The Language of Science: A Primer

Scientific knowledge accrues via the falsification of theory-derived hypotheses.

Falsification is the process of proving a hypothesis wrong.

Falsifiability is the extent to which a hypothesis, or approach to research, can be falsified.

Falsifiability requires sufficient transparency (i.e., full disclosure, open materials, open data, and pre-registration)

Falsifiability is a non-optional essential aspect of the scientific approach (or else one is a historian rather than a scientist).

Falsification is achieved via meticulously executed series of direct replications.

Taxonomy of direct replications with respect to falsifiability (10 shades of falsifiability).

Replication is the activity of carrying out a direct replication.

Replicability is the extent to which a particular effect/hypothesis, or area of research, is replicable.

An effect is said to be replicable if an effect of similar magnitude can be consistently observed, under specified (boundary) conditions, across independent samples and researchers.

Scientific findings must be demonstrably replicable under conditions specified, but not necessarily replicable at will.1




1. For example, the bending of light 1919 solar eclipse findings, which eventually led to the falsification of Newton’s theory of gravitation in favor of Einstein’s General Theory of Relativity, is not replicable at will, but is (and continues to be) demonstrably replicable for solar eclipses satisfying conditions originally specified.

Need for New Code of Ethics Compliance for Professional Researchers in Era of Hyper-Competitive High-Stake Academic Culture

Etienne P. LeBel & Anne Scheel

[Version 2.4; We thank Nick Brown for valuable feedback on a previous version of this blog post.]

Imagine your child is diagnosed with cancer. You have the choice between two drugs: One was developed and tested in a series of registered studies1, the other in non-registered studies. Which one do you choose? You would probably feel that the answer is a no-brainer — you want the drug whose efficacy was based on evidence least influenced by bias.

Extremely high stakes of pharmaceutical research, in the form of billion-dollar revenues generated from FDA-approved drugs, led the World Medical Association (WMA) in 2008 to institute mandatory study registration for all clinical trials reporting evidence on drug efficacy. This was preceded by the International Committee of Medical Journal Editors’ (ICMJE) decision in 2005 whereby non-registered clinical trial studies would no longer be considered for publication. The logic is that the risk posed by researcher biases in the analysis and reporting of study results, including bias in reporting inconclusive or negative studies, is so high that non-registered studies, consequently, simply cannot and should not be trusted.

The modern era of hyper-competitive high-output academic research culture has also led to extremely high stakes for individual researchers in the form of personal rewards such as prestigious jobs, promotion, book deals, outside financial interests, social status, and media attention. Consequently, there are no intellectually honest and defensible reasons against applying this same requirement to all published research involving human subjects. The person who prefers the cancer drug from registered studies cannot simultaneously dismiss the requirement of study registration for their own psychology studies. Consequently, it follows that all human subjects research not publicly registered should not even be considered for publication in any scientific journal (psychology or otherwise).

Indeed, the latest revised Declaration of Helsinki ethical principles, from 2013, dictates precisely such requirement:

  • 35. Every research study involving human subjects must be registered in a publicly accessible database before recruitment of the first subject.
  • 36. Researchers, authors, sponsors, editors and publishers all have ethical obligations with regard to the publication and dissemination of the results of research. Researchers have a duty to make publicly available the results of their research on human subjects and are accountable for the completeness and accuracy of their reports. All parties should adhere to accepted guidelines for ethical reporting. Negative and inconclusive as well as positive results must be published or otherwise made publicly available. Sources of funding, institutional affiliations and conflicts of interest must be declared in the publication. Reports of research not in accordance with the principles of this Declaration should not be accepted for publication.

Given study registration is not yet mandatory in psychology, however, professional psychologist researchers are not yet complying with these new ethical principles.2 Due to the high-stake personal rewards of the current academic research culture, however, we strongly believe it is time that all professional psychologist researchers abide by such new ethical principles requiring mandatory study registration, in addition to minimal reporting standards, open materials/data, and hypothesis pre-registration.

Anything short of this, given the environment in which researchers operate, fails to adhere to fundamental scientific principles: That is, reporting and testing hypotheses with sufficient transparency and thus falsifiability to maximize the likelihood that we as a research community can conclude a hypothesis is wrong, if it is in fact wrong (which can be easily achieved given new technologies3):

  • Without study registration at a centralized public registry, it is impossible, for us as researchers, to account for the selective file-drawering of “failed” or inconclusive studies.
  • Without a pre-registered method protocol (specified prior to data collection), it is near-impossible for us to account for the multitude of ways researchers may have (un)intentionally exploited analytic and design flexibility to achieve a publishable result.
  • Without minimal reporting standards (e.g., 21-word solutionBASIC 4 Psychological Science reporting standard), we cannot properly evaluate the strength of the reported evidence.
  • Without open materials, we cannot properly scrutinize the experimental design, nor can we conduct diagnostic independent replicability tests.
  • Without open data, we cannot verify the analytic reproducibility or the analytic robustness of the reported results, which need to be independently confirmed before investing precious research resources conducting expensive independent replications.

Being a scientist is a special and precious privilege. It is not an irrevocable right. As credentialed professionals, public intellectuals, and mentors, we have an inordinate amount of influence on citizens, the media and journalists, industry research and corporations, government agencies, NGOs, and other researchers both within and outside our respective fields. But with such importance and respect comes great responsibility.

