Reproducible research

Reproducible research makes its methods, data, and materials sufficiently transparent and accessible for others to independently check how findings were produced.

Reproducibility: The ability to obtain consistent results using the same data, methods, and conditions. This broader property names the central aim of making research independently checkable.
Literate programming: A programming approach that combines executable code with explanatory prose. It keeps analysis steps and their explanations together in research documents.
Preregistration: A time-stamped record of a study's plans made before data collection or analysis. It separates planned tests from decisions made after seeing results.
Replication crisis: A debate about the difficulty of reproducing findings across several research fields. Replication failures intensified demands for transparent methods and materials.
Computational reproducibility: The ability to obtain consistent results by rerunning an analysis with the same data and computational methods. It captures the computational part of reproducible research, not every source of evidence.
Replicability: The ability to obtain consistent results in a new study addressing the same question. It distinguishes rerunning an original analysis from collecting new evidence.
Jupyter Notebook: An interactive document that combines code, text, and computational outputs. Notebooks can record an analysis as an inspectable sequence of steps.
Registered report: A publication format in which a study protocol is peer-reviewed before results are known. It makes methodological scrutiny precede outcome-dependent publication decisions.
Publication bias: The selective publication of research findings based on their direction or statistical significance. It can hide unsuccessful attempts that matter when assessing a finding's reliability.
Empirical reproducibility: The ability to obtain consistent findings through a new empirical study using similar methods. It tests whether findings persist beyond rerunning the original computation.

Linked from 11 pages