Every researcher develops, over the course of a career, a small habit of reaching for their own earlier work. At first, the reaching is innocent. You wrote a paper in 2019 on a topic closely adjacent to what you are writing now, and it makes sense to cite yourself. The prior paper established a claim your current paper depends on, described a method you are reusing, or made an argument you are extending. No serious person objects to citations of this kind. They are the machinery of a coherent research program.
But the habit grows. Somewhere along the way, most researchers reach a point where they begin to cite themselves when the citation is not strictly necessary. The 2019 paper shows up in the literature review of a paper where it is, at best, tangentially related. The method description in 2023 refers back to the method description in 2021, which refers back to the method description in 2019, forming a chain of self-citations that a careful reader would find odd. The researcher is not being dishonest. The papers are theirs; the connections are real. But the reaching has become something more than the argument requires.
There is nothing unusual about this, and it is not strictly wrong. Most researchers self-cite at rates between 5% and 20% of their citations in any given paper. In some fields, where research programs are tightly bounded and build cumulatively on the author’s own earlier findings, higher rates are defensible. Economists and formal theorists often self-cite more than biologists, and this reflects something real about how the disciplines work, not a character flaw in economists.
But there is a version of the habit that has shaded into something else, and it is worth being specific about what that something else is.
The most honest way to describe it is this: self-citation is, in practice, a small act of self-promotion. Every citation to your own work is an impression in someone’s reading queue. The reader clicks through, files the reference, or at least notices the name. Citations are the coin of academic visibility, and the one person who always benefits from a citation to you is you. When an author reaches for their own work, they are reaching, at least partly and at least sometimes, because reaching benefits them. This is not necessarily a bad thing. It is a fact about the structure of academic incentives, and to pretend otherwise is to miss what is actually happening in citation decisions.
The question is not whether self-citation is motivated in part by self-interest. It is. The question is whether the self-interest is disciplined by the standard that applies to any citation: does this reference help my reader understand the current paper? If yes, cite. If no, do not cite, even if you want to.
This simple test is harder to apply than it sounds. The reason is that we are rarely fair judges of our own prior work. Of course our 2019 paper helps the reader. Of course our 2021 methods description is clearer than anyone else’s. Our prior work seems to us more relevant than equivalent work by others because we know our prior work better. This is not arrogance. It is a feature of how memory and relevance interact. What you know best feels most pertinent. A fair test for self-citation requires holding yourself to a standard you would not naturally apply: would I cite this paper if someone else had written it?
Apply this test honestly to your own draft, and most researchers find that a portion of their self-citations, perhaps a third or perhaps half, depending on the paper, would not survive. The 2019 paper is cited because it is yours, not because the reader needs it. You can remove the citation, and the paper is no weaker. You can keep it, and the paper is no stronger, only slightly more visible to anyone who eventually reads it.
There is a second, less-discussed dimension to this. Fields have begun to develop informal norms around what counts as excessive self-citation, and those norms are not always benign. In some quarters, a paper with 15% self-citation is read as serious and programmatic, as evidence of a researcher building on their own prior work in a coherent direction. In other quarters, the same paper is read as self-aggrandizing, as evidence of a researcher padding their citation count. The judgments are not symmetric across subfields, institutions, or career stages. A senior faculty member with 15% self-citation is usually read as building a research program. A postdoc with the same rate is sometimes read as grasping. The norms are uneven, and they disadvantage people who are already in less secure positions.
This creates a practical dilemma. A researcher who self-cites at a rate the discipline considers normal is fine. A researcher who self-cites below that rate, because they are scrupulous or because they hold themselves to the fairness test above, is competing for visibility with researchers who are not as scrupulous, and losing. The scrupulous researcher’s citation count grows more slowly. In a system that rewards citation counts, the scrupulous are penalized for their scruples.
I do not have a clean answer to this. It is one of the several places where the structure of academic rewards runs against the standards the profession publicly endorses. The editors of this Almanac are not naive about the conflict. We are also not willing to recommend that researchers cynically maximize their self-citation rate because the system rewards it. That way lies a particular kind of hollowing out.
What I can offer is a practical procedure, which is the closest thing to an ethical answer available under the current incentives.
First, when you draft a paper, cite whatever prior work, yours or anyone else’s, genuinely helps your reader follow the argument. Draft honestly. Do not preempt your own generosity toward other people’s work, and do not preempt your own honesty about whether your prior paper is actually useful to the reader.
Second, before submission, read your citation list with the fairness test in mind. For each self-citation, ask: would I cite this paper if someone else had written it? If the answer is no, remove the citation. Do this with each self-citation in isolation. Do not bargain by saying, “well, I’ve only self-cited three times in this paper, I can afford one more.” That is a political calculation, not a citation decision.
Third, if the test leaves you with a lower self-citation rate than you would like to see, accept that. The reward for doing this well is a paper that a careful reader will trust more, and a citation list that holds up on rereading. These are small, slow rewards. They are real.
Fourth, be especially careful about self-citing in the introduction and the literature review. These are the places where citations shape how the reader frames your paper, and where you claim your paper’s location in an ongoing conversation. Self-citations here are the most consequential and the most visible. Make sure every one of them earns its place.
There is a further temptation, which is to ask coauthors, collaborators, and former students to cite you. This is a different ethical territory, and most researchers sense correctly that it is worse than excessive self-citation. Asking others to cite you converts a personal vanity into a social obligation, and the obligation creates small resentments that compound. The best researchers I have known do not ask. They write papers that are worth citing, and they let the citations come, or not, on their own.
A closing observation. The self-citation habit is, at its root, an anxiety habit. It is a form of self-assertion in a profession that often fails to recognize its members, a small way of making sure your name appears in the footnotes of the field. The cure is not to pretend the anxiety is illegitimate. The cure is to notice the anxiety, to feel it, and then to write the paper as if you did not need it. Over time, this gets easier. Researchers who are further along in their careers self-cite somewhat less, on average, than researchers who are earlier. This is not because the older ones have transcended ambition, but because they have been cited enough times by others that they no longer need to cite themselves as urgently. The point is not to wait until you reach that stage. The point is to behave now as you will behave then.
Write the paper as if your citation count does not depend on it. Mostly, it does not.