Testing Linear versus Logarithmic Regression Models: A Comment
指出Aneuryn-Evans和Deaton提出的用于分析Cox统计量蒙特卡洛研究的渐近公式无效，因为Cox统计量通常不具有奇异联合渐近分布。
In a recent article, Aneuryn-Evans and Deaton propose asymptotic formulae for analysing Monte Carlo studies of the Cox statistics for testing non-nested hypotheses. This note shows the invalidity of those formulae by demonstrating that, in general, the Cox statistics do not have a singular joint asymptotic distribution.