This code may be of use to people using the Casualty, ETU and Demographic data released by the HackEbola event organizers to look for covariates of the spread of Ebola in West Africa. We were not able to find any non-spurious results with this data, and our main lesson to others using this data is that the holes in the missing data should be filled before any findings are drawn. There is too much reason to suspect non-random attrition in the data, and it currently produces biased, low-powered results. The work of patching the data is not an easy one, or one that can be done from the US. It will require boots on the ground collecting data in West Africa (or possibly cellphone data collection).
Built With
- python
- stata
Log in or sign up for Devpost to join the conversation.