In a paper subtitled, “A simulation study of editors, reviewers, and the scientific publication process,” political scientist Justin Esarey writes:
Under any system I study, a majority of accepted papers will be evaluated by the average reader as not meeting the standards of the journal. Moreover, all systems allow random chance to play a strong role in the acceptance decision. Heterogeneous reviewer and reader standards for scientific quality drive both results.
He concludes:
A peer review system with an active editor (who uses desk rejection before review and does not rely strictly on reviewer votes to make decisions) can mitigate some of these effects.
This seems reasonable to me. As a reviewer, I give my recommendation but I recognize that the decision is up to the editor. This takes the pressure off me: I feel that all I have to do is provide useful information, not to make the decision.
Esarey’s paper also includes some graphs which are pretty good, but I won’t include them here because I’m so bummed that he doesn’t label the lines directly—I can’t stand having to go back and forth between the lines and the legend. Also I don’t like graphs where the y-axis represents probability but the axis goes below 0 and above 1. It’s all there, though, at the link.
I like the paper. I haven’t read the details so I can’t comment on Esarey’s specific models, but the general features seem to make sense, so it seems like a good start in any case.
Re the space on the chart – I don’t like when geometric elements of the chart collide with the boundary of the chart, so I often have space below zero. Here is one example for histograms, https://andrewpwheeler.wordpress.com/2014/03/27/the-baseline-in-histograms-outliers/.
I fully agree with this. The data elements shouldn’t collide with the edges, so I want a little room. That being said, I could see for something that’s bounded between a and b (or 0, 1 in the specific case of probability) it could be useful to have light-grey horizontal lines (or something equally low-key) at the boundary values so that at any point on the graph you can see exactly where the value is relative to the boundaries.
If it’s important, I changed the figure with probabilities so that 0 and 1 are the boundaries of the y-axis. The new version is available on the link Andrew posted (http://jee3.web.rice.edu/peer-review.pdf).
I like those gray lines. I was thinking that the problem with using the line showing the scale of x and the line indicating 0 be the same is that it would be weird to have the 0 values foregrounded on the scale line. I use crime data too and it is really important to know what is 0 or near 0 and what is empty. You are not really showing Y of less than 0, you are really separating the scale from the 0.
The strong role of random chance seems unacceptable, especially given the typically long review times. It cannot benefit science to have 6 month review times where the outcome is in no small part random.
Based on what I saw in my simulations, it’ll be hard to eliminate chance as a factor using any of the traditional institutions of peer review. We could go to the PLoS model, or back to the old model of unilateral editor acceptance, but I’m sure those institutions have problems as well. It’s an interesting puzzle…
Great name, by the way.
Isn’t a “political scientist” doing simulations of peer review the third sign of the apocalypse? The first two being ‘Econophysics’ and Economists doing endless statistical econ of Econ Journal citations?
Anon:
I wouldn’t want political scientists only studying peer review, or mostly studying peer review, or even often studying peer review. But the occasional study of the topic can be helpful.
I take it the phrase “self licking ice cream cone” isn’t common in academia.
This has inspired me to write an article on how only the smartest reviewers pass papers with the phrase “navel gazing” in the title. The paper is title: “A causal analysis of reviewers IQ and navel gazing”. I will be submitting it to reviewers shortly.
Political scientists study institutions of governance. Peer review is the system that governs science. So…
“[Peer Review] improves the quality of manuscripts that go through the process (Goodman et al. 1994), and can help to identify the most impactful contributions to science (Li and Agha 2015).”
Goodman: http://www.ncbi.nlm.nih.gov/pubmed/8198342
Li: http://www.ncbi.nlm.nih.gov/pubmed/25908820
I’m not sure I buy the claim that what was measured in these studies corresponds to “quality” and “impactful contribution to science”, rather than some kind of “fitting in with the crowd” (ie prevailing bias).
+1
Andrew as medical doctor:
Nurse: Doctor Gelman, this patient is flatlining!
Andrew: Uch, that EKG is so poorly labeled. I don’t even see the units anywhere.
all systems allow random chance to play a strong role in the acceptance decision
Some folks with a lot of patience and desire to “succeed” exploit it on routine basis: Everything, mo matter how mundane, gets sent to the Top 3 journals (in succession, if needed), then to the Top 10, etc. The end result is that they end up only publishing “brilliant science”.
Yep. journal publication and the association of *which journal* as equivalent to *how brilliant* is just broken flat out. It’s like an airplane with no wings. Lots of people are climbing around on the fuselage saying it needs just a little patch-up here and there. And then there’s those of us facepalming and shaking our heads.
The only reasonable alternative in my opinion is something like “archive.org” whose whole goal is simply to record/store and provide permanent reference addresses for people’s publications. let “how briliant” be a function of *CONTENT*, reviewer reputation, open advocacy, and open criticism.