The multiplicity of analysis strategies jeopardizes replicability: lessons learned across disciplines

For a given research question, there are usually large variety of possible analysis strategies acceptable according to the scientific standards field, and concerns that this multiplicity plays an important role in non-replicability findings. Here, we define general framework on common sources uncertainty arising computational analyses lead multiplicity, apply within overview approaches proposed across disciplines address issue. Armed with framework, set recommendations derived therefrom, researchers will be able recognize applicable their field use them generate findings more likely replicated future studies, ultimately improving credibility process.