Consequently, it follows that insufficiently transparent, and hence insufficiently falsifiable, research should be considered professionally unethical for the following reasons:

  • When the public funds research, taxpayers provide money in good faith that the funded projects will advance knowledge and help address societal problems. Non-falsifiable research wastes public funds which could otherwise be spent on social services and programs that reduce suffering and safe lives.
  • Non-falsifiable research also wastes additional public funds spent misguidedly trying to replicate and build upon such research.
  • Non-falsifiable research also leads to costly and ineffective practical implementation attempts, which can have grave consequences on real-world practical, legal, and political decisions.
  • Non-falsifiable research wastes the time of volunteering human subjects and in some cases unjustly puts their well-being at risk.
  • Non-falsifiable research erodes the public’s trust in scientists, will lead to further research funding cuts, and stifles society’s evolution toward evidence-based policy-making.

We propose that all professional psychologists need to abide by the new 2013 Declaration of Helsinki ethical principles, which are consistent with current lower bar country-based professional society code of ethics including the APA, CPA, DGPs, VSNU, and European Code of Conduct for Research Integrity (and as has been previously argued here). This is gravely needed to finally be accountable to the public. Accountable to the fact that all published research actually follows fundamental scientific principles, ensuring the necessary degree of transparency and falsifiability required for scientific progress (building upon existing softer and voluntary initiatives such as the Commitment to Research Transparency and the TOP guidelines).

Such a new ethical code of conduct would explicitly stipulate the following standards for all published scientific research4:

  • Public registration of all studies at a field-relevant centralized registry, which includes a pre-registered method protocol document clearly describing rationale of study, study sample and design, and planned data analytic approaches (e.g., IRB ethics approval documents).
  • Compliance with fundamental reporting standards relevant to the reported research (e.g., BASIC 4, CONSORT standard for experimental studies; STROBE standard for observational/correlational studies)
  • Open materials: Public online archiving of all relevant procedural details, materials, and measures, unless proprietary exclusions apply, to allow for proper scrutiny of experimental design and independent replicability tests.
  • Open data: Public online archiving of all relevant data, raw or transformed data, unless proprietary or confidentiality exclusions apply, to allow for verification of analytic reproducibility and analytic robustness of reported results.

Compliance with this new code of ethics could be implemented by having each stakeholder in a researcher’s ecosystem (i.e., via journals, professional societies, funding agencies, university employment contracts) require that individual researchers explicitly consent to following such code. This is akin to the Hippocratic Oath for medical professionals, guided by the more general Hippocratic Oath proposed for all scientists (see also here). Upon taking such oath, violation of the new ethical standards should be considered unethical and should be investigated as researcher misconduct by the appropriate stakeholder(s) involved.

We urgently need, at this current moment, to have a serious discussion within the psychological research community about the minimum scientific standards that need to be met to be an ethical researcher in this modern era of high-stake hyper-competitive high-output academic research culture. This discussion should incite calls to action to ensure that all stakeholders vigilantly enforce compliance to this new code of ethics. Otherwise, the reputation of all professional psychologists will continue to be tarnished, extensive research waste and direct and indirect harm to society will continue, and the public’s trust in science will be further eroded.




1. “Registered studies” as in studies registered in public centralized study registries prior to data collection, such as
2. We must emphasize, however, that a growing minority of psychologists have made admirable efforts to pre-register and provide open materials/open data for some or all of their studies.
3. E.g. technologies to safely store and share data and materials, preregister studies, establish a reproducible workflow, conduct multi-lab collaborations, verify the accuracy of one’s own and others’ reported results, and make manuscripts publicly available for pre-publication peer feedback.
4. These standards should not be misconstrued as guaranteeing scientific knowledge, but rather as minimal standards that need to be in place to allow the possibility of achieving valid and generalizable knowledge about how our world works.

A simpler and more intuitive publication bias index?

At this past SPSP, Uri Simonsohn gave a talk on new ways of thinking about statistical power. From this new perspective, you first determine how large a sample size you can afford for a particular project. Then, you can determine the minimum effect size that can reliably detected (i.e., 95% power) for that sample size (e.g., d_min = .73 can be reliably detected with n=50/cell). I believe that this approach is a much more productive way of thinking about power for several reasons, one being that it substantially enhances the interpretation of null results. For instance, you can conclude (assuming integrity of methods and measurement instruments) that the effect you’re studying is unlikely to be the size of the minimum effect size reliably detectable for your sample size (or else you would have detect it). That being said, it is still possible the effect exists but is much smaller in magnitude, which would require a much larger sample size to reliably detect.

In this post, I use the core ideas from this new approach to come up with a simpler and more intuitive way of gauging publication bias for extant empirical studies.

The idea is simple. If a study reports an observed effect size smaller than the minimum effect size reliably detectable for the sample size used, then the study likely suffers from publication bias and should be interpreted with caution. The further away the observed effect size is from the minimally detectable effect size, the larger the bias. Let’s look at some concrete examples.

Zhong & Liljenquist’s (2006) Study 1 on the “Macbeth effect” found a d=.53 using n=30/cell. At this sample size, however, only effect sizes as large as d=.95 (or greater) are reliably detectable with 95% power. On the other hand, Tversky & Kahneman’s (1981) Framing effect study found a d=1.13 using n=153/cell. At that sample size, effect sizes as small as d=.41 are reliably detectable. See Table below for other examples:
minimum effect size

The new bias index can be calculated as follows:  minimum effect size - bias-equation

(And note we’d want to calculate a 95% C.I. around the bias estimate, given that bias estimates should be more precise for larger Ns all else being equal.)

To shed more light on the value of this simpler publication bias index, in the near future I will calculate these for studies where replicability information exists and test empirically whether the index predicts lower likelihood of replication.