The Purpose page provides a better look at the reasons why it would be helpful to have a common platform researchers can use to gather predictions as to the effects their studies will find.
There are many existing prediction markets. We think there is value in each approach but want to focus on getting independent estimates from different agents. In a prediction market, respondents would be able to view each other’s best guesses before making their own prediction, and this could lead to biases.
This project is generously supported by the Alfred P. Sloan Foundation (grant number G-2019-12325) and an anonymous foundation.
There is a tension between collecting predictions for more studies and risking survey fatigue on the part of expert forecasters. One of the biggest challenges in building a predictions platform for social science research is that it might impose a large burden on forecasters, similar to referee requests. An advantage of a centralized predictions platform is that we can ensure no individual receives an abundance of requests for predictions. Still, to further mitigate risk, we are doing a “soft launch” of the platform which will focus on gathering predictions for a few key initial studies. This will enable us to tweak parameters like the incentives offered and the frequency of prediction requests and adapt going forward. While we are still in this phase, we will be only approving a limited set of studies. Please see the discussion of these initial studies on the Purpose page for more information.