March 2009 Issue • Volume 37 • Issue 3

to print a pagePrint This Page

Are Journal Accept Rates
as Low as They Look?

Paula England, Stanford University, and former editor of ASR (1994–1996)

Authors aspiring to publish in a sociology journal typically understand that, in the best case, an article gets accepted only after an invitation to revise and to resubmit (an R&R). They often want to know the probability that an author sending an article to this journal will eventually get it accepted by this journal. But, oddly enough, this is not what ASA journals' "accept rates," previously published annually in Footnotes but now online, tell us.

Here is how ASA (and some other scholarly journals) compute their accept rates. The basic concept is to take acceptances during the year as a ratio of all decisions—positive and negative—made in the year. ASA puts all decisions in the denominator, including accepts, rejections, conditional accepts, and invitations to revise and resubmit. In effect, original submissions and revisions (after an R&R or conditional accept) count as separate manuscripts for purposes of the accept rate. A manuscript that ultimately gets accepted counts twice—as one accept and one nonaccept. If we want the accept rate to answer the question I posed above, a better procedure would be to only put final decisions in the denominator—accepts and rejects decided during the year. Thus, every paper would only enter the statistics once, counting as an acceptance regardless of how many revisions it went through or a reject if it was ultimately rejected, either originally or after a revision.

For Example

Consider the following hypothetical—a journal in which all papers submitted are eventually accepted, but every paper goes through one R&R decision on the way. An author submitting would know her or his paper was sure to be accepted eventually, so calling the accept rate 100% makes sense in this scenario, and this is what we would get if only final decisions were in the denominator. However, the way ASA calculates its journals' accept rates, the rate is only 50% despite the fact that every paper is ultimately accepted. If every paper required one R&R and one conditional accept, the rate would drop to 33%. Thus, under the present way of calculating rates, differences across editors within a journal, between journals, or between disciplines may be affected by how many revisions editors typically require before acceptance.

How much difference
would it really make if only
final decisions were put in
the denominator?

Clearly accept rates would be higher if only final decisions were in the denominator (the numerator is the same under either system). How much difference would it really make if only final decisions were put in the denominator? To find out I asked the editors of two ASA journals and the journal of Sociologists for Women in Society to share their 2008 statistics with me so I could see what differences it makes to calculate accept rates with only final decisions included in the denominator. (Thanks to Randy Hodson and Vincent Roscigno, editors of the American Sociological Review (ASR); Gary Alan Fine, editor of Social Psychology Quarterly; and Dana Britton, editor of Gender & Society, for the data from which I calculated the numbers.) In 2008, ASR’s official rate was 8.25%, calculated using ASA’s method, with a denominator including final accepts and rejects, as well as the intermediate decisions allowing revision. If the denominator had included only final decisions, the accept rate would have been 11.42%. The second rate is 38% higher than the first (the difference between the two over 8.25 is .38). Similar computations for Social Psychology Quarterly show that their official 2008 accept rate of 9.43% would be 15.96% if only final decisions were in the denominator, a 69% increase. If I apply the ASA method to Gender & Society statistics, its accept rate would be 9.67%; with only final decisions in the denominator, it is 11.88%, which is 23% higher.

Arguments For and Against

An argument sometimes made for the status quo is that, when trying to convince an interdisciplinary tenure and promotion committee that a colleague has published in very selective journals, the lower the rate the more useful for the case. However, even the more realistic accept rates that I calculated above using only final decisions as the base show that our journals are extremely selective. A downside of the current system is that it gives authors an unrealistically low idea of their chances that their paper will ultimately be accepted by a journal. Moreover, the rate as now calculated is reduced when editors increase the typical number of revisions required before papers are ultimately accepted, even if the probability of eventual acceptance does not change. I suggest that we change how ASA calculates accept rates, taking a given year’s number of accepts as a percent of all final decisions made that year (accepts and rejects). logo_small


Back to Front Page of Footnotes