01
Ask: what is the question?
Read the title and abstract. What exactly did researchers want to understand? Was it temperature, energy use, comfort, or health?
Research & knowledge / Learning guideNo experience needed
How to read
a research paper.
You don't need a research degree to ask good questions about evidence. Start here.
Your first five minutesWhere to
Where to
begin.
Most people do not need to read an academic paper line by line. A few careful checks go a long way.
02
Look at who and where.
Was the study done in a single home, an entire city, a laboratory, or a simulation? Which climate, building type, and season?
03
Check how it was measured.
Actual energy bills? Indoor sensors? Satellite surface temperatures? Modeling? The method determines what conclusions are possible.
04
Separate results from interpretation.
Read the figures and results. Then compare them with the discussion and conclusions. Look for magnitudes, uncertainty, and sample size.
05
Find the limitations.
A finding may apply only to specific walls, seasons, equipment, or climates. A correlation does not by itself prove a cause.
06
Visit the original paper.
Use the DOI link to check the original publication. A publisher page may be paywalled; look for a legal open-access version when available.
A research paper at a glanceThe main
The main
sections explained.
Abstract
A very short outline of the question, approach and main finding. Start here, but do not stop here.
Introduction
Why the question matters and what earlier studies did or did not show.
Methods
Who or what was studied, how data were collected, and what analytical tools were used.
Results
The measured or modeled findings, including charts, tables and uncertainty.
Discussion
What the researchers think the results mean and how they compare with other work.
Limitations
What the study cannot establish and where generalization may be inappropriate.
References & DOI
Where the evidence came from and how to find the original publication reliably.
Which kind of paper is this?Different papers.
Different papers.
Different jobs.
A headline saying “scientists found” does not tell you how evidence was gathered. These common publication types answer different questions.
Original research article
Researchers collect or analyze data to answer a particular question. This might involve indoor temperature sensors, energy bills, satellite images, laboratory tests or simulations.
For example
Researchers place sensors in ten homes during a heatwave and report what the instruments measured.
Check: How many homes? Were there comparison homes? Was this measured or modeled?
Review article
Authors describe and discuss earlier studies on a topic. A general or narrative review can give useful background, but how papers were chosen may be less systematic.
For example
An article explains what past studies say about roof insulation, windows and summer comfort.
Check: Which studies were included? Was the selection method explained?
Systematic review
Researchers follow a documented process to search for, select and assess studies on a defined question. It can help reveal where findings agree and where evidence is weak.
For example
A team searches databases using pre-set criteria to find studies testing urban tree cooling.
Check: Which databases and years? Were weaker studies evaluated?
Meta-analysis
Researchers statistically combine results from multiple studies when the data are suitable. It may give a more precise overall estimate, but combining very different studies can mislead.
For example
Researchers combine comparable studies to estimate a typical measured response, while checking why results differ.
Check: Were the measurements comparable? How much did results vary?
Preprint
A manuscript shared publicly before formal journal peer review. It can provide early access to useful ideas, but its methods or conclusions may change.
For example
A researcher shares a new analysis of heat-pump performance while it is being evaluated by a journal.
Check: Has a reviewed version appeared? Is this the latest version?
News story or blog
Reporting can make research accessible, but it is usually not the original study. Headlines may simplify findings or omit important limits.
For example
A website says “Trees reduce city heat by 10 degrees!” without saying which temperature, where or under what conditions.
Check: Can you click through to the original study and verify the claim?
Helpful further reading: NIH guide to research methods ↗ and Cochrane guidance on meta-analysis ↗.
Common mistakesRead results
Read results
with care.
Air temperature ≠ surface temperature
A satellite may show that a black rooftop is 50°C while a lighter rooftop is 38°C. That does not mean the living room became 12°C cooler or that the outdoor air dropped by 12°C. The camera is measuring the surface, not the room or the surrounding air.Ask: What temperature was measured—surface, outdoor air, or indoor air?
Association ≠ proof of cause
Imagine researchers compare a leafy neighborhood at 27°C with a less green area at 30°C. It is tempting to say the trees caused all 3°C of difference. But the areas might also have different building density, traffic, elevation or wind. Without additional tests, we cannot know how much difference comes from trees alone.Ask: Did researchers compare equivalent conditions or rule out other explanations?
Model prediction ≠ measured savings
A computer model predicts that upgrading a home’s insulation will save $300 a year. That is a prediction based on assumed weather, energy prices, thermostat settings and construction quality. The family’s actual bills might show a different saving.Ask: Is this a predicted saving, or did someone measure it in a real home?
One result ≠ a universal rule
A study finds that exterior shading helped reduce overheating in houses in a hot, sunny climate. That does not tell us the exact savings for an apartment in Toronto, or for a house where winter heating matters more. The same design can have different benefits and drawbacks in different places.Ask: Is the building, climate and situation similar to mine?
Now put it into practiceOne question.
One question.
Evidence to explore.
Open a paper in our library, read its summary, and compare the explanation with the original study.
Explore reviewed studies ↗