Managing conflicts of interest in conference peer review

Define conflicts in writing

A policy that says “avoid conflicts” is not enforceable. List what counts, so reviewers and chairs apply the same rule.

• Co-authorship within the last [three] years.
• Same institution or department.
• Supervisor, student or close collaborator relationships.
• Financial or commercial interest in the work.
• Personal relationships that could reasonably affect judgement.
• Committee-specific cases, for example the proposer of a symposium reviewing its submissions.

Collect declarations at three points

Ask authors to name people who should not review the work when they submit. Ask reviewers to declare conflicts when they accept. Ask them again when they open an assigned abstract, because the author list is only visible then.

Enforce before scoring

A conflict should block the assignment or hide the abstract, not appear as a note after a score has been entered. Where the system allows it, use the author and institution data on the record to flag likely conflicts automatically, and let a chair confirm them.

When the chair is conflicted

Programme chairs and track leads submit abstracts too. Decide in advance who takes over decisions on their submissions and record that in the policy, so the process does not depend on someone noticing.

A sample conflict-of-interest policy

Adapt this outline to your community. Keep it to one page and publish it on the reviewer page and next to the submission form, so that authors and reviewers see the same rules.

CONFLICT OF INTEREST POLICY 1. Purpose The committee aims to review every abstract fairly and to be seen to do so. 2. What counts as a conflict A reviewer has a conflict with an abstract if, within the last three years, they have co-authored with any author, worked in the same department, supervised or been supervised by an author, or if they have a financial or personal interest in the outcome. 3. Declaring Authors may name up to three people who should not review their abstract. Reviewers must declare a conflict before scoring and must not read or score the abstract. 4. Enforcement The programme chair checks declarations and reassigns affected abstracts. Conflicted reviewers cannot see the reviews of that abstract. 5. Committee members Committee members who submit abstracts do not take part in decisions on them. 6. Records Declarations and reassignments are recorded and kept with the abstract.

Three scenarios and how to resolve them

Policies are tested by concrete cases. These three come up in most calls, and it helps to agree in advance how you will handle them.

• A reviewer realises, halfway through reading, that they know the author from a previous collaboration. They stop, declare the conflict, and the chair reassigns the abstract. Any score already entered is discarded.
• The topic is so narrow that almost every qualified reviewer has some connection to the group. Accept minor, historic links, for example a co-authorship seven years ago, but require disclosure and ask a second reviewer without any link to read the abstract.
• A track lead submitted an abstract in their own track. Another track lead or the programme chair takes the decision, and the track lead is hidden from reviews and discussions of that abstract.

Self-declaration and automatic detection

Relying only on reviewers to declare conflicts leaves gaps, because people forget or do not think a link matters. Relying only on software leaves different gaps, because software sees co-authorship and institution but not personal relationships. Use both, and treat the automatic flags as prompts for a human to confirm.

• Self-declaration catches personal and informal relationships, but depends on memory and honesty.
• Automatic detection from author and affiliation data catches shared institutions, but produces false positives and misses spelling variants.
• Author-named exclusions catch rivals and sensitive cases, but can be abused; the chair should review unusual requests.
• A final check by the programme chair before scoring closes the loop and creates a record.

Conflicts in single-blind and double-blind review

The review model changes how conflicts work. In single-blind review, reviewers see the authors and can identify conflicts themselves. In double-blind review, reviewers cannot see the authors, so they cannot know about a conflict, and the system or the chair must check for them. In that case authors’ exclusion lists and automatic matching carry more weight, and reviewers should be told to report if they suspect they know the author from the content, such as a distinctive dataset. Decide which model you use before the call opens, since it affects what authors must provide and which fields are hidden.

What to do when a conflict is found late

Sometimes a conflict emerges after scores have been entered, or even after decisions. Do not ignore it, and do not quietly change the outcome. Record what was found and when. If the conflicted reviewer’s score affected the decision, ask another reviewer to assess the abstract independently and let the chair decide with that information. If the decision is already communicated and would change, contact the author, explain honestly and correct it. A visible correction protects your process more than a hidden one, and the next chair will know what to do in a similar case.

A reviewer declaration form

A short form keeps declarations consistent. It can be part of the acceptance email or the reviewer’s account page.

Reviewer declaration Name: [ ] I have read the conflict of interest policy: [yes] For each assigned abstract: [ ] No conflict [ ] Conflict (choose): co-author · same institution · supervisor or student · financial interest · personal relationship · other Comment (optional, visible to the chair only): [ ]

Telling authors about the policy

Authors should know that a conflict policy exists and how it protects them. State it in one paragraph next to the submission form, and say how they can ask for someone not to review their work. Doing so early reduces complaints later and shows that the committee takes fairness seriously.

Keep a record

Store declarations, overrides and the reason for any exception with the abstract. If an author later challenges a decision, the committee can show what was declared and how it was handled. Tell reviewers what will be recorded, and keep it visible only to those who need it.