Multiple constraint and truncated skew models

Abstract

Attention is focussed on a truncated version of the extended two-piece skew-normal distribution, studied by Arnold et al. [On multiple constraint skewed models. Statistics. 2009;3(3):279–293]. When the truncation point is set equal to zero, the resulting model can be viewed as a flexible extension of the half-normal model. Properties of the truncated distribution are investigated and corresponding likelihood inference is considered. The methodology is applied to data set involving non-negative observations and it is verified that the fit of the truncated model compares favourably with that of the half-normal.