Do senior authors get listed for reviewing rather than doing?
Written against the design standard.
1. The question
Honorary authorship has been studied mostly by asking authors. The CRediT taxonomy now records, in machine-readable form, what each author says they did. An author whose only declared role is “Writing: review and editing” appears on the byline without a declared hand in the work. Whether that pattern concentrates among senior authors is a direct, behavioral test of the gift authorship hypothesis, using the authors’ own declarations.
2. Prior work, and the search behind the novelty claim
Log key P16.
corpus regex: CRediT|contributor roles? taxonomy
PubMed: ("CRediT"[tiab] OR "contributor roles taxonomy"[tiab]
OR "contributor role taxonomy"[tiab]) AND (authorship[tiab]
OR "gift authorship"[tiab] OR "honorary authorship"[tiab] OR seniority[tiab])Two hundred and eighty PubMed records, twenty-six in the corpus. The relevance-ordered top ten are about authorship and credit in general, such as the gendered nature of authorship (PMID 34516891), shared first authorship (PMID 31607826) and authorship and contributorship (PMID 24644440). None in the top ten relates declared roles to author seniority. This is the weakest novelty claim of this set and says so: the space is large and the top ten is not the whole of it. Probe PD widens the search before any data are retrieved.
3. Hypothesis
The probability that an author’s declared roles contain no substantive contribution rises with the author’s seniority, conditional on author position and team size.
Null. The coefficient on seniority is 0.
What would falsify it. A coefficient at or below 0.
4. Design and unit
Cross-sectional, author-level. One author on one article is one unit; articles and authors both recur, and the model treats them as crossed.
5. Sample
PMC open-access research articles, 2020 to 2025, whose JATS carries role elements with the CRediT vocabulary for every listed author. Articles where only some authors have roles are excluded and counted, because a missing role cannot be read as no contribution.
6. Exposure
Seniority at the article’s publication date: years since the author’s first indexed publication in OpenAlex. OpenAlex author disambiguation merges and splits people; section 10 carries this.
7. Outcome
Minimal role set: every declared role is in {Writing: review and editing, Funding acquisition, Resources}. Every other CRediT role, including Supervision, counts as substantive. Counting Supervision as substantive works against the hypothesis, because supervision is a legitimately senior role; the choice is made so that a positive result cannot be produced by it.
Roles are read from JATS role elements and their vocab-term attributes, not from free-text contribution statements (S2).
8. Validation gate
The author checks role extraction against the article’s own contribution section for 100 articles. Extraction must match the article on at least 98 percent of author-role pairs; this is parsing, not judgment, and should be near perfect.
9. Analysis
In R.
glmer(minimal ~ ns(seniority_years, 3) + position + log(team_size)
+ (1 | article) + (1 | author), family = binomial, data = d)Position is first, last, or middle. Preregistered alongside: the same model restricted to middle authors, where position carries least information, and a sensitivity analysis dropping authors whose OpenAlex record implies a career longer than 50 years at the article date, which marks records that probably merge more than one person.
10. Threats
| threat | handling | removed or disclosed | standard |
|---|---|---|---|
| OpenAlex author disambiguation | Implausible careers excluded in sensitivity; merged identities bias seniority upward for common names | Disclosed | S3 |
| Journals that require CRediT differ from those that do not | Result generalizes to CRediT-declaring articles only | Disclosed | |
| Self-reported roles can themselves be generous | Biases toward the null for the hypothesis, since inflated roles hide minimal ones | Disclosed | |
| Team size and seniority are correlated | Team size in the model | Removed |
11. Probes
| probe | computes | goes to | what would change the design | status |
|---|---|---|---|---|
| PA | PMC articles with complete CRediT roles, per year | 5 | fewer than 5000 articles narrows the window | not run |
| PB | OpenAlex author match rate for those authors | 6 | under 90 percent matched makes the exposure unreliable | not run |
| PC | the author’s 100 extraction checks | 8 | under 98 percent the parser is fixed first | needs the author |
| PD | a second, wider novelty search on author seniority and contribution statements | 2 | an existing study of this relation makes this a replication, labeled as one | not run |
12. What this cannot show
- Whether any given author deserved authorship.
- Anything about articles that do not declare roles, which are most of the literature.