Information and Model Misspecification - Classical Statistics vs. Info-Metrics

Information and Model Misspecification - Classical Statistics vs. Info-Metrics

Explore the challenges of model misspecification in statistical analysis through this insightful seminar by SFI External Professor Amos Golan. Delve into the info-metrics framework and its approach to handling imperfect and incomplete information across various scientific disciplines. Compare classical statistical methods with info-metrics, focusing on constraint specification and criterion selection. Examine a practical example investigating power law distributions using Shannon entropy and Empirical Likelihood, highlighting the complexities of identifying misspecification in seemingly equivalent predictions. Gain valuable insights into the nuances of statistical modeling and inference under conditions of uncertainty.