Replication (science)

from Wikipedia, the free encyclopedia

In the philosophy of science, replication refers to the repetition of a scientific study . The aim is to control and review the reported research results. Ultimately, replication achieves two things: on the one hand, it increases the acceptance of the successfully replicated study and, on the other hand, it disciplines researchers to work carefully in their study and to refrain from fraud and falsification in science .

The replication must be distinguished from the reproduction , which is based on the data of the existing study and checks its analysis. The term reproducibility is often used more generally, which also includes replicability.

Replication types

A distinction is made between two types of replication: replication in the narrower sense, that is, under the same test conditions, and replication in the broader sense, that is, under changed test conditions.

Replication under the same experimental conditions (direct replication)

In the case of a replication under the same experimental conditions, an experiment is carried out as precisely as possible under the conditions of a previous experiment. If it is not an experiment, but field and empirical studies , the goal is to demonstrate the effects again. The reproducibility of the first study can only be demonstrated if the same data or the processes underlying the data generation are used .

The reason can be doubts about the results of a previous empirical study. A professionally prepared study in the context of empirical social research always shows in its descriptive text the conditions and instruments / methods that ultimately led to the test result. This makes it easier for a subsequent scientist to check the "original experiment" through replication.

Replication under changed test conditions (conceptual replication)

In the case of a replication under changed test conditions, an experiment is carried out in a (mostly slightly) modified form of a previous experiment.

In psychology, attempt replications are also often undertaken by the scientists who carried out the "first attempt". An example is the Milgram experiment on obedience to authorities by Stanley Milgram . Milgram conducted the experiment under various conditions, including: a. the degree of authority of the experimenter was varied.

Innovation versus replication

There is an inherent tendency towards the new in science. The interest in replicative research is often weakened accordingly. To explicitly support replication-oriented studies, a 3 million euro funding program was set up in the Netherlands in 2016.

Dissemination in the social sciences

Third-party replication does not rank high in the social sciences, although it is essential to credible science. A meta-study published in the American Economic Review as early as 1986 showed that non-replicability was more the rule than the exception in economic research.

The basic requirement for replication is a sufficient degree of transparency, i.e. the data must be stored for the interested public. Although recognized as a problem, only a few journals require authors to submit the data used in their study (including processing). Even in cases where a data archive is available, it is rarely possible to replicate the associated studies.

Only a few scientific journals like the International Journal for Re-Views in Empircal Economics (IREE) specialize in replication or, like the Journal of Applied Econometrics, make at least part of the volume of replications available. The Center for Statistics Georg-August University Göttingen offers the Replication Working Papers , a working paper on series of replicated studies.

There is an increasing rethinking of the replication of results in psychological science . The reason for this is the “replication crisis” , which describes the fact that many findings could not be replicated in costly replication attempts. In this sense, the implementation of open science practices (e.g. availability of study materials) is increasingly being used in order to promote and facilitate replications in psychology.

literature

  • Milgram, S. (2007). The Milgram Experiment - On Obedience to Authority (15th Edition). Reinbek near Hamburg: Rowohlt.
  • Bortz, J. & Döring, N. (2006). Research methods and evaluation for human and social scientists (4th edition). Heidelberg: Springer.

Web links

Individual evidence

  1. ^ A b McCullough, BD & McGeary, Kerry Anne & Harrison, Teresa D .: Lessons from the JMCB Archive . In: Journal of Money, Credit and Banking . tape 38 , no. 4 , 2006, p. 1093–1107 , doi : 10.1353 / mcb.2006.0061 (English).
  2. ^ Pesaran, M. Hashem: Introducing a replication section . In: Journal of Applied Econometrics . tape 18 , no. 1 , 2003, p. 111 , doi : 10.1002 / jae.709 (English).
  3. ↑ accordingly: Bortz, 2006, p. 88
  4. ↑ Wired : Error! How science deals with its mistakes from September 9, 2016, loaded on February 6, 2017
  5. ↑ DeWald, William, Jerry Thursby and Richard Anderson: Replication in empirical economics: the Journal of Money project . In: American Economic Reviw . tape 76 , no. 4 , 1986, pp. 587-603 , JSTOR : 1806061 (English).
  6. ^ Georg-August-Universität Göttingen - Replication Working Papers. uni-goettingen.de, accessed on September 2, 2015 (English).