We focus on gathering data from impact evaluations and synthesizing that information through meta-analysis.
The hardest part of doing a meta-analysis is gathering the data. Each paper must be read by two different people, who manually extract information from the papers; if they disagree on any aspect, a third person is called in to arbitrate.
This process doesn’t scale very well and it is a large part of why meta-analyses are often out of date. With many new studies coming out, we would like to be able to streamline this process.
Machine learning allows us to do so. We can extract information about the programs and studies as well as information about the effect sizes they found. For each extracted piece of information, we will also generate a probability that the information is correct.
At the very minimum, this will greatly reduce the amount of time it takes to identify basic characteristics of studies, such as where they were done and which methods they used. It is also the only way to really stay abreast of the latest research for a variety of areas. Given that the methods should be scalable to much of health, education, and economics, we will build this tool in a general way so that its results can inform policy even in developed countries.
You can think of this as a ScienceScape for meta-analysis.
To support this initiative, we have a crowdfunding campaign. Please consider donating.