By transparent peer review in this context, we mean that authors are "opted-in" for transparent peer review unless they opt-out. If authors opt for transparency and their article is published, then the peer reviewers’ reports, authors’ responses, and editors’ decisions will accompany the published article. Reviewers also have the option to disclose their names alongside their reports but this is not mandatory.  The peer review history is freely available via a page hosted by Publons linked from the published article, for example, \citep{chen} and this also allows engagement with the peer review process to be measured.  Each component of the peer review history has a DOI, ensuring each element is fully citable. For those reviewers who choose to sign their reviews, the DOIs can also be added to their ORCID records. The layout of the peer review page reflects the sequence of peer review from original submission, initial peer review, revision, re-review to acceptance of the version of record. Furthermore, each peer review element can be endorsed by, or commented on, by Publons users.
The first journal to join the pilot was Clinical Genetics \cite{graf2018}. Since then, the initiative has grown steadily at Wiley, adding a further 10 journals in January 2019  \citep{moylan2019} and reaching a total of 40 journals for Peer Review Week, September 2019 \cite{graf2019}. This September 2020, the theme of Peer Review Week is 'Trust' \cite{meadows2020}. Transparent peer review is, of course, a significant step towards bringing greater trust to research publishing and recognition to all involved in preparing and disseminating research. Making the peer review history visible and discoverable is key to improving the quality and efficiency of research communication and improving researchers' trust in a journal's processes while also potentially reducing fraud in peer review.
65 Wiley journals are now offering transparent peer review across a range of disciplines. Here we reflect on what we have learnt during the past two years and share our findings. We present data on the transparent peer review pilot from 27 journals across a range of subject disciplines that have participated for a sufficient length of time to enable robust data collection. In order to minimize bias we assessed similar data on a set of 29 comparable 'control' journals that did not introduce transparent peer review across the same time period. We wanted to understand the effect of introducing transparent peer review to a journal in terms of turnaround times and willingness of reviewers to agree to review.

Methods

Data from 27 journals across a range of subject disciplines (earth and environmental sciences; life sciences; medicine; nursing, dentistry and healthcare; physical sciences and engineering; social and behavioural sciences and veterinary medicine) that have been participating in the transparent peer review pilot were compared to data from 29 'control' journals that did not introduce transparent peer review. The control journals were sufficiently similar in terms of subject discipline, impact factor, number of submissions and number of publications that no significant differences were detected by using t-tests and Chi2 tests 
We evaluated data from the journals that joined the pilot pre- and post- the introduction of transparent peer review and compared this to data from the control journals for the same time period (1st September 2018 to 31st March 2020). The majority of journals implementing transparent peer review began doing so in September 2019 .   Articles in the control  group were randomly assigned "pre- or post- transparent peer review status" based on the month of their publication and the distribution of "pre- or post- transparent peer review" status in the test group during that month. We use the annotation 'control journal - pre-TPR' and 'control journal - post-TPR' when comparing data from the control journals  pre- or post- the introduction of transparent peer review with those journals that went on to adopt the model. Note that none of the control journals actually introduced transparent peer review. For inclusion in the study, articles from both groups of journals must have reached a final decision.   Data from a total of 74160 articles was analysed.

Results

Information on the number of articles included in the study from journals that adopted transparent peer review and the control journals that did not is summarised in Table 1.