Related Experiment Videos
An Introduction to Model Selection
1University of Göttingen
Journal of Mathematical Psychology
|March 29, 2000
Summary
This paper introduces frequentist and Bayesian model selection methods like Akaike information criterion (AIC) and Bayesian information criterion (BIC) for beginners. It discusses selection bias and uses population data for clear illustration.
Area of Science:
- Statistics
- Statistical Modeling
Background:
- Model selection is crucial in statistical analysis.
- Nonspecialists require accessible introductions to complex statistical concepts.
Purpose of the Study:
- To introduce frequentist and Bayesian model selection methodologies.
- To explain the underlying principles of criteria such as AIC and BIC.
- To highlight the importance of understanding selection bias.
Main Methods:
- Explanation of frequentist model selection criteria (e.g., AIC, bootstrap, cross-validation).
- Introduction to Bayesian model selection criteria (e.g., BIC).
- Illustrative example using complete population data.
Main Results:
- Demonstration of how model selection criteria function.
- Examination of effects typically obscured in sample-based analyses.
- Discussion of selection bias in model selection.
Conclusions:
- The paper provides a foundational understanding of model selection for those with basic statistics knowledge.
- Accessible explanation of key frequentist and Bayesian approaches.
- Emphasis on practical considerations like selection bias.