“Conflict of interest” is not one category but two independent classification schemes layered on top of each other. One axis measures certainty — how confirmed the risk of bias actually is: actual, potential, or perceived. The other axis measures source — where the competing interest comes from: financial, non-financial, personal, or institutional. A single real-world situation is classified on both axes at once — for example, an investigator’s equity stake in a sponsor company is a financial conflict that may be actual, potential, or perceived depending on the specific facts. Confusing the two axes is the most common source of “what are the types of COI” mismatches between different training materials and policies. This page defines both, shows how they combine, and explains why institutions typically manage perceived conflicts with the same seriousness as actual ones. See CASRAI’s Conflict of Interest (COI) in Research entry for the underlying definition this page builds on.
Axis one: actual, potential, and perceived conflicts of interest
This axis answers “how confirmed is the risk?” — it is a spectrum of certainty, not three unrelated categories.
- Actual conflict of interest. The competing interest already exists and directly bears on the decision, judgment, or research task at hand right now — there is no “if” about whether the two interests are in tension, only about how the conflict is managed. Example pattern: a grant reviewer is asked to score an application submitted by their own spouse’s lab.
- Potential conflict of interest. The competing interest exists, but the circumstances that would make it bear on a specific decision have not yet arisen, or may never arise. The risk is real enough to require disclosure and monitoring, but nothing currently in front of the person is actually compromised. Example pattern: a researcher is in early-stage consulting discussions with a company that might, months later, sponsor a study in the researcher’s own field — no sponsored study exists yet, but if one is proposed, the existing relationship would become an actual conflict.
- Perceived (or “apparent”) conflict of interest. A reasonable, informed observer could conclude that the relationship creates a risk of bias, even where the person involved is confident — and may in fact be correct — that no bias exists. The defining feature of a perceived conflict is that it is evaluated from the outside: it does not depend on whether the individual believes themselves to be objective, because self-assessment of one’s own bias is exactly what disclosure and independent review exist to check. Example pattern: a journal peer reviewer previously co-authored a paper with the submitting author eight years ago and has had no contact since — the professional history alone is often enough to trigger a perceived-conflict disclosure requirement under many journal policies, independent of whether it would actually sway the review.
Canada’s Tri-Council Policy Statement on the ethical conduct of research involving humans (TCPS 2), Chapter 7, is one of the more explicit national research-ethics policies on this point: it frames conflict of interest as arising whenever a person or institution is placed in a “real, potential or perceived” conflict between their research-related duties and personal, institutional, or other interests, and directs that prospective research participants be informed of real, potential, or perceived conflicts so they can make an informed decision about participating — treating disclosure of the perceived category as equally necessary to an informed-consent process, not optional or lesser. See TCPS 2 (2022), Chapter 7.
Why a perceived conflict still has to be disclosed and managed
This is the point most COI training gets asked about directly, and the answer follows from how research-integrity frameworks define the underlying test in the first place. As CASRAI’s conflict of interest entry sets out, the operative question institutions, funders, and journals apply is not “did this actually bias the outcome,” but whether a reasonable person, aware of the relationship, could conclude it created a risk of undue influence. A perceived conflict satisfies that test by definition — the relationship is visible and plausible enough that a reasonable outside observer would question it — even where no actual bias occurred or ever would have. Three practical reasons institutions still require disclosure and often some form of management for purely perceived conflicts:
- Self-assessment isn’t a reliable filter. The whole reason disclosure-then-independent-review exists as a two-step process, rather than trusting individuals to self-certify their own objectivity, is that people are systematically poor judges of their own bias. A perceived conflict routed through the same review step lets someone other than the conflicted party make that call.
- Public and institutional trust depends on appearances, not just outcomes. A funder, journal readership, or research participant has no way to verify that an undisclosed relationship didn’t influence a result — only that the relationship existed and wasn’t disclosed. Managing perceived conflicts protects the credibility of the finding itself, independent of whether bias occurred.
- Non-disclosure of a known, disclosable relationship is treated as a violation in its own right. Per ICMJE guidance (reused from the verified position already cited on CASRAI’s conflict of interest page), a purposeful failure to disclose a relationship a form specifically asks about is its own form of misconduct, regardless of whether the underlying relationship actually affected the work. That standard applies the same way to a relationship a reasonable person would flag as merely perceived.
In practice, “management” of a confirmed-perceived-only conflict is often lighter than for an actual one — published disclosure alongside the article or decision is frequently sufficient, without the recusal, divestiture, or independent-monitoring steps a confirmed actual conflict may require. The disclosure obligation is what stays constant across the certainty axis; the intensity of the management response is what typically scales with it.
Axis two: financial, non-financial, personal, and institutional conflicts of interest
This is the axis most “what are the 4 types of conflict of interest” queries are actually asking about — it classifies conflicts by where the competing interest comes from, independent of how certain the resulting risk is.
- Financial. The competing interest is money, equity, or anything readily convertible to it — employment or consulting income, honoraria, stock or options, patents and royalties, paid expert testimony, or research funding and in-kind support from a party with a stake in the outcome. In the US, the Public Health Service’s financial conflict of interest (FCOI) regulation at 42 CFR Part 50, Subpart F defines a specific “significant financial interest” disclosure threshold rather than leaving it to judgment. Full treatment: CASRAI’s conflict of interest entry.
