Write a good review

A good review does two things. It tells the editors whether the work is sound, and it tells the authors how to make it better. It is specific, fair and polite.

Before you start

• Do you have a conflict of interest? If you work with the authors, compete with them, or stand to gain or lose from the outcome, decline.
• Is it your field? If only part of it is, say so in your confidence and in your comments to the editors.
• Can you finish by the due date? If not, decline at once so someone else can be asked.
• Keep it confidential. Do not share the submission, do not use its ideas, and do not put it into a public AI tool.

Read twice

1. First reading: what is the question, what is the answer, and do you believe it?
2. Second reading: go through section by section and note where you stopped believing, or stopped understanding.

A structure that works

1. Summary

Two or three sentences that say, in your own words, what the work does and finds. This shows the authors you understood it, and helps the editors.

This abstract reports a solar dryer built from local materials and compares it with open sun drying of mango slices over two seasons. The dryer halves the drying time and keeps more vitamin C.

2. Strengths

What is good about it. Be exact. Every submission has something.

The comparison over two seasons is the strength of this work, and the results are reported with clear numbers.

3. Main concerns

The few things that affect whether the conclusions hold. Number them. For each, say what the problem is, why it matters, and what would settle it.

1. The sample size for each run is not given, so the reader cannot judge how firm the difference is. Please state the number of runs and give a measure of spread.

4. Smaller points

Clarity, missing references, figures, language. A numbered list, with the place: “Section 2, second paragraph”.

5. Recommendation

Choose it to match what you wrote. If you list three serious concerns, “accept” will confuse the editors.

Use the whole form

The review form has four parts, and each does a different job. A review that fills in only the recommendation leaves the editors with nothing to weigh.

Part	What it is for	Who reads it
Notes on passages	A remark about one place in the text.	Editors and authors
Scores, with a short reason for each	How the work does on four separate questions.	Editors and authors
Comments to authors	Your reading of the whole work: summary, strengths, concerns.	Editors and authors
Comments for the editors only	What the authors should not read.	Editors

What each score means

Score each question by itself. A submission can be very clear and not new, or important and badly explained. Four different scores tell the editors more than four equal ones.

	Originality	Soundness	Clarity	Significance
The question	Is it new?	Are the methods and the reasoning right?	Is it well written and well organised?	Does it matter to the field?
5	A new question, method or result.	The method answers the question, and the conclusions follow from the results.	Can be read once and understood.	Others will use or build on it.
4	A clear step beyond what is known.	Sound, with small gaps that do not change the conclusions.	Clear, with a few places to tidy.	Useful to people working on this topic.
3	A known approach applied to a new case.	Mostly sound. One point needs to be checked or explained.	Understandable with effort.	Of some interest, mostly local or narrow.
2	Largely repeats known work.	A weakness that puts the main conclusion in doubt.	Hard to follow in several places.	Little that others could use.
1	Nothing new that you can find.	The method cannot answer the question, or the conclusions do not follow.	You could not work out what was done.	You cannot see who it is for.

Good practice:

• Give a reason for any score of 1, 2 or 5. One sentence under the score is enough: “Soundness 2: the two groups were measured in different seasons.”
• Use the whole scale. If everything you review gets a 4, your scores tell the editors nothing.
• Score the work as its kind. An abstract is a summary. Do not mark its soundness down because the full method is not there, when the method it names is a sensible one.
• Your scores and your recommendation must agree. Soundness 2 with Accept will be queried.
• Confidence is about you, not the work. If the statistics are outside your field, say 2 - Low and tell the editors which part you could not judge.

Notes on passages, or comments?

Use a note on the passage when	Use your comments when
A number, unit or reference is wrong in one place.	A concern affects the conclusions.
One sentence cannot be understood.	The whole section is hard to follow.
A figure or table is not explained where it appears.	Something is missing from the work altogether.
A claim at this point needs a source.	You are summarising, or naming strengths.

• Write a note that can be acted on. “Unclear” does not help. “Which of the two dryers does ‘it’ refer to here?” does.
• A few good notes are better than fifty. For language, mark two or three examples and ask for a careful reading of the rest.
• Say the important things twice. A serious concern belongs in your comments, numbered, even when you also marked where it shows.
• Notes do not replace the summary. Authors read your comments first.

Judge the right things

Judge	Do not judge
Whether the question is clear and worth asking	Whether you would have studied something else
Whether the method can answer the question	Whether it is the method you prefer
Whether the conclusions follow from the results	The authors, their institution or their country
Whether a reader could repeat the work	Their English, where the meaning is clear

For an abstract, you are judging a summary. Ask whether the question, the method and the result are there and believable. Do not ask for what only a full paper could hold.

For a poster, judge the four sections as you would an abstract.

Wording that helps

In place of	Write
“The methods are weak.”	“The method does not say how the slices were chosen. Please describe the sampling.”
“This is badly written.”	“Section 3 was hard to follow. Stating the aim at the start of the section would help.”
“The authors clearly do not know the literature.”	“The work would be stronger if it discussed [the specific studies].”
“This is wrong.”	“I could not see how the result in Table 2 follows from the data in Table 1. Could the authors explain?”
“Cite my papers.”	Suggest a reference only where it truly matters, whoever wrote it.

Write as you would wish to be written to. Many authors are early in their careers, and many are not writing in their first language.

Comments for the editors only

Use these for what the authors should not read:

• a suspicion of plagiarism or of work sent to two places, with your evidence;
• an ethical concern;
• how sure you are, and which parts you could not judge;
• a plain statement of your view, if you softened it for the authors.

Do not put criticism here that the authors need in order to improve the work.

Common mistakes

Mistake	Why it hurts
One line: “Good paper, accept.”	Gives the editors nothing to decide on.
A list of faults and no strengths	Reads as hostile; authors stop listening.
Asking for a different study	Review the work that was done.
Rewriting the English line by line	Point to a few examples and ask for a careful reading.
A recommendation that does not match the comments	The editors cannot tell what you think.
Sending it late	Authors and other reviewers wait on you.

Checklist

• I summarised the work in my own words.
• I named at least one strength.
• Each concern says what, why, and what would settle it.
• I scored all four questions separately, and gave a reason for each low or top score.
• My notes on passages each say what to change.
• My scores, my comments and my recommendation agree.
• My confidence says honestly how well I know the subject.
• Nothing in my comments to the authors reveals who I am, where the call reviews without names.
• I would be content to receive this review.