Editorial process in scientific journals: analysis and modeling

The editorial handling of papers in scientific journals as a human activity process is considered. Using recently proposed approaches of human dynamics theory we examine the probability distributions of random variables reflecting the temporal characteristics of studied processes. The first part of this paper contains our results of analysis of the real data about papers published in scientific journals. The second part is devoted to modeling of time-series connected with editorial work. The purpose of our work is to present new object that can be studied in terms of human dynamics theory and to corroborate the scientometrical application of the results obtained.

💡 Research Summary

The paper treats the editorial handling of scientific manuscripts as a human‑activity process and applies concepts from human dynamics theory to characterize its temporal patterns. The authors focus on the waiting time (t_w) between a manuscript’s receipt and its final acceptance (or the latest revision date when acceptance is missing) and investigate the statistical distribution of this variable across several journals.

Data were collected from publicly available web pages of three Elsevier journals—Physica A, Physica B, and Information Systems—as well as from the Ukrainian journal Condensed Matter Physics. The sample sizes range from 262 records (Condensed Matter Physics) to over 4,500 records (Physica A and B). For each paper the authors computed (t_w) and built histograms with 5‑day bins. Visual inspection of log–log and semi‑log plots revealed a single, asymmetric peak followed by a long tail, suggesting non‑Poisson behavior.

Two candidate models were fitted to the empirical distributions: (1) a log‑normal distribution, parameterized by a location (t_c) and scale (\omega); and (2) a power‑law with an exponential cutoff, (P(t_w)=A,t_w^{-\alpha}\exp(-t_w/t_0)), with exponents (\alpha=1) and (\alpha=3/2). Non‑linear least‑squares fitting yielded adjusted coefficients of determination (\bar R^2) close to 0.97 for the log‑normal fit of Physica A and 0.95 for the power‑law with (\alpha=1). Similar performance was observed for the other journals; the (\alpha=3/2) variant performed slightly worse (≈0.92). Importantly, both families share a leading tail behavior (P(t_w)\sim t_w^{-1}), a hallmark of “critical” or “super‑critical” queueing regimes where the traffic intensity (\rho=\lambda/\mu>1).

To explore the mechanistic origin of the observed tail, the authors constructed a minimalist simulation that omits the peer‑review stage. Manuscripts arrive at a constant rate, and an editorial board meets periodically (interval (T), e.g., 15 days for Physica A) to decide on acceptance or rejection. The issue size is limited by a fixed number of papers per issue. By varying the input rate (\lambda) relative to the service capacity (\mu), three regimes are examined: sub‑critical ((\rho<1)), critical ((\rho=1)), and super‑critical ((\rho>1)). The simulated waiting‑time distributions are either sharply truncated or nearly uniform, lacking the heavy tail seen in the empirical data. This contrast indicates that the peer‑review process—characterized by author revisions, reviewer delays, and priority‑based decision making—is the key driver of the long‑tailed waiting‑time statistics.

The study concludes that (i) editorial processing can be fruitfully analyzed within the framework of human dynamics; (ii) both log‑normal and cutoff‑power‑law models adequately capture the empirical waiting‑time distributions, with the tail consistently following a (t^{-1}) scaling; and (iii) the peer‑review stage is essential for generating the observed non‑Poisson behavior. Limitations include incomplete data (rejected manuscripts are not publicly recorded), occasional inconsistencies in date fields, and the relatively narrow set of journals examined. Future work should broaden the dataset, incorporate rejection statistics, and refine the simulation to include more realistic reviewer‑author interactions.

Overall, the paper provides a quantitative baseline for assessing editorial efficiency, suggests statistical tools for scientometric monitoring, and opens a new avenue for applying human dynamics theory to scholarly communication processes.