Recommendations for increasing the transparency of analysis of pre-existing datasets
Secondary data analysis, or the analysis of pre-existing data, can be a powerful tool for the resourceful researcher. Never has this been more true than now, when technological advances allow for easier sharing of data across labs and continents and the mining of large sources of “pre-existing data”. However, secondary data analysis is often ignored as a methodological tool, either when developing new open science practices or improving analytic methods for robust data analysis. In this paper, we hope to provide researchers with the knowledge necessary to incorporate secondary data analysis into their toolbox. Specifically, we define secondary data analysis as a tool and in relation to other common forms of analysis (including exploratory and confirmatory, observational and experimental). We highlight the advantages and disadvantages of this tool. We describe how engagement in transparency can improve and alter our interpretations of results from secondary data analysis and provide resources for robust data analysis. We close by suggesting ways in which subfields and institutions could address and improve the use of secondary data analysis.