Monday, July 27, 2015

Median publication delays at 38 APA journals

Last week, I had a paper accepted at the Journal of Personality and Social Psychology.  This acceptance is good for me, as JPSP is one of the more prestigious journals in my field.  However, given how grueling the review process has been, it's hard for me to feel happy about this acceptance -- based on my records, this paper spent about 17 months in review, and right now it has only been accepted, not published.

Of course, I am far from the only person whose paper has spent a long time in the limbo between acceptance and publication.  In fact, based on two analyses of papers in PubMed, this experience seems distressingly common.  For example, Steve Royle found that papers submitted to cell biology journals in 2013 and indexed by PubMed take about 100 days to go from received to accepted and another 120 days to go from accepted to published, for a total of 220 days.  In another analysis, Daniel Himmelstein analyzed the time between acceptance and publication for 3,476 journals indexed by PubMed in 2014.  I didn't see an overall median lag time, but most of the lags seem to be between 50 and 60 days.

Both of these analyses focus primarily on biology journals, and primarily on journals indexed by PubMed.  For example, if you try search for "J Pers Soc Psychol", the PubMed abbreviation for JPSP, on the Himmelstein site, you will not find this journal listed -- possibly because JPSP does not report the receipt and acceptance dates for each article in PubMed.  This leads me to my question: Do the Royle and Himmelstein analyses reflect the typical delays at psychology journals?

If my experience with my most recent paper is typical, maybe not.  In her comment on Himmelstein's analysis, social psychologist Liz Page-Gould seems to share my intuition:


Intuition, however, is not the same as evidence.  I therefore decided to find some hard data on time to publication myself.

I did a search in the primary psychology database, PsycINFO, for all papers published by the American Psychological Association between 2009 and 2014.  I limited my search to APA journals for three reasons:
  1. Many (but not all) important journals in psychology, such as Psychological Bulletin and Psychological Methods, are published by APA.  Of greatest interest to me, one of these journals is JPSP.
  2. Whereas many journals do not include submission and acceptance information in PsycINFO, most APA journals do include this information (indeed, APA publishes a few journal-level metrics using some of this information on its website).  Good on APA.
  3. EBSCO, which provides UW-Madison access to PsycINFO, limits the number of records that can be exported at a time to 25,000, and publisher was a convenient way to obtain a grouping of articles under this limit.
To be clear -- I didn't limit my search to APA journals out of a desire to pick on APA.  Rather, I limited my search to APA journals because most APA journals report the information that I wanted in PsycINFO.

My search resulted in information about 16,746 articles published in 38 journals between 2009 and 2014.  For each article, I extracted the submission date, acceptance date, print publication date, and the online publication date if this was different than the print date.  This allowed me to compute, both overall and for each journal, statistics on the time from submission to acceptance (the review time), acceptance to online publication (the online lag), acceptance to print publication (the print lag), and submission to the soonest of the online or print publication (the total time).  I also examined changes between 2009 and 2014 in these statistics.


Overall statistics

Below are the minimum, first quartile, median, third quartile, and maximum (i.e., the 0%, 25%, 50%, 75%, and 100% quantiles) for overall review time, online lag, print lag, and total time.


Here are boxplots of this same information.


I have a few observations about the table and figure:
  1. The median time to review a published article is quite long -- between 7 and 8 months
  2. Once an article is reviewed, the lag to publish a journal article online is consistently smaller than the lag to publish the print version
  3. The median total time to publish an article after it has been submitted is two weeks shy of a year
These statistics support Liz Page-Gould's observation that the timecourse of biology publication is much shorter than the timecourse of psychology publication (at least for APA journals).  349 days is about one and a half times the 220 figure noted by Steve Royle.  Even if we focus just on the time between article acceptance and publication, the times are quite high, and consistently higher than those noted by Daniel Himmelstein for journals indexed by PubMed.


Statistics by journal

Below is a figure showing the review times, online lags, print lags, and total times for all 38 journals in my search.  I have also created a spreadsheet of quantiles by journal, which you can view here.



Focusing in on JPSP, my experience of a review time of about 17 months (510 days) is long -- above the 75% quantile -- but not outrageously long in comparison to the typical review time of 284 days.  Indeed, only four journals of those that I sampled -- Developmental Psychology, Journal of Educational Psychology, Journal of Applied Psychology, and Psychological Methods -- have longer median lag times than JPSP.


Trends over time

Finally, I examined trends over time in review time, online and print lag times, and total time.  The following graph shows loess fits and 95% confidence bands for the relationship between publication date and each of my four variables. 

Click to embiggen

Interestingly enough, there is a sharp decline in total time (the purple line) that begins at around 2010 and ends partway through 2012.  This decline in total time starts around the time that APA journals started adopting "online first" publication policies (the start of the blue line), and seems to be due to the decrease in the lag between acceptance and online publication, which grew increasingly fast in 2011 until leveling out in 2012.  These results suggest that the adoption of online first publication policies have cut the total time between submission and publication.


Conclusions

Based on my analysis of APA journals, the time between submission and publication is long -- usually around a year, and quite a bit longer than at biology journals.  Indeed, given that many papers are rejected at the first journals to which they are submitted, the median time from first submission to eventual publication may be even longer than is represented in my data.  At least we can take cold comfort in the fact that the introduction of online first policies seems to have somewhat lessened overall publication delays.

Why are the delays between submission and publication so long?

I can only speculate as to the reasons, but I think two factors contribute to long publication delays in psychology:
  1. Most reviewers and editors  have unreasonably high standards during the review process.  A manuscript needs to be "perfect" in order to meet these standards, delaying even papers with otherwise amazing data.
  2. Reviewers and editors have an unfortunate habit of requesting the collection of additional data to fix the problems that they perceive in manuscripts.  Sometimes these data can be collected quickly, but sometimes they cannot.
One can argue that, as long as the review process is conducted efficiently and results in improvements in manuscripts, delays due to the review process may be justified.  I'll come back to this argument in a later post, but I think there are reasons to suspect that, for reasons that are not necessarily the fault of either reviewers or editors, the review process as it works in psychology is both inefficient and not always helpful.

What are the consequences of long publication delays?

The most obvious consequence is that long publication delays slow scientific communication, and therefore scientific progress.  However, one consequence that is often overlooked is that publication delays are harmful to the career development of graduate students and early career researchers.

Let's assume that the typical psychology graduate student spends six years in graduate school.  The median time of 349 days from submission to acceptance represents about a sixth of this time in graduate school.  Given that graduate students need to have about 5.6 publications to obtain the most prestigious research jobs (Stenstrom, Curtis, & Iyer, 2013), year-long publication delays can be extremely disruptive to an aspiring scientist's career trajectory.  Small wonder many graduate students are so embittered to the publication process.


References

Stenstrom, D. M., Curtis, M., & Iyer, R. (2013). School rankings, department rankings, and individual accomplishments: What factors predict obtaining employment after the PhD? Perspectives on Psychological Science, 8, 208–217. http://doi.org/10.1177/1745691612474316

No comments:

Post a Comment