The Studies On Measurement Of Replicability: A New Approach To Solve Replicability Crisis

The Reproducibility Spectrum,which proposed by Roger Peng,provides a way to measure the degree of replicability and constitutes a new approach to control replicability crisis.Since the issue of replicability has become prominent in the new century that has led to replicability crisis,the concept of replicability constitutes a scientific epistemology and major scientific issue.In 2011,Roger Peng proposed the Reproducibility Spectrum,which three levels of reproducibility are classified according to the degree of reproducibility: code,code and data,and linkable and executable code and data,it differs from the traditional standard of replicability,just replicability or no-replicability.The Reproducibility Spectrum has clarified the pursuit of higher levels of replicability,influenced the improvement of research quality in a variety of scientific fields such as epidemiology,economics,and artificial intelligence,has become a widely accepted framework for measuring degree of replicability.The differences in the degree of reproducibility is the inherent rationale for classifying the three levels of reproducibility.Firstly,code is the lowest level of reproducibility,implementing code sharing and pre-registration can help reduce and eliminate analysis flexibility,and avoid arbitrary code changes,which may lead to different conclusions from the same data,so it is the element that must be controlled in the first place.Secondly,code and data is the second level of reproducibility,achieving simultaneous open code and open data can clearly reveal the analytical process of forming conclusions from the data,and providing researchers with an effective way to understand a research.Thirdly,linkable and executable code and data is the highest level of reproducibility,which specifies the conditions for third parties to check reproducibility,such as software and hardware versions,parameters,etc.,so that others can reproduce data analysis.The mechanism by which the Reproducibility Spectrum solves the replicability crisis is that it requires open code and data that can open the black box to reveal the scientific process,thus providing a way to discover errors by backtracking the research process.Firstly,the transparency of the code and data reveals the research process and helps to achieve a shift from a focus on the academic papers to the research process.Secondly,the transparency of code and data provides a basis for re-examining the scientific research process,a robust system of accountability will reduce the risk of trust in the scientific research system and incentivize researchers to be rigorous in their research.Thirdly,the transparency and openness of code and data will help backtrack the process of scientific research,so as to identify possible scientific errors,and improve the quality of research.