Saturday, July 05, 2008

Science Cannot Be Secret

This blog post does a great job laying out some of the core concerns about complexity and proprietary information in accountability systems. Our mantra has been "simple is better, unless it's wrong". In the VA work we do, we might prefer a simpler model that has "strong" assumptions. However, we think it is important to develop model enhancements that deal with cases in which the assumptions cannot be met. We look at the results of both models to see what difference violating the assumption makes. I very much agree with the author of this blog that this is a policy decision and that the educational agency (state, local, etc.) must be engaged with the model design and be able to understand and explain any and all modeling choices.


No comments: