Replication Rate by Journal OpenAlex 2-Year Mean Citedness
OpenAlex 2-year mean citedness is a single current snapshot from the OpenAlex API (as of 2026), so it does not vary by publication year; the basis is fixed to that snapshot.
Rate = success / (success + failure + reversal); inconclusive and unrecorded outcomes excluded, where success means the outcome recorded in the database. Unit: replication effect (n = 6,931 — originals whose journal is matched in the impact-factor dataset). How this is defined.
4,462 papers plotted · 456 excluded (no recent impact factor) · 838 excluded (inconclusive outcome)
Replication rate by OpenAlex 2-year mean citedness quintile
All papers with a publication-year impact factor, split into five equal-count bins. Hover a bar for its journal-cluster bootstrap 95% interval.
Replication rate by OpenAlex 2-year mean citedness
The same papers as above, grouped into equal-width impact-factor intervals (the top bin is a catch-all for IF ≥ 16). Unlike the equal-count quintiles, this shows the trend against a true impact-factor scale; most papers fall in the low bins.
Per-journal replication rate vs OpenAlex 2-year mean citedness
One point per journal with at least 15 plotted papers. Point size reflects the number of papers; whiskers are 95% Wilson CIs.
Metric sources. The 2-year and 5-year impact factors and the CiteScore-style 4-year metric are self-computed from the SciSciNet-v2 / OpenAlex citation graph — not the official Clarivate Journal Impact Factor™ or Scopus CiteScore, which are proprietary. They differ from the vendor figures in two ways that pull in opposite directions. Their denominator counts every document a venue publishes, where the real JIF counts only “citable items” (articles and reviews), so journals carrying heavy front matter or conference abstracts are pushed down sharply — the effect is severe enough that Cancer Research, which publishes thousands of meeting abstracts, lands near 0.6. Their numerator, meanwhile, draws on a broader citation graph than Web of Science, which pushes figures up. Deflation usually wins: against the citable-denominator metric below, 85% of journals here come out lower (median ratio 0.80), but 12% come out higher. Treat them as internally consistent relative indicators rather than as JIF estimates, and prefer the citable-denominator metric when comparing across journals of different types. OpenAlex 2-year mean citedness is OpenAlex’s own published impact-factor-like metric, fetched from the OpenAlex API as a single current snapshot (it does not vary by year). SCImago SJR is a prestige-weighted metric (citations weighted by the citing journal’s standing) taken per publication year from SCImago; papers before 1999 use the 1999 ranking. Citations per citable document (SCImago, 3-yr) is a CiteScore-style ratio — SCImago’s Total Citations (3 years) divided by its Citable Docs (3 years) — per publication year (pre-1999 uses 1999). Because its denominator counts only citabledocuments (excluding news, editorials and other front-matter), it is normalized like the real JIF/CiteScore and, unlike the self-computed IFs above, does not deflate high-front-matter journals such as Nature and Science.
Matching. Journal names in the replications database are matched to the impact-factor dataset by normalized name; SCImago SJR is joined by ISSN with a name fallback (~94% of matched journals). Books, preprints, and conference proceedings have no journal metric and are excluded.
Attribution. SJR data: SCImago, (n.d.). SJR — SCImago Journal & Country Rank [Portal], retrieved from scimagojr.com (used non-commercially). OpenAlex data is CC0.
Data: replications_database_2026_08_01_120356.csv; metrics generated 2026-08-03.