%{init:{"flowchart":{"useMaxWidth":true},"themeVariables":{"fontSize":"22px"},"flowchart":{"nodeSpacing":50,"rankSpacing":60},"width":1100,"height":500}}%%
flowchart LR
  A["<b>Pass 1</b><br/>15–20 min<br/>Main Argument"] --> B["<b>Pass 2</b><br/>15–20 min<br/>Evidence"]
  B --> C["<b>Pass 3</b><br/>Optional<br/>Deep Methods"]
  style A fill:#4a7c6f,color:#fff,stroke:#334155
  style B fill:#b7943a,color:#fff,stroke:#334155
  style C fill:#b44527,color:#fff,stroke:#334155

Total: 30–40 minutes for core understanding.

Pass 1: The Big Picture

What to Read in Pass 1

%%{init:{"flowchart":{"useMaxWidth":true},"themeVariables":{"fontSize":"22px"},"flowchart":{"nodeSpacing":50,"rankSpacing":60},"width":1100,"height":500}}%%
flowchart LR
  T["<b>Title &<br/>Abstract</b><br/>~2 min"] --> I["<b>Introduction</b><br/>~10 min"]
  I --> C["<b>Conclusion</b><br/>~5 min"]
  C --> S["<b>Skip</b><br/>everything<br/>else for now"]
  style T fill:#4a7c6f,color:#fff,stroke:#334155
  style I fill:#4a7c6f,color:#fff,stroke:#334155
  style C fill:#4a7c6f,color:#fff,stroke:#334155
  style S fill:#64748b,color:#fff,stroke:#334155

The Title Tells You a Lot

“Waking Up the Golden Dawn: Does Exposure to the Refugee Crisis Increase Support for Extreme-Right Parties?”

  • Topic: Refugee crisis and far-right voting
  • Question: Does exposure increase support?
  • Setting: Greece, Golden Dawn party

Golden Dawn rally, Athens (2012). Photo: Steve Jurvetson, CC BY 2.0.

Reading the Abstract

  • The question: What are they asking?
  • The answer: What did they find?
  • The method: How did they study it?
  • So what: Why does it matter?

Golden Dawn decoded: Refugee exposure → ~2 pp increase (≈44% rise at the mean) via natural experiment across Greek islands. Mere exposure can fuel anti-immigrant politics.

What to Look for in the Intro & Conclusion

Introduction:

  • Why this matters: Major crisis, European far-right surge
  • Previous research: Mixed contact-theory evidence
  • What’s new: Sudden 2015 arrivals as natural experiment

Conclusion:

  • Key result: ~2 pp increase in Golden Dawn vote share
  • Mechanism: Strongest where refugees were visible
  • Limitation: Can’t determine if effect is permanent

Honesty about limitations = good research.

Pause and Reflect

After Pass 1, write down three things:

  1. The main question in your own words
  2. The main answer in your own words
  3. One thing you’re confused about

This becomes your foundation for Pass 2.

Pass 2: The Evidence

What to Read in Pass 2

  1. All figures and tables (~10 min)
  2. Research design section (~10 min)

Goal: Understand HOW they answered the question.

How to Read Figures

You don’t need to understand the statistics!

  • What is being compared?
  • What pattern do they show?
  • Does this support the main argument?

Figures tell stories — look for visual patterns first.

Golden Dawn: Key Visual Pattern

Parallel trends at the municipal and township level. Treated islands diverge sharply after the 2015 refugee crisis. Source: Dinas et al. (2019), Figure 2.

Reading Tables Without Panic

  • Rows: Different groups or time periods
  • Columns: Different model specifications
  • Numbers: Estimated effects
  • Stars (*): Statistically significant results

Focus on: direction (positive/negative) and consistency.

Golden Dawn Results: What to Notice

  • Positive numbers throughout → exposure ↑ votes
  • Larger effects where refugees were visible
  • Consistent across specifications

Translation: The finding is robust-holds up multiple ways.

Understanding the Research Design

Core comparison:

  • Islands with many refugees vs. few/none
  • Were islands similar before refugees arrived?
  • Can we rule out alternative explanations?

Refugees crossing from Turkey to Lesbos, Jan 2016. Photo: Mstyslav Chernov, CC BY-SA 4.0.

Understanding Causality

What Does “Causal” Mean?

Causal claim: X causes Y (not just correlation).

Challenge: How do we know it’s not something else?

The Natural Experiment Logic

%%{init:{"flowchart":{"useMaxWidth":true},"themeVariables":{"fontSize":"20px"},"flowchart":{"nodeSpacing":40,"rankSpacing":50},"width":1100,"height":550}}%%
flowchart LR
  A["<b>Treatment</b><br/>Islands receiving<br/>many refugees"] --> C["<b>Compare</b><br/>Did Golden Dawn<br/>vote share rise<br/>MORE here?"]
  B["<b>Control</b><br/>Islands receiving<br/>few/no refugees"] --> C
  C --> D["<b>Assumption</b><br/>Islands were<br/>similar before<br/>2015"]
  style A fill:#b44527,color:#fff,stroke:#334155
  style B fill:#4a7c6f,color:#fff,stroke:#334155
  style C fill:#b7943a,color:#fff,stroke:#334155
  style D fill:#64748b,color:#fff,stroke:#334155

Visualizing Treatment vs. Control

Proximity to Turkey predicts change in Golden Dawn vote share (left). No such relationship before the crisis - the placebo test (right). Source: Dinas et al. (2019), Figure 4.

