Abstract
Nonnormal data arise often in practice, prompting the development of flexible distributions for modeling such situations. In this article, we describe two multivariate distributions, the skew-normal and the skew-t, which can be used to model skewed and heavy-tailed continuous data. We then discuss some inferential issues that can arise when fitting these distributions to real data. We also consider the use of these distributions in a regression setting for more flexible parametric modeling of the conditional distribution given other predictors. We present commands for fitting univariate and multivariate skew-normal and skew-t regressions in Stata (skewnreg, skewtreg, mskewnreg, and mskewtreg) as well as some postestimation features (predict and skewrplot). We also demonstrate the use of the commands for the analysis of the famous Australian Institute of Sport data and U.S. precipitation data.
Original language | English (US) |
---|---|
Pages (from-to) | 507-539 |
Number of pages | 33 |
Journal | Stata Journal |
Volume | 10 |
Issue number | 4 |
DOIs | |
State | Published - 2010 |
Externally published | Yes |
Keywords
- Distribution
- Heavy tails
- Mskewnreg
- Mskewtreg
- Nonnormal
- Precipitation
- Predict
- Regression
- Skew-t
- Skewness
- Skewnormal
- Skewnreg
- Skewrplot
- Skewtreg
- st0207
ASJC Scopus subject areas
- Mathematics (miscellaneous)