Frequentist model-averaged estimators and tests for univariate twin models

Area of Science:

• Quantitative genetics
• Behavioral genetics
• Statistical genetics

Background:

• Parameter estimates in univariate twin data analysis often neglect model selection uncertainty.
• This uncertainty can impact the reliability of genetic variance estimates.
• Existing methods may not adequately address the full scope of analytical uncertainty.

Purpose of the Study:

• To introduce frequentist model-averaged estimators for univariate twin data analysis.
• To quantify the impact of model selection uncertainty on parameter estimates.
• To improve the accuracy and power of detecting additive genetic variance.

Main Methods:

• Utilized information-theoretic criteria to assign model weights for averaging.
• Conducted simulation studies to evaluate estimator performance.
• Compared model-averaged estimators against individual model estimators and decision-based procedures.

Main Results:

• Model-averaged estimators demonstrated lower mean-squared error for additive genetic variance in simulations.
• This improvement was observed particularly for small to moderate sample sizes.
• Bootstrap tests based on model-averaged estimators showed increased power to detect additive genetic variance.

Conclusions:

• Frequentist model-averaged estimators provide a more robust approach to univariate twin data analysis.
• Accounting for model selection uncertainty enhances the precision of genetic parameter estimation.
• The proposed methods offer superior performance in detecting additive genetic variance compared to traditional approaches.