A Literature Review of Questionable Research Practices in Ecological Modelling for Decisions.
This is a conceptual/theoretical paper aiming to develop a conceptual framework of reproducibility for ecological models underpinning decisions and/or decision support systems for ecology and conservation. A part of this conceptual framework will include a formal, mathetmatical definition of questionable research practices (QRPs) suitable for this context. The framework will utilise statistical decision theory, as well as concepts of decision loss / quality and uncertainty.
The purpose of this framework is to provide a useful way of thinking about issues of reproducibility in ecological models, especially where they are used to inform real-world decisions. At present, the language and concepts of reproducibility are largely focused on Null Hypothesis Signifucance Testing (NHST) or other hypothesis-testing methods, and are thus insufficient. It is also hoped that by synthesising issues of reproducibility, researcher degrees of freedom with statistical decision theory the framework / formal definition of QRPs will provide a way of perhaps measuring or quantifying the expected impacts of researcher degrees of freedom on the reproducibility of decisions.
It is hoped that the framework and definition will also be relevant to other applied scientific definitions where there are systematic and formal decision support systems being developed, such as: business management, applied economics, communication science, and clinical medicine.