Outline: Model Evaluation
What could go wrong with a model? Here are some examples.
Disuniformity
- Discontinuous changes in predicted value with regard to primary regressors.
Bias in estimates
- Spatial heterogeneity/auto-correlation
- Omitted variables - coefficient stability
- Poor model specification
- Bad treatment of time
Bias in error structure
- General heteroskedasticity
- Bad treatment of time
Outline these, and any others you can think of.
Edited by Robert Ross