Skew-symmetric distributions and model choice

Skew-symmetric distributions provide flexible models suitable to fit the distribution of data affected by departures from normality, such as skewness and/orheavy tails. However, when skew-symmetric models happen to be over-parameterizedwith respect to the actual distribution, because either only one or none of the deviationsoccur, their adoption can lead to remarkable losses of efficiency in the estimationof the parameters. Consequently the paper proposes a strategy to identifywhether the actual distribution of the data is a sub-model of a skew-symmetric distribution.