Are scientific journals just parasites? (with Chris Chambers)

Aug 7, 2022 Episode Page ↗
Overview

Spencer Greenberg and Chris Chambers discuss biases in academic publishing, like outcome and publication bias. They propose reforms such as Registered Reports, open science practices, and altering researcher evaluation to foster more robust and trustworthy scientific progress.

At a Glance
14 Insights
1h 19m Duration
19 Topics
5 Concepts

Deep Dive Analysis

Introduction to Science Reform and Outcome Bias

Understanding Outcome Bias and Publication Bias

Critique of Journal Prestige and 'Flashy' Results

Introducing Registered Reports for Bias Neutralization

Mechanism and Benefits of Registered Reports

History and Adoption of Registered Reports

Registered Reports for Exploratory and Confirmatory Science

Addressing Self-Deception in Scientific Research

Critiques of Commercial Publishers and Free Peer Review

Motivations for Scientists to Conduct Peer Review

Publishers' Role: Distribution, Credibility, and Alternatives

The Peer Community Inn (PCI) Initiative

Challenges of Journal Brands as Quality Heuristics

Differentiating Scientific Record from Science Journalism

Incentives and Reforms for Academic Hiring and Promotion

The Role and Misuse of Citation Counts

Rapid Fire: Sci-Hub and Open Access

Rapid Fire: Lowering P-Values and Statistical Debates

Rapid Fire: Data Sharing and Ego Depletion Consensus

Outcome Bias

A cognitive bias where the outcome of an intervention, policy, or decision determines whether that decision was considered good. In science, this can lead to a biased record if judgments about research quality are influenced by the results obtained.

Publication Bias

A specific form of outcome bias in scientific literature where the likelihood of a study being published is influenced by whether its results are statistically significant or conform to expectations. This can lead to a skewed scientific record where negative or inconclusive findings are underrepresented.

Registered Reports

A publishing format designed to neutralize outcome and publication bias by having journals decide whether to publish research *before* the results are known. This involves peer review of a study's protocol, methodology, and theoretical background, leading to an 'in-principle acceptance' regardless of the eventual findings.

Preprints

Scientific manuscripts that are made publicly available online prior to formal peer review and journal publication. They serve as an efficient way to disseminate research freely, allow for community feedback, and support versioning for corrections and updates.

Peer Community Inn (PCI) Initiative

A non-profit, academic-led initiative that manages the peer review process for preprints, publishing these reviews and evaluations openly. It aims to shift control of scientific credibility from commercial publishers back to the academic community, allowing journals to accept these evaluations without further peer review.

?
What is outcome bias in science?

Outcome bias is a cognitive bias where the quality of a decision or study is judged based on its outcome, rather than the rigor of the process, which can lead to a distorted scientific record if inconvenient results are suppressed.

?
How do Registered Reports address outcome bias?

Registered Reports tackle outcome bias by having journals conduct peer review of a study's methodology and theoretical background *before* the results are known, issuing an 'in-principle acceptance' that guarantees publication regardless of the findings.

?
Why do scientists perform peer review for free?

Scientists engage in free peer review due to intrinsic motivations like contributing to public good and advancing their field, as well as more self-serving reasons such as gaining early access to research, building a reputation with editors, and career advancement.

?
What value do commercial publishers add to the scientific process today?

Commercial publishers primarily manage the peer review process and maintain journal brands, but their distribution function is largely superseded by the internet, and their credibility is argued to stem from the free labor of academics rather than their own added value.

?
How can the academic community regain control over scientific publishing credibility?

The academic community can regain control by managing peer review of preprints through non-profit initiatives like Peer Community Inn (PCI), which openly publishes reviews and evaluations, allowing journals to accept these evaluations without further peer review.

?
How should scientific journals differ from newspapers?

Scientific journals should focus on curating a scientific record based solely on the quality and rigor of research, regardless of outcome, while newspapers (or a separate layer) can focus on promoting 'cool' or newsworthy findings, ensuring these two functions are not blurred.

?
Why is it difficult to achieve consensus on topics like ego depletion?

Lack of consensus in fields like psychology, exemplified by ego depletion, often stems from the low quality and rampant biases (publication, outcome, reporting) in much of the existing work, which creates unnecessary uncertainty and a distorted literature.

?
What are the benefits of publishing preprints?

Preprints allow for the free and rapid dissemination of scientific research prior to formal peer review, enabling community feedback, version control for corrections, and providing a transparent record of a paper's evolution, which is valuable for learning.

1. Adopt Registered Reports for Research

For appropriate research, submit a detailed protocol for peer review before conducting the study to secure in-principle acceptance, ensuring publication regardless of results. This neutralizes outcome and publication bias, focusing evaluation on methodological quality and question importance.

