First response time
For authors, the duration of the first review round, or first response time, is probably the factor they are mostly interested in, as this takes up a substantial part of the total manuscript evaluation time and to a large extent determines how much time is lost if the outcome is negative. First response time includes the time taken by the journal for a first evaluation of the manuscript, finding reviewers, the time the latter require to do their work, and the time the editor then requires to evaluate the manuscript in light of the referee reports and to inform authors about the decision.
As can be seen in Table 1, the reported first response time in the SciRev data is on average 13 weeks and varies considerably among scientific fields. It took 8–9 weeks in Medicine and Public health related journals, 11 weeks in Natural sciences and General journals, 14 in Psychology, and 16–18 weeks in Social sciences, Humanities, Mathematics and Computer sciences, and Economics and Business. These figures differ between accepted and rejected manuscripts, with first response time of rejected manuscripts taking, on average, 4 weeks longer.
While writing a peer review may take between 4 and 8 h, in only 19% of all reported cases authors were informed about the outcome in less than a month. In about one third of the cases (32%) authors had to wait 3 months or more and in 10% of the cases even more than 6 months before being informed. Duration differs widely between scientific fields. In Social sciences and Humanities, only 7–8% of the authors were informed within 1 month versus 25% in Natural sciences and 27–28% in Medicine and Public health. In Economics and Business and Mathematics and Computer sciences over one sixth (18%) of authors had to wait 6 months or longer.
It is yet unclear to what extent the long duration of the first review round is the result of the peer review process as such and to what extent it is due to (in)efficient manuscript handling at editorial offices. Given that immediate rejection times are often long (see Table 3 and its discussion below), it seems that inefficiencies at editorial offices also play an important role. The finding that in Medicine and Public Health—where professionalization of journals is relatively high—first response times are the shortest, also points in this direction.
To test this idea further, we looked at the relationship between the journal’s impact factor and first response time. As highly ranked journals generally have more resources at their disposal and thus probably better organized editorial offices, and as reviewers are more motivated to review for those journals, we expected to find a negative relationship. Pearson correlations between first response time and impact factor indeed confirm this expectation. These correlations are significantly negative for all scientific fields combined (P = −0.29) as well as for all scientific fields separately, with General journals (P = −0.51), Mathematics and Computer sciences (P = −0.27), and Natural sciences (P = −0.26) having the highest correlations. The only exception was Humanities, where no significant correlation between first response time and impact factor was found. This might be because this field traditionally values publishing books more than publishing in journals (Ware and Mabe 2015).
Total review duration
Total review duration refers to the time a manuscript is under responsibility of the journal. Besides by the duration of the first review round, total review duration is also determined by the number and duration of subsequent review rounds. Total review duration does not include the time taken by authors to revise and resubmit their manuscript. Given that rejected manuscripts have on average less review rounds, we restrict this analysis to accepted papers.
Table 2 shows that the reported total review duration of accepted manuscripts is on average 17 weeks. Again there are substantial differences between scientific fields. With 12–14 weeks, average total review duration is shortest in Medicine, Public health, and the Natural sciences. It is longest in Economics and Business, where the process takes on average 25 weeks and is twice as long. In Mathematics and Computer sciences, Social sciences and Humanities, total review duration is also long, i.e., 22–23 weeks. Hence the differences in the duration of the review processes we observed for the first review round are also present in the other aspects of the process.
If we split out the data further, we note that in Natural sciences, Medicine, and Public health 13–16% of the manuscripts pass through the entire peer review process within 1 month, that this applies to about two thirds of the manuscripts after 3 months, and to 87–92% of the manuscripts after 6 months. In Mathematics and Computer sciences, Social sciences, and Humanities, these figures are 3–4%, one third and slightly above two thirds, respectively. Whereas only 8% of the authors in Medicine had to wait more than 6 months, this applies to one third of authors in Social sciences and Economics and Business.
