Analysis of covariance by assuming a skew normal distribution for response variable

Year 2016, Volume: 45 Issue: 6, 1805 - 1818, 01.12.2016

Afshin Fallah Zahra Goodarzi

Abstract

The traditional theory of analysis of covariance (ANCOVA) is based on
normality assumption, while in many real world applications the data
violate normality and this theory is not adequate. In this paper, we
expand a model for analysis of covariance with a skew normal response
variable. The maximum likelihood estimates of the model parameters
are provided via an EM algorithm. We also developed asymptotic confidence intervals for parameters. A simulation study is performed to
assess the performance of the proposed model. The methodology is
illustrated using a real data set.

Keywords

Analysis of covariance, skew normal, maximum likelihood, EM algorithm

References

• . . .