There seems to be a conceptual error in the acceptance- (also decline-) rate calculation in OJS.
In my opinion (and understanding what an "acceptance rate" should measure), when rejecting an article, the acceptance rate should go DOWN. However, I noted that if I reject an article, the acceptance rate doesn't change at all!
Why is this? It's because in OJS (I saw this in JournalStatisticsDAO.inc.php) the acceptance rate is defined as submissionsAccept / numSubmissions.
I.e. the denominator includes ALL submitted manuscripts, even those where no final decision has been made yet. Thus, rejecting an article keeps the acceptance rate unchanged, which is counterintuitive to most people.
The common understanding of the acceptance rate is that as denominator you count only those articles where a final decision reject or accept has been made by the editor, i.e. submissionsAccept / (submissionsAccept + submissionsDecline).
The difference between the two definitions is substantial for a journal where we have a substantial (and changing) backlog of undecided articles / articles under revision. The acceptance rate is then mainly affected by the size of the backlog / new submissions rate, rather than the editor actions.
The reason why this is so important is that the acceptance rate is commonly seen as a metric which tells authors how selective the journal is, and also as an indirect measure for the journals' quality (low acceptance rate = better (as more selective) journal).
However, if the acceptance rate is measured the way OJS defines it, a low acceptance rate is potentially an indication of a long backlog of undecided articles, thus the exact opposite of a quality criterion! If you want to know how many are undecided, then look at the undecided rate, but don't let the number of undecided articles confound the acceptance or rejection rate calculations.
I welcome some discussion on this, but in my opinion this should be fixed. As a minimum, people (authors, editors) must be warned that the OJS acceptance rate is calculated the way it is, which is different from how most journals would calculate their acceptance rate.