The total time a manuscript is with the journal is determined by the time a journal takes for a review round and by the number of review rounds. As mentioned in the Background-section, there are indications that the number of review rounds has increased in recent years. In our data, the number of review rounds on average amounts to 2.03, with Psychology (2.23), General journals (2.18), Economics and Business (2.16), and Social sciences (2.15) showing a higher average number of review rounds.
Total review duration correlates significantly and negatively (−0.27) with a journal’s impact factor, thus indicating that total review duration is shorter for higher impact factor journals.
Immediate (desk) rejection time
Immediate rejection time is the time an editor takes to inform authors that he or she is not interested in the manuscript (and will therefore not send it to reviewers). Our figures clearly show that immediate rejection time is a major source of unnecessary time loss in the peer review process (Table 3). On average, an immediate rejection in Medicine takes 10 days, closely followed by Natural sciences, Public health, and Engineering, taking 11–12 days. Journals in Psychology, Social sciences and Mathematics and Computer sciences take half as long, i.e., 15–17 days. These are relatively high averages, given that in many cases an inspection of the abstract is sufficient to decide that a paper does not fit.
On the positive side, in half (50%) of the reported immediate rejection cases, the editor informed the author(s) within 1 week. However, the data also show that in 17% of cases authors had to wait more than 4 weeks to be informed of the rejection. Several authors even had to wait for more than 3 months, or withdrew their manuscripts after hearing nothing for an even longer period. These are clearly unacceptable practices.
The situation is best in Medicine, where 62% of authors are informed about an immediate rejection within 7 days, followed by Natural sciences and Public health where this figure is 54%. Immediate rejection time is longest for authors in the Social sciences and Mathematics and Computer sciences, where in about 30% of reported cases it took the editor 4 weeks or more to inform author(s) that he or she was not interested in the manuscript and would not to send it to reviewers. There is a significant negative correlation (−0.18) between immediate rejection time and the journal’s impact factor, which indicates that journals with a higher impact factor have editors who work faster and editorial offices that are more professionally organized.
Reviewers are generally blamed for long processing times, but our findings indicate that manuscript handling at editorial offices plays an important role too. If editors take a month for an immediate rejection decision, they are probably also slow in finding reviewers and processing referee reports.
The average number of referee reports is about 2.2 in all scientific fields (see Table 4). This correspondence is remarkable, given the substantial differences between fields in other respects. There is slight variation in the experienced quality of the referee reports between the fields [as indicated on a scale running from 0 (very bad) to 5 (excellent)]. Authors report the quality of the reports to be somewhat higher in Natural sciences, Engineering, and Public health (3.7), and lower in General journals, Psychology, and Economics and Business (3.4). It is interesting that the long review duration in Economics and Business did not translate into referee reports experienced of higher quality.
Authors who were given the opportunity to revise and resubmit their papers were also asked to what extent they perceived the requested changes as difficult and whether they thought their manuscript had improved as a result of the revision. There is a significant positive correlation (0.40) between these factors. When the revision was experienced as more difficult, authors were also more satisfied with the improvement. Regarding the difficulty experienced, revision processes were perceived as easiest in Mathematics and Computer sciences and in Public health (2.6), and as most difficult in Economics and Business (3.3). Regarding the experienced improvement of the manuscript as a result of the revision, authors from Social sciences, Economics and Business, and Humanities reported somewhat higher figures (3.8 and 3.9) compared to the other scientific fields (3.7).
There is a small positive correlation (0.07) between the difficulty experienced regarding the referee reports and the impact factor of the journal. Thus, reviewers of more highly ranked journals tend to make somewhat greater demands on the authors. The degree of improvement experienced regarding the manuscript is not significantly related to impact factor.