- Non-financial. The competing interest carries no direct monetary value but could still shape judgment — academic rivalry or competition toward the same finding, service on an advocacy body with a stake in a particular result, or a strongly held prior public position on the exact question under study. The ICMJE’s Uniform Disclosure Form asks authors, reviewers, and editors to report both financial and non-financial relationships side by side, on the same reasonable-observer standard.
- Personal. Close personal relationships — family, romantic partners, close friendships, or significant personal conflict/rivalry with another party to the decision. Different COI frameworks classify this differently: some, including the general definition on CASRAI’s conflict of interest page, treat personal relationships as a specific instance of the broader non-financial category rather than a fourth category in their own right. Other training materials and institutional policies (particularly ones written for a lay or all-staff audience rather than a research-specific one) list “personal” as a separate, fourth bucket precisely because relationship-based conflicts are common and specific enough to call out on their own rather than leave nested inside a broader “non-financial” label. Both framings describe the same underlying situations; treat “financial / non-financial / personal / institutional” as one common four-way presentation of a taxonomy that is substantively three-way (financial vs. non-financial vs. institutional, with personal relationships as the most frequent non-financial subtype) once the categories are checked against how funder and journal policy documents actually define them.
- Institutional. The institution itself — or a senior official acting through their institutional authority — holds a financial interest that could unduly affect, or appear to affect, decisions the institution is supposed to make impartially about research it oversees. This is conceptually distinct from every individual-level category above: an institutional conflict can exist even when no single investigator has a disclosable interest of their own, and resolving individual conflicts does not resolve an unaddressed institutional one. Full treatment, including the Institute of Medicine’s 2009 framework and worked examples, is on CASRAI’s conflict of interest page.
How the two axes combine: illustrative examples
The following are illustrative composites built to show how the certainty axis and the source axis intersect in practice — they are not real institutions, individuals, or documented cases.
| Source | Actual | Potential | Perceived |
|---|---|---|---|
| Financial | An investigator holds founder equity in the company whose device is the subject of the trial they are currently running. | An investigator is in early licensing talks with a company; no sponsored research relationship exists yet, but one is being discussed. | A peer reviewer held a small amount of stock in the study sponsor two years ago, sold it, and now reviews a manuscript from that sponsor. |
| Personal | A department chair is directly supervising and evaluating their own spouse’s grant-funded position. | A hiring committee member’s close friend has indicated they may apply to a not-yet-posted position on the committee’s panel. | Two co-investigators on a study previously co-authored several papers together and have an obvious long-standing professional friendship visible in their publication history. |
| Institutional | A university’s technology-transfer office holds equity in a spinout company and is simultaneously asked to approve a sponsored-research agreement between that spinout and a university lab. | A university is in early discussions to license a discovery to a company that might later fund additional research on the same technology at the same institution. | A university-affiliated foundation has received a large unrestricted gift from a company, and a researcher at that university later publishes findings favorable to that company’s product, even though no direct funding tie exists between the gift and that specific study. |
Frequently asked questions
What are the 4 types of conflict of interest?
Most commonly cited as financial, non-financial, personal, and institutional — classifying conflicts by where the competing interest comes from. Note that several research-specific frameworks (including CASRAI’s own conflict of interest definition) treat personal relationships as a subtype of the non-financial category rather than a fully separate fourth bucket, making the underlying taxonomy three-way rather than four-way depending on the source consulted. Either framing describes the same set of situations.
What is a perceived conflict of interest?
A situation where a reasonable, informed outside observer could conclude there is a risk of biased judgment, whether or not any actual bias exists. It is evaluated from the observer’s perspective, not the involved person’s own assessment of their objectivity, and most institutional and journal policies require it to be disclosed on the same footing as an actual conflict, even though the resulting management step is often lighter.
Is a perceived conflict of interest still a “real” conflict?
Yes, in the sense that matters for disclosure obligations. Research-integrity frameworks generally do not require proof of actual bias before treating a relationship as disclosable — the risk of undue influence, as a reasonable observer would assess it, is the operative test. See the section above on why perceived conflicts are still managed.
What’s the difference between a potential and a perceived conflict of interest?
A potential conflict is about timing: the competing interest exists, but the specific circumstances that would make it actually bear on a decision haven’t happened yet. A perceived conflict is about the observer’s viewpoint: it may already be fully “live” in the sense that the relationship and the decision coexist right now, but the question is whether a reasonable outside party would see a risk of bias, independent of whether one is actually present.
How is a conflict of interest different from a conflict of commitment?
A conflict of interest concerns bias risk — whether an outside interest could compromise objective judgment. A conflict of commitment concerns time and obligation — whether outside activities draw so much of a person’s professional time and effort that they compromise their primary institutional duties, independent of any bias question. See CASRAI’s Conflict of Commitment (COC) entry for the full distinction.
References
- Government of Canada, Panel on Research Ethics, Tri-Council Policy Statement: Ethical Conduct for Research Involving Humans – TCPS 2 (2022), Chapter 7, “Conflicts of Interest” (ethics.gc.ca)
- 42 CFR Part 50, Subpart F, “Promoting Objectivity in Research” (PHS financial conflict of interest regulations)
- ICMJE, “Disclosure of Financial and Non-Financial Relationships and Activities, and Conflicts of Interest”
- Institute of Medicine (National Academies), Conflict of Interest in Medical Research, Education, and Practice (2009)
