Mathematics
8 years ago

New gradient test improves accuracy of statistical models for small samples.

Statistical Modelling Society eBooks
Gradient test for generalised linear models with random effects.
Antonio Hermes M. da Silva-Júnior, Jochen Einbeck, Peter Craig

Paper Summary

The gradient test was developed for choosing parameters in statistical models with random effects. This test doesn't need the Fisher information matrix and has a chi-squared distribution. It was shown to work well in simulations, outperforming other common tests, especially with small sample sizes.

Related papers

Theme background image
Mathematics
4 years ago

New method reduces errors in comparing data with unequal variances

Multicollinearity
Regularization Methods
Statistical Control
Theme background image
Computer Science
5 years ago

New test revolutionizes high-dimensional data analysis for accurate results.

Model-Based Clustering
Scaling Limits
Skew Distributions
3 views
Economics, Econometrics and Finance
5 years ago

New test method offers faster, less biased inferences in statistical modeling.

Spatial Econometrics
Regularization Methods
Missing Data
1 view