Reproducibility: Difference between revisions

Revision as of 03:16, 14 May 2025

Latest revision as of 19:39, 3 August 2025

(14 intermediate revisions by 8 users not shown)

Line 2:

{{About|the reproducibility of scientific research results|reproductive capacity of organisms|fertility|and|fecundity|reproducibility in the context of computer software|Reproducible builds}}

'''Reproducibility''', closely related to '''replicability''' and '''repeatability''', is a major principle underpinning the [[scientific method]]. For the findings of a study to be reproducible means that results obtained by an [[experiment]] or an [[observational study]] or in a [[statistical analysis]] of a [[data set]] should be achieved again with a high degree of reliability when the study is replicated. There are different kinds of replication<ref>{{Cite journal|last1=Tsang|first1=Eric W. K.|last2=Kwan|first2=Kai-man|date=1999|title=Replication and Theory Development in Organizational Science: A Critical Realist Perspective|url=http://dx.doi.org/10.5465/amr.1999.2553252|journal=Academy of Management Review|volume=24|issue=4|pages=759–780|doi=10.5465/amr.1999.2553252|issn=0363-7425|url-access=subscription}}</ref> but typically replication studies involve different researchers using the same methodology. Only after one or several such successful replications should a result be recognized as scientific knowledge.

== Types of Reproducibility ==

There are different kinds of replication studies, each serving a unique role in scientific validation:

Direct Replication – The exact experiment or study is repeated under the same conditions to verify the original findings.

Conceptual Replication – A study tests the same hypothesis but uses a different methodology, materials, or population to see if the results hold in different contexts.

Computational Reproducibility – In data science and computational research, reproducibility requires making all datasets, code, and algorithms openly available so others can replicate the analysis and obtain the same results.

== Importance of Reproducibility ==

Reproducibility serves several critical purposes in science:

Verification of Results – Confirms that findings are not due to random chance or errors.

Building Trust in Research – Scientists, policymakers, and the public rely on reproducible studies to make informed decisions.

Advancing Knowledge – Establishes a strong foundation for future research by validating existing theories.

Avoiding Bias and Fraud – Helps detect false positives, publication bias, and data manipulation that could mislead the scientific community.

Challenges in Achieving Reproducibility

Despite its importance, many studies fail reproducibility tests, leading to what is known as the replication crisis in fields like psychology, medicine, and social sciences. Some key challenges include:

Insufficient Data Sharing – Many researchers do not make raw data, code, or methodology openly available, making replication difficult.

Small Sample Sizes – Studies with limited sample sizes may show results that do not generalize to larger populations.

Publication Bias – Journals tend to publish positive findings rather than null or negative results, leading to an incomplete scientific record.

Complex Experimental Conditions – In some cases, small variations in laboratory settings, equipment, or researcher expertise can affect outcomes, making exact replication difficult.

== Real-World Applications of Reproducibility ==

Medical Research – Reproducibility ensures that clinical trials and drug effectiveness studies produce reliable results before treatments reach the public.

AI and Machine Learning – Scientists emphasize reproducibility in AI by requiring open-source models and datasets to validate algorithm performance.

Climate Science – Climate models must be reproducible across different datasets and simulations to ensure accurate predictions of global warming.

Pharmaceutical Development – Drug discovery relies on reproducing experiments across multiple labs to ensure safety and efficacy.

== Improving Reproducibility in Science ==

To enhance reproducibility, researchers and institutions can adopt several best practices:

Open Data and Code – Making datasets and computational methods publicly available ensures that others can verify results.

Registered Reports – Some scientific journals now accept studies based on pre-registered research plans, reducing bias.

Standardized Methods – Using well-documented, standardized experimental protocols helps ensure consistent results.

Independent Replication Studies – Funding agencies and journals should prioritize replication studies to strengthen scientific integrity.

With a narrower scope, ''reproducibility'' has been defined in [[computational science]]s as having the following quality: the results should be documented by making all data and code available in such a way that the computations can be executed again with identical results.

In recent decades, there has been a rising concern that many published scientific results fail the test of reproducibility, evoking a reproducibility or [[replication crisis]].

==History==

Line 67 ⟶ 15:

''Replicability'' and ''repeatability'' are related terms broadly or loosely synonymous with reproducibility (for example, among the general public), but they are often usefully differentiated in more precise senses, as follows.

WhenTwo major steps are naturally distinguished in connection with reproducibility of experimental or observational studies: when new data isare obtained in the attempt to achieve it, the term ''replicability'' is often used, and the new study is a ''replication'' or ''replicate'' of the original one. Obtaining the same results when analyzing the data set of the original study again with the same procedures, many authors use the term ''reproducibility'' in a narrow, technical sense coming from its use in computational research. ''Repeatability'' is related to the ''repetition'' of the experiment within the same study by the same researchers. Reproducibility in the original, wide sense is only acknowledged if a replication performed by an ''independent researcher team'' is successful.

Two major steps are naturally distinguished in connection with reproducibility of experimental or observational studies:

When new data is obtained in the attempt to achieve it, the term ''replicability'' is often used, and the new study is a ''replication'' or ''replicate'' of the original one. Obtaining the same results when analyzing the data set of the original study again with the same procedures, many authors use the term ''reproducibility'' in a narrow, technical sense coming from its use in computational research.

''Repeatability'' is related to the ''repetition'' of the experiment within the same study by the same researchers.

Reproducibility in the original, wide sense is only acknowledged if a replication performed by an ''independent researcher team'' is successful.