2. Prioritize Quality Over Flashy Results

As a researcher, prioritize conducting high-quality, rigorous work over seeking exciting or novel results, as the scientific record should be determined by work quality, not outcomes. This helps prevent a lopsided literature and self-deception.

3. Embrace Transparency in Research

Be transparent about all analyses, including exploratory ones, and avoid reinventing history to present post-hoc discoveries as predictions. This ensures the scientific record is accurate and prevents misleading conclusions.

4. Publish All Rigorous Research Findings

Publish the outcomes of all rigorous research, including negative, inconclusive, or assumption-challenging findings, potentially using formats like Registered Reports. This prevents others from repeating mistakes and makes the best use of scientific resources.

5. Share Research Data Publicly

Publish data sets, even if not fully used in your research, with appropriate anonymization, especially for publicly funded work. This allows the broader scientific community to benefit from collected data and advances knowledge.

6. Shift Peer Review Control

Advocate for and participate in initiatives that shift control of the peer review process from commercial publishers to the academic community, such as using preprints and non-profit peer review entities. This reduces publishing costs and ensures the process serves scientific, not corporate, interests.

7. Reform Researcher Evaluation Criteria

As an academic, advocate for and implement changes in hiring, promotion, and assessment panels to value open science practices (e.g., data sharing, pre-registration) and the quality of research, rather than relying on crude heuristics like journal prestige or grant money.

8. Utilize Preprints for Dissemination

Publish research as preprints on public websites before formal journal submission to rapidly share findings, initiate community discussion, and allow for versioning and corrections. This makes science more accessible and dynamic, moving beyond an 18th-century fixed record model.

9. Exercise Caution with Citation Counts

View citation counts as a measure of short-term impact, not necessarily an indication of research quality. Avoid relying on them as a superficial metric for evaluating work or scientists.

10. Ensure Balanced and Ethical Citations

As an author, ensure citations are properly balanced and avoid strategically omitting contradictory work. As a reviewer, avoid coercing authors into inappropriate citations.

11. Practice Responsible Science Communication

When communicating science to the public or journalists, ensure press releases and discussions are as careful, factual, and free of spin as possible. This is crucial as many journalists may lack deep scientific training.

12. Distinguish Science Record from News

Recognize that the scientific record should publish research based solely on quality, regardless of outcome, while a separate layer can highlight ‘cool’ or newsworthy findings. This prevents blurring and distortion of scientific priorities.

13. Embrace Disagreement and Lack Consensus

Recognize that lack of consensus is normal and healthy in many areas of science, especially for newer theories, and is not necessarily a sign of failure. This fosters critical thinking and ongoing scientific inquiry.

14. Continuously Learn and Seek Understanding

Make a habit of constantly taking in information from a wide variety of sources (scientific papers, articles, books) to build broad knowledge and deeply understand the structure and models of various phenomena. This fosters intellectual growth and expertise.

The scientific record should not be determined by the results. It should be determined by the quality of the work going into it.

Chris Chambers

You are the easiest person to fool in your own research.

Chris Chambers (attributing Richard Feynman)

Credibility does not come from publishers. They're a proxy for it. Only they're sort of a heuristic that we use.

Chris Chambers

Eliminating outcome bias in science is so much more important than deciding whether to use a Bayesian hypothesis test or a frequentist t-test.

Chris Chambers

The traditional journal-based publishing system is stuck in the 18th century of a permanent fixed record, which is there and that's that. It's just not the way science works.

Chris Chambers

If it does, it's really nothing like it's proposed at the moment in terms of theory. It's a warning, I think, to the psychological field that we need to work much harder at controlling our own biases and avoiding fooling ourselves.

Chris Chambers

Registered Reports Publication Process

Chris Chambers
  1. Authors write and propose a full theoretical background and detailed methodology (protocol).
  2. The protocol is submitted for peer review *before* the study is conducted.
  3. The journal evaluates the quality of the method, importance of the question, ethical standards, and other elements contributing to scientific quality, *without knowing the results*.
  4. Based on this evaluation, the journal issues an 'in-principle acceptance' (IPA) if the protocol is approved.
  5. Authors then conduct the research, knowing that the outcomes will not determine whether or not it gets published.
  6. Upon completion, authors submit the finished study.
  7. The same reviewers and journal assess whether the protocol was followed and if the conclusions are based upon the evidence.
  8. If the protocol was followed and conclusions are evidence-based, the study is published.
approximately 300
Number of journals offering Registered Reports format A small percentage of the thousands of journals in science
2-5%
Percentage of overall empirical submissions that are Registered Reports at a typical journal Still a minority of submissions
0.05
P-value threshold traditionally used for making claims Considered too liberal and can lead to too many false discoveries
0.005 or even lower
Proposed lower p-value threshold Suggested to reduce false discoveries
30-40%
Estimated replication rate for social science papers from top journals (not registered reports) Based on large-scale replication initiatives, or 1 in 3 to 1 in 4