Motivations and Methods for Replication in Geography: Working with Data Streams

Abstract

The article begins with a historical account of Hudson’s rural settlement theory and the various attempts to replicate Hudson’s research. Harvey’s exhortation “by our theories you shall know us” is discussed as a motivation for replication. Motivations not considered are the detection of fraud, mendacity, and incompetence, because these are the domain of reproducible research. Replication research in medicine, psychology, economics, and criminology is reviewed. The varying distinctions between replication and reproducible (R&R) research in each discipline are described. In each discipline the essential papers that have defined the “replication crisis” and the strategies that researchers have presented are discussed. These strategies include recommendations for systematic reviews and the standardization of research protocols, including the PRISMA and STROBE protocols that are now the accepted format for research in medicine. All four disciplines recommend the use of a formal meta-analysis following the systematic reviews of previous research contributions. There follows a brief discussion of a case study of a meta-analysis in geography that represents a model for others to follow and, second, the suggestion that geographically weighted regression analyses can be seen as a method for replicating the validity of a model across space. The article concludes with a review of the recently developed technology of computer and Jupyter Notebook as a way of facilitating research replication.