Exercise: Spot the Comparison

Prompt: In the Golden Dawn study, the treatment is “islands receiving many refugees” and the control is “islands receiving few.” What could go wrong with this comparison? Brainstorm two possible threats to validity with a partner.

Estimated time: 5 minutes

Pass 3: Methods (Optional)

Pass 3: When and How

Do it if:

  • Writing about the paper in detail
  • Preparing for in-depth discussion
  • Planning similar research

What to expect:

  • The methods/analysis section in full detail
  • The hardest part of the paper
  • Can take 1–4 hours depending on complexity

Skip if you just need the main ideas. Strategy: Go slow, look up terms, ask for help.

Reading Critically

Critical Reading

Don’t just accept everything!

  • Is the comparison fair?
  • Are there alternative explanations?
  • Does the evidence support the conclusion?
  • What are the limitations?

Good papers acknowledge their limits. No single study proves anything definitively-science advances through accumulation.

Critical Questions for Golden Dawn

  • Could tourism have declined on exposed islands?
  • Might media coverage have differed across islands?
  • Would the effect persist over time?
  • Could economic factors explain the results?

The authors address some-but not all.

Making Reading Stick

Recognition ≠ Recall

What feels like learning:

  • Highlighting key passages
  • Rereading sections
  • Recognizing terms when you see them again

What actually sticks:

  • Reconstructing the argument from memory
  • Explaining it in your own words
  • Noticing what you can’t yet say

Evidence: Dunlosky et al. (2013) reviewed 10 study techniques. Highlighting and rereading rated low utility; practice testing and spaced practice rated high utility.

Close the Book

After Pass 2, close the paper and write from memory:

  1. The research question (one sentence)
  2. The main finding (one sentence)
  3. The core comparison or design (one sentence)

Then check what you missed.

The act of retrieval — even imperfect — is what builds durable understanding (Karpicke & Roediger, 2008).

Reading Across Papers

For a literature review, don’t summarize papers one at a time.

  • Read 3–5 papers on the same question
  • Where do they agree? Where do they disagree?
  • Whose evidence is stronger, and why?
  • What question would the next paper need to answer?

Adler (1972) calls this syntopical reading — and it’s where your own ideas start.

AI Can Help-But Has Limits

AI CAN:

  • Explain jargon in plain language
  • Clarify confusing sentences
  • Define statistical concepts

AI CANNOT:

  • Replace actually reading the paper
  • Evaluate evidence critically for you
  • Guarantee factual accuracy-AI can invent citations that don’t exist and misrepresent findings

Always verify before trusting.

AI Prompts: Good vs. Bad

Good prompts:

  • “What does ‘difference-in-differences’ mean simply?”
  • “Rephrase this sentence: [paste sentence]”
  • “Did I understand correctly? [your explanation]”

Bad prompts:

  • “Summarize this entire 30-page paper” → skips your learning
  • “Give me references on this topic” → AI invents citations
  • “Write a critique of this paper’s methods” → plagiarism risk

Practical Tips

Create a Reading Template

For each paper, write one sentence for each:

  1. Research question
  2. Main finding
  3. Method used
  4. One strength
  5. One limitation
  6. Why it matters

What to Focus On

You can skip:

  • Every equation or statistical test
  • Every citation in the literature review
  • Every technical detail in the appendix

You must understand:

  • Main question and answer
  • Basic logic of the comparison
  • Key evidence (direction and consistency)
  • Major limitations

Reading Gets Easier Over Time

Author’s illustration of a typical learning curve.

Key Takeaways

Remember

  1. Confusion is normal-even for experts
  2. Don’t read linearly-use the multi-pass strategy
  3. The question is your compass-find it first
  4. Figures tell stories-you don’t need advanced stats
  5. Read critically-ask questions, identify limits
  6. Reconstruct from memory-recognition isn’t recall
  7. Use AI wisely-as assistant, not replacement

Reading is a skill. You improve with practice.

References

Works Cited

Adler, M. J., & Van Doren, C. (1972). How to read a book (Rev. ed.). Simon & Schuster. (Original work published 1940)

Dinas, E., Matakos, K., Xefteris, D., & Hangartner, D. (2019). Waking up the Golden Dawn: Does exposure to the refugee crisis increase support for extreme-right parties? Political Analysis, 27(2), 244–254. https://doi.org/10.1017/pan.2018.48

Dunlosky, J., Rawson, K. A., Marsh, E. J., Nathan, M. J., & Willingham, D. T. (2013). Improving students’ learning with effective learning techniques: Promising directions from cognitive and educational psychology. Psychological Science in the Public Interest, 14(1), 4–58. https://doi.org/10.1177/1529100612453266

Karpicke, J. D., & Roediger, H. L. (2008). The critical importance of retrieval for learning. Science, 319(5865), 966–968. https://doi.org/10.1126/science.1152408

Keshav, S. (2007). How to read a paper. ACM SIGCOMM Computer Communication Review, 37(3), 83–84. https://doi.org/10.1145/1273445.1273458