The terms reproducibility and replicability sometimes appear even in the scientific literature with reversed meaning,<ref>{{cite arXiv|title=Terminologies for Reproducible Research|last1=Barba|first1=Lorena A.|year=2018|class=cs.DL |eprint=1802.03311}}</ref><ref>{{cite web|title=Replicability vs. reproducibility — or is it the other way round?|last1=Liberman|first1=Mark|url=https://languagelog.ldc.upenn.edu/nll/?p=21956|access-date=2020-10-15}}</ref> as different research fields settled on their own definitions for the same terms.<ref>{{cite journal|title=Brooke on the Merton Thesis: A Direct Replication of John Hedley Brooke’sBrooke's Chapter on Scientific and Religious Reform.|last1=Van Eyghen|first1=Hans|last2=Van den Brink| first2=Gijsbert |last3=Peels | first3=Rik|year=2024|journal=Zygon |volume=59| issue=2| url=https://www.zygonjournal.org/article/id/11497/#!}}</ref>

==Measures of reproducibility and repeatability==

In chemistry, the terms reproducibility and repeatability are used with a specific quantitative meaning.<ref>{{Cite journal |last= |first= |title=IUPAC - reproducibility (R05305) |url=https://goldbook.iupac.org/terms/view/R05305 |access-date=2022-03-04 |website=[[International Union of Pure and Applied Chemistry]]|doi= 10.1351/goldbook.R05305|doi-access=free|url-access=subscription}}</ref> In inter-laboratory experiments, a concentration or other quantity of a chemical substance is measured repeatedly in different laboratories to assess the variability of the measurements. Then, the standard deviation of the difference between two values obtained within the same laboratory is called repeatability. The standard deviation for the difference between two measurement from different laboratories is called ''reproducibility''.<ref name="ASTM E177">{{cite web|url=https://www.astm.org/Standards/E177.htm |title=Standard Practice for Use of the Terms Precision and Bias in ASTM Test Methods |year=2014 |author=Subcommittee E11.20 on Test Method Evaluation and Quality Control |publisher=ASTM International |id=ASTM E177}}{{Subscription required}}</ref>

These measures are related to the more general concept of [[variance component]]s in [[metrology]].

Line 106 ⟶ 51:

Reproducible research is key to new discoveries in [[pharmacology]]. A Phase I discovery will be followed by Phase II reproductions as a drug develops towards commercial production. In recent decades Phase II success has fallen from 28% to 18%. A 2011 study found that 65% of medical studies were inconsistent when re-tested, and only 6% were completely reproducible.<ref>{{Cite journal|last1=Prinz |first1=F. |last2=Schlange |first2=T. |last3=Asadullah |first3=K. |doi=10.1038/nrd3439-c1 |title=Believe it or not: How much can we rely on published data on potential drug targets? |journal=Nature Reviews Drug Discovery |volume=10 |issue=9 |page=712 |year=2011 |pmid=21892149 |doi-access=free}}</ref>

Some efforts have been made to increase replicability beyond the social and biomedical sciences. Studies in the humanities tend to rely more on expertise and hermeneutics which may make replicability more difficult. Nonetheless, some efforts have been made to call for more transparency and documentation in the humanities.<ref>{{Cite journal |last1=Van Eyghen |first1=Hans |last2= Van den Brink |first2=Gijsbert |last3= Peels |first3= Rik |title=Brooke on the Merton Thesis: A Direct Replication of John Hedley Brooke's Chapter on Scientific and Religious Reform |journal=Zygon: Journal of Religion and Science |volume=59 |issue=2 |year=2024|url=https://www.zygonjournal.org/article/id/11497/| doi=10.16995/zygon.11497|doi-access=free }}</ref>

==Noteworthy irreproducible results==

Line 112 ⟶ 59:

In March 1989, [[University of Utah]] chemists Stanley Pons and Martin Fleischmann reported the production of excess heat that could only be explained by a nuclear process ("[[cold fusion]]"). The report was astounding given the simplicity of the equipment: it was essentially an [[electrolysis]] cell containing [[heavy water]] and a [[palladium]] [[cathode]] which rapidly absorbed the [[deuterium]] produced during electrolysis. The news media reported on the experiments widely, and it was a front-page item on many newspapers around the world (see [[science by press conference]]). Over the next several months others tried to replicate the experiment, but were unsuccessful.<ref>{{cite journal|title=Physicists Debunk Claim Of a New Kind of Fusion|newspaper=New York Times|last=Browne|first=Malcolm|url=http://partners.nytimes.com/library/national/science/050399sci-cold-fusion.html|date=3 May 1989|access-date=3 February 2017}}</ref>

[[Nikola Tesla]] claimed as early as 1899 to have used a high frequency current to light gas-filled lamps from over {{convert|25|mi|km}} away [[Wireless energy transfer|without using wires]]. In 1904 he built [[Wardenclyffe Tower]] on [[Shoreham, New York|Long Island]] to demonstrate means to send and receive power without connecting wires. The facility was never fully operational and was not completed due to economic problems, so no attempt to reproduce his first result was ever carried out.<ref>[[Margaret Cheney (author)|Cheney, Margaret]] (1999), ''Tesla, Master of Lightning'', New York: Barnes & Noble Books, {{ISBN|0-7607-1005-8}}, pp. 107.; "Unable to overcome his financial burdens, he was forced to close the laboratory in 1905."</ref>

Other examples which contrary evidence has refuted the original claim: