Peer Review Best Practices

A thorough, fair, and timely peer review is essential to the integrity of scholarly publishing. This guide outlines the ethical principles and practical techniques that distinguish high-quality reviews.

Ethical Responsibilities

Peer reviewers are trusted custodians of unpublished research. The Committee on Publication Ethics (COPE) identifies the following core obligations:

Obligation	What it means in practice
Confidentiality	Do not share, discuss, or cite the manuscript outside the review process
Integrity	Declare conflicts of interest promptly; do not review work by close colleagues or competitors
Timeliness	Complete the review by the agreed deadline or request an extension in advance
Constructiveness	Provide feedback that helps authors improve the work, whether or not you recommend acceptance
Impartiality	Base your recommendation on the scientific merit of the work, not personal preference

caution

Downloading a manuscript for review does not grant permission to use it in your own research, share it with colleagues, or disclose it to third parties — including AI writing tools — without explicit journal permission.

Structuring Your Review

A well-structured review makes it easy for editors to reach a decision and for authors to act on your feedback. Use the following sections:

1. Summary

Write 2–4 sentences describing:

• The paper's main research question or hypothesis
• The methods used
• The principal findings
• The claimed significance

This demonstrates you have read and understood the paper, and it signals to the author whether your interpretation matches their intent.

2. Major Concerns

List issues that must be resolved before the paper can be accepted. Number each concern so authors can respond point-by-point. Major concerns typically include:

• Unsupported or overstated conclusions
• Critical gaps in methodology
• Missing controls or comparisons
• Statistical errors that invalidate key results
• Reproducibility problems (missing data, code, or protocols)

3. Minor Concerns

List issues that should be addressed but would not alone prevent acceptance:

• Unclear or ambiguous language
• Figures that lack axis labels or legends
• Missing or incorrectly formatted references
• Inconsistencies between text and tables

4. Recommendation

Select one of the standard OJS options:

Recommendation	Meaning
Accept Submission	Ready to publish with no or trivial changes
Revisions Required	Acceptable after authors address specific concerns
Resubmit for Review	Substantial revision needed; revised version requires re-review
Decline Submission	Fundamental flaws that cannot be addressed by revision

Your recommendation should follow logically from the concerns you have listed.

Constructive vs Destructive Feedback

Constructive	Destructive
"The sample size (n = 12) is insufficient to support the claim in line 45. A power analysis or acknowledgment of this limitation is needed."	"The sample size is way too small. This is a serious flaw."
"Figure 3 lacks axis labels and units, making it impossible to interpret the scale."	"The figures are poor quality."
"The literature review omits recent work on X (e.g., Smith 2022, Jones 2023) that directly addresses the research question."	"This has been done before."

Focus on the work, not the author. Phrases like "the authors should…" or "it would strengthen the paper if…" maintain a professional tone.

tip

If you would be uncomfortable saying a comment aloud to the author at a conference, revise it before submitting.

Common Errors to Watch For

Methodological

• Sampling strategy not described or inappropriate for the research question
• Lack of a control group or baseline comparison
• Instrumentation or measurement tools not validated
• Potential confounds not acknowledged

Statistical

• Inappropriate statistical test for the data type or distribution
• Multiple comparisons without correction (e.g., Bonferroni, FDR)
• Confidence intervals or effect sizes absent when significance is reported
• Correlation reported as causation

Logical

• Conclusions that exceed what the data can support
• Cherry-picked results presented without full context
• Circular reasoning (using the conclusion to justify the methodology)
• Failure to address alternative explanations

Avoiding Bias

Reviewers must guard against common cognitive biases:

Bias type	Description	Mitigation
Confirmation bias	Favouring results that align with your own views	Focus on methodology, not conclusions
Affiliation bias	Being overly positive or negative based on the author's institution	Remind yourself of double-blind purpose
Novelty bias	Rewarding surprising results regardless of rigour	Apply the same methodological standard to all papers
Scope creep	Asking authors to do additional experiments beyond the stated scope	Limit requests to what is necessary to support existing claims

If you realise mid-review that you hold a conflict of interest, stop reading, notify the editor via the OJS Discussion thread, and decline the assignment.

Confidentiality Obligations

• Do not share the manuscript or any part of it with colleagues, students, or collaborators without the editor's permission.
• Do not use ideas, data, or methods from the manuscript in your own work before it is published.
• Do not disclose that you are reviewing (or have reviewed) this specific submission.
• Post-publication, you may mention your role as reviewer only if the journal operates a fully open review policy that already discloses reviewer identity.

Suspected Plagiarism or Data Fabrication

If you encounter evidence of research misconduct, follow these steps:

1. Do not alert the authors directly.
2. Document your concerns with specific examples (page numbers, figures, suspect passages).
3. Open the Discussion thread in OJS and contact the editor confidentially.
4. Do not submit your review until the editor has acknowledged the concern and provided guidance.

Common indicators include:

Indicator	Possible misconduct
Passages that read differently from the rest of the paper	Plagiarism
Figures with suspiciously clean or identical noise patterns	Image manipulation
Data distributions that are implausibly uniform	Data fabrication
Results that exactly match a previous paper by different authors	Duplicate submission or self-plagiarism

Editors are responsible for investigating misconduct; reviewers are responsible for reporting it. COPE provides detailed flowcharts for editors at publicationethics.org.

Further Reading

• Learning OJS – Reviewing — Completing the review form and submitting your recommendation
• COPE Peer Review Guidelines — Full ethical guidelines for peer reviewers
• PKP Community Forum — Ask questions and find answers from the OJS community