Thomas Lumley • 1/22/2026

Gauss is Not Mocked

This article delves into statistical methods for testing multiparameter hypotheses in survey statistics and regression models. It contrasts the Rao-Scott tests, known for better Type I error control, with 'intrinsically-weighted' tests, examining their power and performance under complex sampling designs. The analysis is based on a Masters project investigating likelihood, Wald, and score test variants.

0 comments

#Statistical Inference #Regression Analysis #Survey Statistics