Beyond Open Access: Improving Quality in Research
Improving scientific publishing requires more than just removing barriers.
Updated May 20, 2025 | Reviewed by Jessica Schrader
There’s a myriad of research-related issues that make their way into public discourse. One that pops up every now and then concerns the academic publishing process. Discussions on the topic will vary, ranging from the publish-or-perish mindset instilled in academic publishing to the inaccessibility of quality research—including research that is publicly funded—to the gatekeeping that occurs in the scientific review process. These topics are usually of inherent interest to me, given that some of my time is spent as an academic researcher. But they’re also of interest to many people who aren’t academic researchers, especially in light of the public’s current negative attitude about experts and the media and the increasing trend of people “doing their own research.”
I recently came across an article by Kulldorff (2025) , which focuses on the rise and fall of scientific journals and a potential way forward. That potential way forward involved four pillars: (1) open access publishing, (2) open peer reviewing, (3) paying peer reviewers, and (4) removing article gatekeeping (which concerns removing the acceptance/rejection process, especially the desk rejections process) [1] .
What was most interesting was not that it was merely a thought piece about an issue that has been thought-pieced ad nauseam . Rather, it was actually a first attempt to apply the “way forward” that the author proposes, published in a journal that is seeking to apply those pillars. The article is worth reading, as I’m not going to use this post merely to recap Kulldorff’s arguments [2] . What I am going to do here is push back a bit on Kulldorff’s fourth pillar—arguing that some level of gatekeeping is necessary—while also proposing a fifth pillar to address another pressing issue in research publishing: the growing tendency to prioritize analytical complexity over clarity and necessity.
Some Gatekeeping Can Be Beneficial
I was fully on board with Kulldorf’s pillars until I made it to the fourth one. The full wording of the pillar is as follows:
Removal of article gatekeeping, letting an organization’s scientists freely publish all their research results in a timely and efficient manner.
While removing barriers to publication sounds appealing in theory, it overlooks a key role that editorial oversight can play in ensuring the quality and rigor of published research. And so, my counterargument here is that some article gatekeeping is necessary and can serve a valid purpose in the scientific process. There are times when a rejection is called for as a means of forcing the author to improve (1) the strength of the arguments underlying the study [3] or (2) the quality of evidence/analytical techniques used in the study [4] .
Some of my own research serves as a testament to that, with a specific case being Grawitch et al. (2023) . Our original attempt to publish involved a single study (rather than the three-study finished product) and less sophisticated theoretical and logical foundations underpinning the research. So, some of the rejections served to assist us in refining our arguments and improving the strength of our inferences, both of which ultimately led to a higher-quality finished product.
That all being said, there was also a lot of questionable gatekeeping involved in the process, from desk rejections with little to no rationale to lengthy reviewer rants on tangential topics unrelated to our specific studies, to bizarre assertions made about the implications of our research. Sadly, my experience with this single manuscript was not uncommon.
As such, some degree of gatekeeping seems necessary, especially given the increased use of AI to generate junk science ( Schultz, 2025 ) [5] . The increasing volume of published research and variability in its quality necessitates some sort of formalized gatekeeping system. Such a system should have clearly defined (and limited) parameters and minimize use of the desk rejection as a gatekeeping tool [6] .
While I generally agree with the direction in which Kulldorf is arguing published research should head, I would contend that the four pillars are incomplete. As a basis for understanding what’s missing, I’ll rely on a recent argument put forth by Spector (2024) [7] . In it, he argued the field of industrial/organizational psychology has become too beholden to statistical modeling, describing such analyses as a a “trojan horse.” The reasoning for his argument is that, in the vast majority of cases, modeling amounts to nothing more than trying to make correlational results look more sophisticated than they really are [8] . Such analyses lead to claims (and visuals) that imply causality when causality is not at all tested [9] .
The overuse of modeling is just one symptom of a broader issue—a research culture that prioritizes analytical sophistication over clarity and necessity. Many studies rely on complex techniques, not because they are required to answer the research questions, but because the use of such analytical techniques can enhance publishability. Structural modeling, multilevel analyses, or intricate machine learning techniques may sound impressive, but if they don’t add meaningful insight beyond a traditional statistical technique, they risk muddying rather than clarifying our understanding [10] .
And so, my fifth pillar would be as follows:
Analyses should be no more complex than is necessary to answer the research questions.
When analytical complexity becomes a virtue rather than a necessity, research loses clarity, accessibility, and its ability to advance knowledge.
Kulldorff’s proposal offers a compelling vision for reforming scientific publishing, addressing key issues such as accessibility, transparency, and the inefficiencies of peer review. However, no reform is without trade-offs. While removing barriers to publication could accelerate the dissemination of knowledge, it also risks weakening the quality control mechanisms that help refine research before it enters the public domain. Some degree of gatekeeping is necessary—not to suppress ideas (as gatekeeping sometimes does now), but to ensure that published research meets basic standards of rigor and validity.
At the same time, the push for analytical sophistication as a marker of quality has created a different set of challenges. The overuse of complex analytical techniques—whether structural modeling, hierarchical methods, or machine learning—often adds layers of sophistication without adding real insight. If complexity is prioritized for its own sake rather than as a means to answer meaningful research questions, we risk making scientific findings less transparent, less replicable, and ultimately, less useful.
Scientific publishing should evolve, but in a way that preserves the integrity of research rather than introducing new problems. Any meaningful reform must strike a balance between accessibility and rigor, openness and quality, innovation and clarity. Ensuring that research remains both rigorous and interpretable is not an obstacle to progress—it’s the very foundation of scientific advancement.
[1] I will admit I was a bit confused as to what this was originally, but reading the entire piece will make it clearer what the author was referring.
[2] Besides, I was already predisposed to agree with much of what he argued, given my extensive dissatisfaction with the existing publication process, that anyone who know me will attest to.
[3] This would most notably concern the argument presented in the introduction.
[4] This would most notably concern the number of studies included – which I would contend also requires some caveats – and the specific analyses used to analyze the data (inappropriate analytical techniques would potentially become an issue).
[5] This begs the question of how useful gatekeeping really is, but I would imagine the problem would be even worse without some gatekeeping. One might further argue that some of Kulldorf’s other pillars – especially around peer review – would help to keep this issue somewhat in check.
[6] This is when the editor or action editor rejects a submission out of hand, often with no reasoning provided (other than that it doesn’t fit the journal ort reviewers won’t review it favorably). In cases where an action editor finds egregious errors that would warrant such a rejection, such errors should be clearly specified and argued in the rejection. Any other reason for a desk rejection can become easily abused.
[7] I know, two months in a row I am referring to Paul’s work, but it again fits nicely with my own argument.
[8] I will also admit I am guilty of this, using structural equation modeling and especially path modeling in several of my prior studies.
[9] I have sought to find a way to represent such noncausal models a different way (e.g., more Venn diagram like), but I have yet to find anything that is easily presented and interpretable.
[10] More sophisticated analyses also make it more difficult for reviewers to identify errors, faulty conclusions, and the like because it can be challenging for journals to find subject matter experts who also happen to be experts in the various advanced statistical techniques that may be used.
Share this post Facebook Bluesky Linkedin Email
There was a problem adding your email address. Please try again.
By submitting your information you agree to the Psychology Today Terms & Conditions and Privacy Policy
Matt Grawitch, Ph.D. , is a professor at Saint Louis University (SLU), serving within the School for Professional Studies (SPS).
Get the help you need from a therapist near you–a FREE service from Psychology Today.
This article is part of the Bringwise Psychology Journal — daily insights on human behavior, mental health, and personal growth.