Abstract
This paper examines a number of statistics that have been proposed to test the normality assumption in the tobit (censored regression) model. It argues that a number of commonly proposed statistics can be interpreted as different versions of the Lagrange multiplier, or score, test for a common null hypothesis. This observation is useful in examining the Monte Carlo results presented in the paper. The Monte Carlo results suggest that the computational convenience of a number of statistics is obtained at the cost of poor finite sample performance under the null hypothesis.
Original language | English |
---|---|
Pages (from-to) | 521-532 |
Number of pages | 11 |
Journal | Journal of Applied Statistics |
Volume | 31 |
Issue number | 5 |
DOIs | |
Publication status | Published - 2004 |
Keywords
- tobit models
- normality
- statistics
- econometrics
- business mathematics