Reproducibility (to a limited extent also traceability ) generally means the possibility of repeating something or doing it again (e.g. a process, an experiment), or going the same way again . Reproducible also means the possibility of producing something repeatedly . In changing contexts the word can mean comprehensibility , objectivity or availability , which often leads to confusion.
Depending on the subject, the term has a deeper or more precise meaning.
In business, the term reproducibility or repeatability describes the potential of a business model to be repeated in different sales markets . This factor determines how marketable and globally competitive a product or company is. Closely related to this is the aspect of scalability , i.e. how easy it is to expand and grow a business model.
In science, reproducibility or replicability in general has a high priority. It means the repeatability of empirical-scientific research results. Only when a finding is reproducible and a replication study comes to similar, if not the same results, does the first study gain credibility.
In a replication study, an observational study, experiment or measurement is repeated under the same conditions as the initial study. If the results generated from the data obtained lead to the same conclusions or, within the framework of the expected random deviations, agree with those of the first study, the replication is successful and the results are considered to be replicable .
In order for a replication study to be possible, sufficient logging of the experimental set-up and implementation of the experiment or the procedure for selecting the observation units is part of good scientific practice.
Soil science, geology
In soil-related specialist areas, reproducibility is particularly important for rock and soil samples . Because often strong local variance of soil parameters must be particularly vigilant to ensure that the sample to be taken or in the choice of measurement points hole for the environment representative is. Data collections and geographic information systems should be thoroughly analyzed for this condition in order to avoid wrong conclusions for construction and other projects. The temporal variability due to the season, precipitation or groundwater must also be considered.
In the chemical analysis (or analytical chemistry ) is in quantitative analysis of the Eurachem instead of the expression precision a different nomenclature recommended. There are different degrees of reproducibility, which are described by the terms comparative precision , laboratory precision and repeatability . For all three it is precisely specified under which conditions an experiment must be repeated in order to measure the respective change in the results.
The International Union of Pure and Applied Chemistry defines "reproducibility":
“The closeness of agreement between independent results obtained with the same method on identical test material but under different conditions (different operators, different apparatus, different laboratories and / or after different intervals of time). The measure of reproducibility is the standard deviation qualified with the term 'reproducibility' as reproducibility standard deviation. In some contexts reproducibility may be defined as the value below which the absolute difference between two single test results on identical material obtained under the above conditions, may be expected to lie with a specified probability. Note that a complete statement of reproducibility requires specification of the experimental conditions which differ. "
In psychology, as a result of broad critical discussions, an increasing professional interest in replications developed, which in recent years has led to several large replication series.
Art and cultural studies
In the arts and cultural sciences , reproducibility describes the possibility of manually or technically copying , recording, depicting or duplicating an art object (or more generally a product) , or producing it as a series from the outset (see series production ). In this sense, the expression was addressed in particular by Walter Benjamin in his essay The work of art in the age of its technical reproducibility (1935/36). The term “reproducibility” is closely related to the pairs of terms original and copy , original and forgery, as well as unique and series product .
Reproducibility is not just a technical problem, but also a problem of legitimacy ( copyright ). Is the imitation allowed to happen at all and can it be equal to the original? Generally speaking, if writing is to capture the voice of an author, then every reading of this writing is only a substitute for that voice. But if the author's voice is viewed as a model for any reader, then every reading is an original. This has been true for music and theater since the 18th century , where every reproduction is an original, while the reproduction of a unique piece in the visual arts is only a substitute or a forgery to this day. In the case of signed art prints, on the other hand, a number of reproductions are authorized by hand.
In technology , reproducibility is an essential aid in eliminating programming errors, e.g. B. in software , or design errors , z. B. in electrical appliances . One tries to trace the processes that led to an error as precisely as possible, so that this error occurs again in the attempt and thus the problem can be narrowed down. In software projects, this helps in later " debugging " and removing the error.
In mechanical engineering, reproducibility means restoring a workpiece within its tolerances and its specific properties.
Open source software
In the case of open source software , the reproducible or deterministic compilation of the source code to the finished binary file is a method to establish a complete chain of trust and thus to be able to prove that the compiled file was generated from the published source code and has not been manipulated in between, for example to add back doors .
A constant compilation environment and tools ensure that everyone who compiles the source text correctly receives a binary file that is exactly the same bit . The resulting binary file is thus a reproducible end product (English. Reproducible build ) and can be created in the same way and easily compared by everyone.
The repeatability of test results is a fundamental requirement for scientific work. In addition, there is a public interest in ensuring that research results are secured, non-reproducible research results are recognized and intentional falsification of results is uncovered. In the media, proven forgeries, as they were known from various disciplines, caused a sensation. The Science and Technology Policy Office in Washington also took up the topic of reproducibility in 2014 on the occasion of a broad initiative Strategy for American Innovation : “Given recent evidence of the irreproducibility of a surprising number of published scientific findings, how can the Federal Government leverage its role as a significant funder of scientific research to most effectively address the problem? "
- Jim Giles: The trouble with replication. In: Nature . Vol. 442, Issue 7101, 2006, pp. 344-347.