2014 年 13 巻 p. 158-171
Meta-analyses are studies that bring together data or results from multiple independent studies to produce new and over-arching findings. Current data curation systems only partially support meta-analytic research. Some important meta-analytic tasks, such as the selection of relevant studies for review and the integration of research datasets or findings, are not well supported in current data curation systems. To design tools and services that more fully support meta-analyses, we need a better understanding of meta-analytic research. This includes an understanding of both the practices of researchers who perform the analyses and the characteristics of the individual studies that are brought together. In this study, we make an initial contribution to filling this gap by developing a conceptual framework linking meta-analyses with data paths represented in published articles selected for the analysis. The framework focuses on key variables that represent primary/secondary datasets or derived socio-ecological data, contexts of use, and the data transformations that are applied. We introduce the notion of using variables and their relevant information (e.g., metadata and variable relationships) as a type of currency to facilitate synthesis of findings across individual studies and leverage larger bodies of relevant source data produced in small science research. Handling variables in this manner provides an equalizing factor between data from otherwise disparate data-producing communities. We conclude with implications for exploring data integration and synthesis issues as well as system development.