Opinion

My Predictive Model Has Required Significant Recalibration

A Harvard political scientist updates her priors in public and dislikes the experience with methodological rigor.

I voted against Trump twice. I stand by both votes as expressions of my values. I also maintain a predictive model, and that model has required significant recalibration.

This is uncomfortable. Discomfort, however, is not evidence. If it were, my 2017 syllabus would have been a peer-reviewed journal.

What My Model Predicted

The model predicted institutional decay, policy chaos, and measurable deterioration across several governance indicators. Some of that happened. Some of it did not. Some indicators improved in ways the model treated as impossible because the model was built by someone who found the possibility irritating.

"A model that cannot survive contact with outcomes is not a model. It is a mood with citations." Prof. Eleanor Watts

I am not becoming a supporter. I am becoming less wrong in public. My department has asked whether this could happen more quietly. It cannot.