The time from the first submission date to the final decision date is not only influenced by the time the manuscript is at the editorial office or being reviewed, but also by the time authors take to revise their manuscript. It is therefore important to look also at the duration of the revision time. Table 5 shows that authors who received a revise-and-resubmit on average take 39 days to revise their manuscript, but there is substantial variation among the fields. Authors in Economics and Business take longest to revise their manuscripts: on average 64 days to prepare and submit a revised version. This is substantially longer than authors in Natural sciences, Engineering and Mathematics and Computer sciences (32–34 days) and in Public health (29 days). Apparently, in Economics and Business it is not only the editors who take more time.
Table 5 also shows the percentage of manuscripts revised within a specific number of days. While 18% of authors in Engineering, Mathematics and Computer sciences and Public health revise their manuscript within 7 days, this applies to 9–10% of authors in Social sciences and Humanities and only 3% of authors in Economics and Business.
Regarding the relationship between the journal’s impact factor and the time authors take to revise their manuscript, we expected authors who received a revise-and-resubmit from a high-level journal to be more motivated to complete the revision of their manuscript quickly. However, no significant correlation was found between revision time and the journal’s impact factor.
Rating of peer review experience
The SciRev questionnaire gives authors the opportunity to provide an overall rating of the review experience on a scale from 0 (very bad) to 5 (excellent); see Table 6 for details. Authors of accepted manuscripts give the peer review process a much higher rating (4) than authors of rejected manuscripts (2.2). Moreover, the rating of the peer review process is negatively related to total review duration. This correlation is −0.43 for both accepted and rejected manuscripts.
To determine how the various factors might affect the satisfaction of authors with the peer review process, we turn to the results of multivariate analyses (see Table 7).
The first columns show the results of Model 1, which contains all relevant variables. Model 2 contains the same variables but also the significant interactions between the variables.
As can be seen in Model 1, all variables, except impact factor, are significantly related to authors’ rating of the peer review process of their manuscript. As expected, authors of accepted manuscripts rate the process significantly more positive than authors of rejected manuscripts. Authors tend to suffer from attributional bias: if their paper is rejected, they often blame this on situational factors such as incompetent reviewers and uninterested editors; but if it is accepted they tend to attribute this to their own expertise and competence in writing high-quality papers (Garcia et al. 2016).
Authors also value speed of the peer review process. When the duration of the first review round is shorter and there are fewer review rounds, authors give the process a significantly higher rating. Authors who receive more referee reports also tend to be more positive about the process. Their perception might be that their manuscript has been dealt with more seriously and thoroughly. Authors from countries where English is the first language rate the peer review process less positive than authors from other countries. It is possible that these authors have higher expectations of the process and are more critical regarding aspects that do not meet their expectations.
Taking into account other factors, authors in Economics and Business, Social sciences, Psychology, and Mathematics and Computer sciences are more positive about the peer review process than authors in Natural sciences, Medicine, Public health, and especially General journals.
When we include the significant interactions in the model (Model 2), the sign and significance of the main effects stay the same. The interaction analysis shows that the negative effect of a longer duration of the first review round and the negative effect of more review rounds are less profound for accepted papers. Hence it seems that authors are willing to accept extensive revision work if this is rewarded with the acceptance of their paper. At the same time, they seem especially disappointed if the manuscript is still rejected after a long review process.
The negative interaction between a paper being accepted and the number of referee reports indicates that authors of rejected papers may consider a higher number of reports as a sign that their paper was taken seriously and might be content with extensive feedback. For obvious reasons, authors of accepted papers are more positive when the journal has a higher impact factor. Authors from English-speaking countries are less negative about the peer review process when their paper is accepted and when they receive more referee reports but find a long process more problematic. This might reflect that they have higher expectations that their paper will be accepted and that the peer review process will be short and efficient compared to authors from non-English-speaking countries.
When the duration of the first review round is longer, or when the impact factor of the journal is higher, authors are more concerned about a higher number of review rounds. In those cases, they might expect a smooth continuation of the process and be more disappointed when this proves not to be the case. A longer duration of the first review round is considered less negative by authors who receive more referee reports.