Good if make prior after data instead of before
9 days ago
- #Model Selection
- #Bayesian Statistics
- #Alien Existence Debate
- The article critiques a Bayesian approach to model checking and selection, arguing it presents well-known issues as novel without referencing existing solutions.
- It highlights the problem of 'model space' insufficiency and critiques the proposed solution of ignoring the issue without discussing alternatives or disadvantages.
- The discussion extends to Bayesian updating, emphasizing that priors should evolve with new data, a fundamental Bayesian principle.
- A detailed critique of applying Bayesian reasoning to the existence of aliens illustrates the complexity of defining priors and likelihoods in real-world scenarios.
- The article suggests that simplifying complex, infinite possibility spaces into discrete categories can lead to misleading conclusions if not carefully managed.
- It concludes that practical Bayesian analysis requires adjusting priors based on data to avoid cognitive dissonance and achieve more accurate posterior probabilities.