The Distributed Meta-Analysis System is an online tool to help scientists analyze, explore, combine, and communicate results from existing empirical studies. It's primary purpose it to support meta-analyses, by providing a database for empirically estimated models and methods to integrate their results. The current version supports a range of tools that are useful for analyzing empirical climate impact results, but we hope to expand its applicability to other fields including social sciences, medicine, ecology, and geophysics.
Scientists and policy advisors often struggle to synthesize the quickly evolving state of empirical work produced in widely divergent contexts. The process of combing through the scientific literature for a meta-analysis can very difficult and costly, particularly where a wide range of factors are included in empirical relationships and the diverse methodological tools are used. As a result, meta-analyses are done infrequently, and the modelers, scientists, and policy advisors often left with out-of-date estimates. These factors encourage researchers performing meta-analyses to produce as general a result as possible, despite the multiple targeted questions that could be asked of any body of empirical estimates.
Technology has a clear and ready ability to support this integration of knowledge. First, DMAS can handle the mechanical aspects behind combining estimates, allowing empirical estimates to be incorporated into many different meta-analyses. Scientists engaged in empirical research, or in the review of others’ work, will be able to input rich numerical and analytical descriptions of empirically estimated relationships. Those descriptions would include the form and conditions of their estimated empirical relationship, the strength and applicability of their result, and the characteristics and reliability of their data. These models can then be quickly compared against and combined with previously described models of the same or related parameters.
Second, the database of empirical relationships can be expanded simultaneously by many scientists working independently to detail their empirical findings. Ultimately we hope that DMAS can be a forum where whole communities of researchers can contribute to producing better results. The tool would allow any scientist with published results to input their new empirical estimates, and instantly see how the compare to past results, and how combined with past results they form a "best-estimate" of the parameters of interest. By combining the efforts of many scientists on the Internet, the capacity to keep best-estimate empirical relationships current increases drastically. Unlike many crowd-sourcing projects, the vetting of this information is made much easier by connecting estimates with published work. This connection to the published literature further supports the construction of comprehensive meta-analyses.
Finally, the tools allow these empirical relationships to be applied to particular contexts, and to combine different kinds of results in a coherent framework. For example, models of the impact of temperature changes on agricultural yields can be first aggregated to build a best-estimate of the effect, then integrated across future temperature scenarios by country to generate an expected affect, and finally combined to estimate a social cost of climate change.